USF Libraries
USF Digital Collections

Improved methodologies for modeling storage and water level behavior in wetlands

MISSING IMAGE

Material Information

Title:
Improved methodologies for modeling storage and water level behavior in wetlands
Physical Description:
Book
Language:
English
Creator:
Nilsson, Kenneth
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Analytical techniques
Bathymetry
Frequency analysis
Hydrologic models
Water storage
Wetlands
Dissertations, Academic -- Civil & Environmental Engineering -- Doctoral -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Wetlands are important elements of watersheds that influence water storage, surface water runoff, groundwater recharge/discharge processes, and evapotranspiration. To understand the cumulative effect wetlands have on a watershed, one must have a good understanding of the water-level fluctuations and the storage characteristics associated with multiple wetlands across a region. An improved analytical method is presented to describe the storage characteristics of wetlands in the absence of detailed hydrologic and bathymetric data. Also, a probabilistic approach based on frequency analysis is developed to provide insight into surface and groundwater interactions associated with isolated wetlands. The results of the work include: 1) a power-function model based on a single fitting parameter and two physically based parameters was developed and used to represent the storage of singular or multiple wetlands and lakes with acceptable error, 2) a novel hydrologic characterization applied to 56 wetlands in west-central Florida provided new information about wetland hydroperiods which indicated standing water was present in the wetlands 62% of the time and these wetlands were groundwater recharge zones 59% of the time over the seven year study, 3) the smallest extreme value probability distribution function was identified as the best-fit model to represent the water levels of five wetland categories in west-central Florida, 4) representative probability models were developed and used to predict the water levels of specific wetland categories, averaging less than 10% error between the predicted and recorded water levels, and 5) last, based on this probability analysis, the various wetland categories were shown to exhibit similar means, extremes and ranges in water-level behavior but unique slopes in frequency distributions, a here to for new finding. These results suggest that wetland types may best be differentiated by the regular variability in water levels, not by the mean and/or extreme water levels. The methods and analytical techniques presented in this dissertation can be used to help understand and quantify wetland hydrology in different climatological or anthropogenic stress conditions. Also, the methods explored in this study can be used to develop more accurate and representative hydrologic simulation models.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Kenneth Nilsson.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0003458
usfldc handle - e14.3458
System ID:
SFS0027773:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Improved Methodologies for Modeling Storage and Water Level Behavior in Wetlands by Kenneth Allan Nilsson A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Civil Engineering Department of Civil and Environmental Engineering College of Engineering University of South Florida Major Professor: Mark A. Ross, Ph.D. Jeffrey A. Cunningham, Ph.D. Jeffrey S. Geurink, Ph.D. Terrie M. Lee, M.S.E. Rafael A. Perez, Ph.D. Mark C. Rains, Ph.D. Kenneth E. Trout, Ph.D. Date of Approval: March 23, 2010 Keywords: analytical techniques, bathymet ry, frequency analysis, hydrologic models, water storage, wetlands Copyright 2010 Kenneth Allan Nilsson

PAGE 2

ACKNOWLEDGMENTS I want to thank my committee members Dr. Jeffrey Cunningham, Dr. Jeffrey Geurink, Ms. Terrie Lee, Dr. Rafael Perez, Dr. Ma rk Rains and Dr. Kenneth Trout for their respective support and for providi ng invaluable insight on al l aspects of my work. I especially want to thank Dr. Kenneth Trout fo r all of his help with the development of this dissertation. He was always around with an open door policy to discuss and provide valuable insight into the various topics covered in this work, and to listen to my diatribe when other parties were not around. A dditionally, I graciously acknowledge the contributions of Mr. Michael Hancock a nd Mr. Doug Leeper for their valuable contributions throughout the development of this dissertation. Furthermore, I would like to thank my parents and family for their unconditional support and understanding through this experience. Also I would like to tha nk all my friends for providing stress relief, and special thanks goes out to Heather, Tov and Russell for listening to my rants and for providing much n eeded relief and escape when the stress of this project was overwhelming. I sincerely appreciate everyones support, understanding and help.

PAGE 3

i TABLE OF CONTENTS LIST OF TABLES............................................................................................................... v LIST OF FIGURES.......................................................................................................... vii ABSTRACT....................................................................................................................... ix CHAPTER 1 INTRODUCTION........................................................................................ 1 CHAPTER 2 A GENERAL MODEL TO REPRESENT MULTIPLE WETL AND STAGE-STORAGE BEHAVIOR.................................................................. 6 2.1. Introduction....................................................................................................... 6 2.2. Theory and Methodology..................................................................................8 2.2.1. V-h Power-Function W etland Model.................................................8 2.2.2. Model Development Data Set............................................................ 9 2.2.3. V-h Model Shape Param eter Development...................................... 14 2.3. Methods...........................................................................................................15 2.3.1. Model Evaluation Techniques......................................................... 15 2.3.2. General Shape Parameter Development.......................................... 16 2.4. Specific V-h Model Param eters...................................................................... 17 2.5. V-h Model Param eter Sensitivity Analysis..................................................... 19 2.6. Results and Discussion................................................................................... 22 2.6.1. General Shape Parameter Analyses................................................. 22 2.6.1.1. Individual Wetland Categories......................................... 25 2.6.1.2. Wetland Groups................................................................ 27 2.6.2. General V-h Model Perform ance..................................................... 27 2.6.2.1. Individual Wetland Categories......................................... 27 2.6.2.2. Wetland Groups................................................................ 29 2.6.2.3. Volume Over Prediction ( VOP) Analysis...........................33 2.7. Validation Data Set Analysis.......................................................................... 33 2.7.1. Validation Data Set..........................................................................38 2.7.2. General V-h Model Perform ance with Validation Data................... 39 2.7.2.1. Storage Predictions ( V ) .....................................................39 2.7.2.2. Stage Predictions (h) .........................................................39 2.7.3. Discussion of the Validation Application........................................ 40 2.8. Conclusions.....................................................................................................41 2.9. Acknowledgements.........................................................................................44

PAGE 4

ii CHAPTER 3 HYDROLOGIC CHAR ACTERI ZATION OF 56 WESTCENTRAL FLORIDA ISOLATED WETLANDS USING A PROBABILISTIC METHOD......................................................................................................................... .45 3.1. Introduction..................................................................................................... 45 3.2. Description of Study Area..............................................................................49 3.2.1. Precipitation Patterns.......................................................................55 3.2.2. Wetland Classifications................................................................... 56 3.2.3. Wetland Hydrogeologic Setting....................................................... 60 3.3. Wetland and Upland Water Elevation Data.................................................... 61 3.4. Methods...........................................................................................................70 3.4.1. Hydrologic Evaluation..................................................................... 70 3.4.1.1. Empirical Distribution (Frequency) Development...........70 3.4.1.2. Relative Water Level Development.................................. 72 3.4.1.3. Frequency of Water Levels at Dry Bed ( DB ) and Nor mal Pool ( NP )..........................................................................73 3.4.2. Wetland Category and Group Com parisons KolmogorovSmirnov Tests............................................................................................73 3.4.2.1. Wetland Category Comparisons.......................................74 3.4.2.2. Wetland Group Comparisons............................................75 3.5. Results............................................................................................................. 75 3.5.1. Wetland Hydrologic Evaluatio n (Frequency Analysis) ................... 75 3.5.1.1. Well Ground Elevation Datum ( GE ) ................................75 3.5.1.2. Wetland Dry-Bed Datum ( DB ) .........................................80 3.5.1.3. Dry Bed ( DB ) and Nor mal Pool (NP ) Relative Frequency Identification................................................................ 84 3.5.2. Wetland Category Water-Level Data Comparisons Kol mogorov-Smirnov Tests....................................................................... 85 3.5.2.1. Wetland Category Comparisons.......................................85 3.5.2.2. Regional Wetland Groups................................................. 87 3.6. Discussion.......................................................................................................89 3.6.1. Wetland Hydrologic Evaluatio n (Frequency Analysis) ................... 89 3.6.2. Frequency of Water Levels at Dry Bed ( DB ) and Norm al Pool ( NP )...................................................................................................92 3.6.2.1. Analytical Model Application........................................... 93 3.6.3. Combined Wetland Water-Level Data Comparisons...................... 94 3.7. Application: Impacted Wetland Identification................................................ 95 3.8. Conclusions...................................................................................................102 3.9. Acknowledgments......................................................................................... 104 CHAPTER 4 THE EXTENT AND PR EVALE NCE OF GROUNDWATER RECHARGE/DISCHARGE CONDITIONS IN WEST-CENTRAL FLORIDA ISOLATED WETLANDS............................................................................................... 105 4.1. Introduction................................................................................................... 105 4.2. Methods.........................................................................................................107 4.2.1. Head Differences between Paired Wetland and Upland Water Levels ............................................................................................107 4.2.2. Seasonal Group Water Level Analyses..........................................109

PAGE 5

iii 4.3. Results........................................................................................................... 111 4.3.1. Head Difference between Wetland and Upland Water Levels (Surficial Aquifer) .................................................................................... 111 4.3.1.1. Standard Statistical Analyses.......................................... 111 4.3.1.2. Frequency Analyses........................................................112 4.3.2. Seasonal Group Water-level Conditions........................................ 113 4.3.2.1. Standard Statistical Analyses.......................................... 113 4.3.2.2. Frequency Analyses........................................................115 4.3.2.3. Wilcoxon Rank Sum Tests.............................................. 116 4.4. Discussion.....................................................................................................117 4.4.1. Head Differences between Paired Wetland and Upland Water Levels ............................................................................................117 4.4.1.1. Complete Data Set.......................................................... 117 4.4.1.2. Seasonal Data Sets.......................................................... 118 4.4.1.3. Consistent Recharge Feature Spatial Locations..............120 4.4.1.4. Recharge Wetland Versus Flow-through Wetland......... 120 4.4.2. Seasonal Group Water Levels........................................................ 122 4.4.2.1. Recharge Characteristics................................................. 122 4.4.2.2. Wilcoxon Rank Sum Tests.............................................. 123 4.5. Conclusions...................................................................................................124 CHAPTER 5 PROBABILITY DENSITY FUNCTION REPRESENTATIONS OF WESTCENTRAL FLORIDA ISOL ATED WETLAND WATER LEVELS.......... 126 5.1. Introduction................................................................................................... 126 5.2. Water-Level Data.......................................................................................... 128 5.2.1. Best-Fit Probability Distribution Identification............................. 129 5.2.2. Wetland Category Empi rical Distributions .................................... 130 5.3. Methods.........................................................................................................132 5.3.1. Smallest Extreme Value Distribution............................................ 132 5.3.2. Smallest Extreme Value Parameter Identification......................... 134 5.3.3. SEV Model Evaluation ................................................................... 135 5.3.3.1. Wetland Category Analyses............................................ 135 5.3.3.2. Individual Wetland Comparisons................................... 136 5.4. Results and Discussion................................................................................. 137 5.4.1. Smallest Extreme Value Parameter Identification......................... 137 5.4.2. Smallest Extreme Value ( SEV ) Distribution Models ..................... 143 5.4.2.1. SEV Model W ater Level Predictions.............................. 143 5.4.2.2. SEV Model Evaluation .................................................... 149 5.4.2.2.1. Wetland Category Evaluation..........................149 5.4.2.2.2. Individual Wetland Comparisons.................... 149 5.4.2.3. Discussion SE V Model Performance........................... 150 5.4.3. SEV Models Probability Plots ..................................................... 151 5.4.4. Theoretical Application................................................................. 154 5.5. Conclusions...................................................................................................155 CHAPTER 6 SUMMARY AND CONCLUSIONS....................................................... 157

PAGE 6

iv REFERENCES................................................................................................................164 APPENDICES.................................................................................................................169 Appendix A: Staff Gauge and W etland Well Data Correlations........................ 170 Appendix B: Wetland Well and Upland Well Data Normality Check............... 172 Appendix C: Monthly Versus Daily Data Comparison ......................................175 Appendix D: SWFWMD White Papers on Wetland Histories........................... 178 ABOUT THE AUTHOR.......................................................................................End Page

PAGE 7

v LIST OF TABLES Table 2.1. Wetland site characteristics and V-h Model perform ance......................... 11 Table 2.2. V-h Model shape param eter and maximum pool depth estimate sensitivity analysis..................................................................................... 21 Table 2.3. Generalized shape parameter evaluation. .................................................. 24 Table 2.4. Lake validation data set fo r general shape param eter evaluation.............. 35 Table 3.1. Wetland identification and physical characteristics. .................................. 53 Table 3.2. Monitoring we ll identification and p hysical properties. ............................ 63 Table 3.3. Wetland well and upland well data collection schedule. ........................... 66 Table 3.4. Wetland and upland water el evation su mmary statistics (NGVD 29)..............................................................................................................68 Table 3.5. Wetland and upland empirical distribution summary statistics adjusted to the ground elev ation at the wetland w ell (GE )........................ 79 Table 3.6. Wetland and upland empirical distribution summary statistics adjusted to the wetland dry-bed elevation (DB ). ....................................... 83 Table 3.7. Dry bed ( DB ) and normal pool ( NP ) probability. ...................................... 84 Table 3.8. Wetland category monthl y water level description. .................................. 86 Table 3.9. Wetland category water-level distribution comparisons, Kol mogorov-Smirnov test results.............................................................. 87 Table 3.10. Regional wetland group monthly data description. ...................................88 Table 3.11. Regional wetland group wate r-level distribution com parisons, Kolmogorov-Smirnov test results.............................................................. 89 Table 3.12. Wetland percentile s exceeding one standard deviation (outliers).............. 99 Table 4.1. Wetland-upland head difference.............................................................. 111

PAGE 8

vi Table 4.2. Wetland surficial aquifer head difference at par ticular frequency indices. .....................................................................................................112 Table 4.3. Seasonal wetland and up land surficial water levels. ................................ 115 Table 4.4. Seasonal wetland and upland surf icial water levels at particular frequency indices. .................................................................................... 116 Table 4.5. Wilcoxon rank sum test results................................................................ 117 Table 5.1. Wetland category monthly data description............................................ 129 Table 5.2. Comparison of alternative probability distributions, AndersonDarling test. ..............................................................................................130 Table 5.3. Wetland category empirical di stribution function statistics per percentile. .................................................................................................131 Table 5.4. SEV distribution function param eters and distribution fit test results.......................................................................................................138 Table 5.5. SEV Model predicted w ater levels and category evaluation.................... 144 Table 5.6. SEV m odel prediction RMSE per percentile ( RMSEP)............................ 150 Table C.1. Wetland well daily versus m onthly data distribution com parisons......... 177

PAGE 9

vii LIST OF FIGURES Figure 2.1 Representative model predic tion comparisons for a) W est-central Florida wetlands, b) St. Denis NWA wetlands, c) Pothole wetlands, and d) Lakes............................................................................... 32 Figure 3.1 Study wetland locations in the northern Tam pa Bay region of westcentral Florida............................................................................................50 Figure 3.2 Annual regional rainfall m easured at th e Tampa International Airport (T.I.A.) and at the Hillsbo rough River State Park (HRSP)........... 56 Figure 3.3 Example of SWFW MD well and staff gauge locations for a typical wetland. ......................................................................................................58 Figure 3.4 Generalized hydrogeologic section and vertical head distribution (m odified from Lee et al., 2009)................................................................ 61 Figure 3.5 Empirical distribution func tion charts representing wetland and upland water levels adjusted to the ground elevation at the well ( GE )............................................................................................................78 Figure 3.6 Empirical distribution function ch arts representing the wetland and upland water levels adjusted to the wetland dry-bed elevation ( DB )............................................................................................................82 Figure 3.7 Wetland category empirical di stribution functions rela tive water levels based on the dry bed datum (DB ).................................................... 86 Figure 3.8 Regional wetland group empiri cal distribution fu nctions, relative water levels based on the dry bed datum ( DB ).......................................... 88 Figure 3.9 Comparison of water-table depth and upland water level range. ............... 90 Figure 3.10 Individual wetland ou tlier distributio n functions. ...................................... 97 Figure 3.11 Wetlands with empi rical distribution outliers ( 1 StD from mean)....... 100 Figure 4.1 Seasonal wetland and upland surficial aquifer water levels. ................... 114

PAGE 10

viii Figure 4.2 Conceptualized interactions of wetlands with (A) groundwater recharge and (B) groundwater flow through (m odified from Lee et al., 2009).................................................................................................. 122 Figure 5.1 Smallest extreme va lue distribution, general case. ..................................133 Figure 5.2 SEV Model best-fit distributi ons, regional group and cypress wetlands. .................................................................................................. 140 Figure 5.3 SEV Model best-fit distributi ons, m arsh and cypress-marsh wetlands................................................................................................... 141 Figure 5.4 SEV Model best-fit distributi ons, hardwood and wet prairie wetlands. .................................................................................................. 142 Figure 5.5 SEV Model predic ted water levels and recorded water levels for the regional and cypress wetlands.................................................................146 Figure 5.6 SEV Model predic ted water levels and recorded water levels for the marsh and cypress-marsh wetlands.......................................................... 147 Figure 5.7 SEV Model predic ted water levels and recorded water levels for the hardwood and wet prairie wetlands......................................................... 148 Figure 5.8 SEV Model probability plots for all wetland categories. ......................... 152 Figure 5.9 SEV Model probability plots for th e wetland groups............................... 153 Figure B.1 Typical wetland and upland we ll probability density functions. ............. 173

PAGE 11

ix IMPROVED METHODOLOGIES FOR MODELING STORAGE AND WATER LEVEL BEHAVIOR IN WETLANDS Kenneth Allan Nilsson ABSTRACT Wetlands are im portant elements of watershe ds that influence water storage, surface water runoff, groundwater recharge/discharge processes, and evapotranspiration. To understand the cumulative effect wetlands ha ve on a watershed, one must have a good understanding of the water-level fluctuations and the storag e characteristics associated with multiple wetlands across a region. An improved analytical method is presented to describe the storage characteristics of wetlands in the absence of detailed hydrologic and bathymetric data. Also, a probabilistic a pproach based on frequency analysis is developed to provide insight into surface a nd groundwater interactions associated with isolated wetlands. The results of the work include: 1) a power-function model based on a single fitting parameter and two physically base d parameters was developed and used to represent the storage of singular or multiple we tlands and lakes with acceptable error, 2) a novel hydrologic characteriza tion applied to 56 wetlands in west-central Florida provided new information about wetland hydro periods which indicated standing water was present in the wetlands 62% of the time and these wetlands were groundwater recharge zones 59% of the time over the seve n year study, 3) the smallest extreme value probability distribution function was identified as the best-fit model to represent the water

PAGE 12

x levels of five wetland categorie s in west-central Florida, 4) representative probability models were developed and used to pred ict the water levels of specific wetland categories, averaging less than 10% error between the predicted and recorded water levels, and 5) last, based on this probability analysis, the various wetland categories were shown to exhibit similar means, extremes a nd ranges in water-level behavior but unique slopes in frequency distributi ons, a here to for new finding. These results suggest that wetland types may best be differentiated by the regular variability in water levels, not by the mean and/or extreme water levels. The methods and analytical techniques presented in this dissertation can be used to he lp understand and quantify wetland hydrology in different climatological or anthropogenic stress conditions. Also, the methods explored in this study can be used to develop mo re accurate and representative hydrologic simulation models.

PAGE 13

1 CHAPTER 1 INTRODUCTION Wetlands are defined as those areas that are in undated or saturated by surface or groundwater at a frequency and duration suffi cient to support, and that under normal circumstances do support, a prevalence of vegetation typically adapted for life in saturated soil conditions [33CFR328.3(b)] (U.S. Environmental Protection Agency 2009). Wetlands play an important role in the hydrology of watersheds impacting water storage, surface water runoff, groundwate r recharge/discharge processes, and evapotranspiration (Bullock and Acreman 2003). These influences are difficult to quantify or model in many sett ings, especially in shallo w water-table environments typified by west-central Florida. In order to understand the individual or cumulative effect we tlands have on a watershed or region, water level records a ssociated with multiple wetla nds across a region or within a specific wetland category must be studied. The hydrologic characte rization of wetlands requires the use of long-term data records to describe the interacti on of different surface and groundwater influences. Monitoring the po oled water fluctuations as well as watertable fluctuations in and around wetlands is critical in evaluati ng surface and groundwater interactions.

PAGE 14

2 An application of the hydrologic characteri zation of wetlands is the development of accurate hydrologic models that can be used to predict watershed responses. In particular, the variability associated with we tland water level fluctuations needs to be understood and quantified. A good representa tion of wetland storage behavior is essential to ensure the respective hydrologic model functi ons reliably (Winter 1999). Without reasonable hydraulic and storage information, hydrologic models may not represent or predict the wa ter balance in the hydrologic system accurately, and may produce inaccurate es timates of stream flows, gr oundwater recharge/discharge, evapotranspiration, flood plain delineation, and/or wetland sustainability. To avoid this shortcoming, water resource engineers and hydr ologists need to better define: 1) the surface and subsurface water level characteri stics associated with wetlands, 2) the movement of water into and out of wetland s, and 3) the surface and subsurface water storage of wetlands in a hydrologic study area. Defining these characteristics for any fin ite hydrologic study area could be a daunting task. For example, approximately 20% of the land surface in the Southwest Florida Water Management District (SWFWMD), which encompasses 25,900 km2 (10,000 square miles) of west-central Florida, and 29% of Florida ov erall are occupied by wetlands (Lee et al. 2009; Southwest Florid a Water Management District 2007). This presents a significant problem for large si mulation models. Detailed bathymetric profiles, used to estimate wetland storage behavior, as well as abundant, long-term and accurate water-level records typically do not exist for most wetlands. Furthermore, traditional methods used to describe wetla nd water-level fluctuations, such as the

PAGE 15

3 statistical mean, median and ranges, pr ovide limited insights into the hydrologic characteristics of a wetland or particular wetland types. This dissertation presents an analysis of long-term wate r-elevation data and improved methods designed to help describe wetland wate r-level fluctuations. Further, the work provides a better means to define the above gr ound storage characteri stics as well as the surface and groundwater inte ractions associated with wetland s. An analytical method is developed to describe the stor age characteristics of wetlands in the absence of detailed hydrologic and bathymetric data. Also, a proba bilistic approach (f requency analysis) is used to provide insights into surface and groundwater interactions associated with isolated wetlands. The scopes of the sp ecific chapters are outlined below. In Chapter Two an analytical model is deve loped and evaluated that can be used to predict the storage behavior of multiple wetlands and lakes when detailed bathymetric data is limited or unavailable. General models were developed based on detailed bathymetry of wetlands and lakes located in west-central Florid a, North Dakota and Canada. A new method employing frequenc ies is introduced in Chapte r Three to characterize the surface water and groundwater levels associ ated with 56 various isolated wetlands located in west-central Florida. The hydrologic characterization of these wetlands utilizes a unique long-term data set comprised of paired wetland and upland monitoringwell water elevations. The data describe the duration of diffe rent water-level elevations

PAGE 16

4 in the wetland when it is flooded, and the dur ation of different water-table elevations below the wetland when they are dry. Empiri cal (frequency) distribut ions were used to identify key probability indices corresponding to the period of wetland inundation, as well as median and extreme water levels. Thes e distributions were then used to compare the hydrologic characteristics of different wetland categories. Additionally, the distributions were used to help identify imp acted wetlands located th roughout the region. In Chapter Four water-elevation records fo r wetlands were paired with groundwater elevation records at upland wells to evaluate the interactive relationships and recharge/discharge charac teristics between the isolated wetlands and surrounding uplands. Long-term wetland and upland monito ring well data were compared as well as specific data relating to the peak dry season (e.g., March-May) and wet season (e.g., JulySeptember) to note differences in behavior. Best-fit probability density f unctions (probability models) were developed in Chapter Five to represent the water levels associated with five distinct wetland categories, and five groups of wetlands in west -central Florida. The model development data sets were comprised of water levels representing all of the individual wetlands within a specific category or group. The combined data sets were developed using water-elevation data normalized to the respective wetland dry bed elevation. The probability models can be used to differentiate the water-level charac teristics associated with different wetland categories and groups, and can be used as a calibration tool for hydrologic modeling applications.

PAGE 17

5 The methodologies defined in this work will provide insight into the hydrologic characteristics of various wetland types, and enhance the modeling capability of wetland storage where available data is scare or does not exist. Th e storage model in conjunction with the frequency analyses and probability models will improve the accuracy of wetland representation in hydrologic models and aid hydrologists in predicting surface water runoff, river stage and discharg e, and groundwater fluctuations. Further, this work will help evaluate the overall hydrologic impact on wetlands subjected to anthropogenic and natural climatological stresses. Last, the methodologies se t forth in this dissertation can be applied to wetlands in other regions ar ound the United States and the world to help understand their behavior and function in different geologic settings.

PAGE 18

6 CHAPTER 2 A GENE RAL MODEL TO REPR ESENT MULTIPLE WETLAND STAGE-STORAGE BEHAVIOR 2.1. Introduction Quantifying the relationship be tween stage and storage volum e for wetlands and lakes is important for developing accurate hydrologic models for environments containing significant wetland or lake features. Hydrologic simula tion models such as the Hydrological Simulation ProgramFORTRAN (HSPF) (Bicknell et al. 2001) require specification of stage-storage behavior of wetlands and lakes in the model domain. Wetland area-depth (A-h ) and volume-depth ( V-h ) relationships pertaining to the standing (pooled) water portion of the wetland basin are typically determined from detailed bathymetric maps or simple geometric models usually specific to each depression (Hayashi and van der Kamp 2000). Extensive a nd costly surveying is the most reliable method to determine accurate bathymetric profil es. However, detailed surveying is often impractical for larger hydrologic model domai ns comprised of hundreds to thousands of wetlands (Lee et al. 2009; Southwest Florida Water Management District 2007). Yet, a reasonable representation of wetland storage be havior is necessary for hydrologic models to function reliably and possess adequate predictive capabilities. Unfortunately, estimating the storage characteristics of each wetland in such a study area, especially when faced with a lack of survey data can present a significant problem.

PAGE 19

7 Analytical models have been utilized in hydrologic studies to predict wetland storage behavior for many years (Singh and Woolhiser 2002). However, most of the time the respective models and/or model parameters were developed to predict the behavior of the individual wetlands contained in a particular study. For instance, OConnor (1989) developed power-function models to simulate th e variations of dissolved solids in lakes and reservoirs, where separate model parame ters were developed for each lake and reservoir. Shjeflo (1968) used a prismoidal formula to verify that wetland volumes developed from specific topogr aphic maps were accurate and Wise et al., (2000) developed a stage-volume relationship for an isolated marsh wetland. Furthermore, Hayashi and van der Kamp (2000) used a power function model utilizing a scaling constant and a dimensionless slope profile to represent the area-depth relations of individual shallow wetlands. The power-functi on model parameters were used to define the size and geometry of specific depre ssions. Later Brooks and Hayashi (2002) modified Hayashi and van der Kamps equati on to estimate the maximum volumes of the individual wetlands. Alt hough these models proved eff ective for their respective purposes, they were not intended to be used in a generalized manner, i.e. to model the storage behavior of multiple wetlands in a study domain. Two primary objectives were established to address these issues. The first was to develop an analytical technique that utilizes a simple pow er-function model to represent the stage-storage relationships of individual wetlands and lakes (i.e. specific powerfunction model). The technique makes use of a single dimensionless fitting or shape parameter that can be used to define a specific wetland stage-storage relationship.

PAGE 20

8 Further, the technique requires limited field data such as th e maximum or reference pool area (based on vegetative cover) obtained from aerial photographs or from polygon coverages in Geographic Information System (GIS) databases, such as the National Wetlands Inventory (U.S. Fish and Wildlife Service 2007b), and the associated maximum pool depth corresponding to that area. The second was to use the power-function mode l to predict the storage behavior of multiple wetlands and lakes in a hydrologic st udy area when detailed bathymetric data is limited or unavailable. The goals of this obj ective were to: 1) deve lop generalized shape parameters (i.e. general stage-storage models ) for specific wetland categories, lakes, and wetland groups, 2) investigate the error of th e predicted stage-storag e relationships using the general shape parameters, 3) test the general shape parameter against an independent validation data set comprised of 21 lakes in west-central Florida, and 4) to aid hydrologic modelers potentially using the stage-storage model by quantify the sensitivity and uncertainty of the shape para meter and the sensitivity of the reference pool depth. 2.2. Theory and Methodology 2.2.1. V-h Power-Function W etland Model Wetland and lake stage-storage relationships can be described via a simple power function relating the wetland pool volume ( V ) to the wetland pool depth ( h ) using a single dimensionless shape parameter ( m ) (Nilsson et al. 2008). Parabolic equations serve as a starting geometric model for these depressions. From this model, it can be shown that a

PAGE 21

9 power function representing the conical rotati on of any profile (i.e convex, planar or concave) with the origin at the deepest point in the wetland can be expressed as: m o oo Vhh h m hA hV ) ( (2.1) where VVh [L3] is the wetland or lake volume corresponding to the respective pool depth, h [L], Ao [L2] is the maximum coverage area when the wetland pool is full, ho [L] is the maximum wetland pool depth, and m is the dimensionless fitting or shape parameter describing the wetland V-h geometric relationship. The V-h power-function model [Eq. (2.1)] will be referred to as the V-h Model hence forth. The wetland V-h relationship is robust, describing a wide range of ge ometries, depending on the value of the m parameter, however the primary assumption for this method is that wetlands are circular in shape. For instance, m = in Eq. (2.1) produces a vertical line at the maximum pool depth, representing cylindrical storage, m = 1 produces a planar curve, and 0 < m < 1 and 1 < m < produce convex and concave volume-stage curves respectively. 2.2.2. Model Development Data Set Specific V-h Models were developed for 42 individua l wetlan ds and lakes. The specific shape parameters presented in Table 2.1 were derived from the specific V-h Models using detailed bathymetric survey data for five cypress wetlands, five marsh wetlands and 17 lakes located in west-central Florida (Haag et al. 2005; Nilsson et al. 2008), as well as five pothole wetlands located in St. Denis National Wildlife Area in Saskatchewan, Canada (Hayashi and van der Kamp 2000), and 10 prairie pothole we tlands located in

PAGE 22

10 North Dakota (Shjeflo 1968). The west-centr al Florida wetlands and lakes were formed by solution weathering of the ka rst terrain and in some instances deep karst collapse, while the pothole wetlands located in the nor thern United States and Canada were formed by glacial scouring. Additionally, Table 2.1 contains the maximum pool areas ( Ao), pool depths ( ho), pool volumes ( Vo) and the individual wetland st orage shape parameters ( m ), and associated summary statistics for all wetlands in the data set.

PAGE 23

11 Table 2.1. Wetland site characteristics and V-h Model performance. Characteristics Model Evaluation Ao ho Vo Upper 80% V o 100% V o VARE VRE Wetland Category (x103 m2) (m) (x103 m3) k m RMSERe l*(%) m RMSERel*(%) (%) (%) West-Central Florida W05 Cypress 35.5 0.6 5.7 23 4.22.9 4.3 2.8 7.7 2.8 W19 Cypress 8.4 0.8 2.8 28 2.50.6 2.5 0.6 1.1 0.2 S63 Cypress 5.1 0.4 0.9 16 2.74.8 2.7 4.6 10.2 4.1 S68 Cypress 23.4 0.5 5.6 17 2.01.7 2.0 1.7 1.4 -0.4 GSC Cypress 6.8 0.5 1.0 18 3.51.2 3.5 1.2 4.2 -2.4 W03 Marsh 29.9 1.5 17.2 35 2.76.2 2.8 5.8 7.1 -0.5 W29 Marsh 26.4 0.9 11.6 29 2.01.9 2.0 1.8 2.2 0.2 HRSP Marsh 9.0 0.7 1.8 24 3.31.5 3.3 1.5 4.3 -2.4 DP Marsh 21.0 2.4 20.5 41 2.51.6 2.5 1.6 2.7 -1.2 GSM Marsh 6.6 0.3 1.2 12 1.81.9 1.8 1.8 1.6 -0.1 Big Fish Lake 4,330 7.3 19,300 13 1.86.6 1.8 6.4 15.6 8.3 Bonnie Lake 133 4.6 253 17 2.33.3 2.3 3.3 10.0 -6.8 Calm Lake 610 10.02,200 12 2.82.0 2.8 1.9 3.8 0.5 Clear Lake 698 8.8 3,900 11 1.61.7 1.6 1.7 2.9 1.1 Garden Lake 72 8.4 177 11 3.41.0 3.4 1.0 1.8 -0.5 Green Lake 423 5.6 440 11 5.16.8 5.1 6.8 22.8 -13.1 Jackson Lake 13,300 9.8 77,100 12 1.72.5 1.7 2.5 4.6 1.1 Letta Lake 2,210 6.7 9,880 11 1.52.0 1.5 1.9 2.4 1.1 Middle Lake 1,110 5.8 3,440 11 1.91.3 1.9 1.3 2.8 1.6 Mound Lake 444 9.7 1,650 12 2.61.2 2.6 1.2 1.5 -0.2 Mountain Lake 272 5.1 632 10 2.22.2 2.2 2.2 6.3 3.0 Neff Lake 1,310 7.8 3,230 10 3.10.4 3.1 0.4 2.8 -1.1 Placid Lake 14,700 18.0106,000 13 2.62.1 2.6 2.1 4.8 2.1 Pretty Lake 420 8.0 1,950 10 1.71.9 1.7 1.9 2.2 -1.0 Reinheimer Lake 158 3.6 187 12 2.92.2 2.9 2.1 8.1 -3.2 Round Lake 57 8.1 143 10 3.21.3 3.2 1.3 3.3 -2.4 Spring Lake 259 15.51,830 12 2.21.3 2.2 1.3 1.3 0.1

PAGE 24

12 Table 2.1. (Continued). Characteristics Model Evaluation Ao ho Vo Upper 80% V o 100% V o VARE VRE Wetland Category (x103 m2) (m) (x103 m3) k m RMSERe l*(%) m RMSERel*(%) (%) (%) St. Denis NWA S92 Pothole 3.2 1.2 1.7 12 2.2 1.8 2.21.8 5.2 -3.4 S104 Pothole 1.2 0.7 0.4 7 2.0 1.0 2.01.0 2.7 -1.7 S109 Pothole 4.1 1.2 2.1 12 2.3 1.2 2.31.2 3.0 -1.3 S120 Pothole 3.2 1.1 1.9 11 1.8 0.8 1.80.8 2.2 -1.4 S125s Pothole 3.9 1.0 2.0 10 2.0 0.4 2.00.4 0.4 0.2 North Dakota 1 Pothole 81 2.6 143 11 1.5 0.6 1.50.6 1.7 -1.1 2 Pothole 174 3.3 403 14 1.4 1.0 1.41.0 1.3 0.3 3 Pothole 352 4.3 850 14 1.8 1.3 1.81.3 3.6 -1.8 4 Pothole 138 3.0 300 13 1.4 0.5 1.40.5 0.8 0.2 C-1 Pothole 198 1.9 278 10 1.4 0.9 1.40.9 1.3 0.1 5 Pothole 105 2.3 178 11 1.4 0.4 1.40.4 0.4 -0.2 5A Pothole 13 1.4 10 11 1.8 0.7 1.80.7 0.6 -0.2 6 Pothole 38 1.5 43 12 1.3 0.8 1.30.8 1.4 0.2 7 Pothole 105 1.5 101 11 1.5 0.8 1.60.8 2.4 1.5 8 Pothole 121 1.1 91 10 1.5 1.4 1.51.3 2.8 1.2 Mean 1,000 4.3 5,590 15 2.31.8 2.31.8 4.0 -0.4 StD 3,040 4.2 20,000 7 0.81.6 0.81.5 4.3 3.1 Min 1.21 0.3 0.41 7 1.30.4 1.30.4 0.4 -13.1 Summary Statistics Max 14,700 18.0106,000 41 5.16.8 5.16.8 22.8 8.3 *RMSERe l = RMSEV-h normalized by the maximum wetland volume ( Vo). k represents the total number of pool stages.

PAGE 25

13 The methodology and specific details used to obtain the wetland bathymetry as well as the data quality for the pothole wetlands in Saskatchewan, Canada and the prairie pothole wetland located in North Dakota are outlined in the respective studies listed above. The cypress and marsh wetland bathymetry data, st age-storage data and classification were provided by the United States Geological Survey (USGS) (Haag et al. 2005). Haag et al. (2005) provides a complete description of the ba thymetry data quality used in the study. The extent of each wetland was determined us ing biological indicators and the respective wetland perimeter elevations. The lake bat hymetry data were provided as TINs by Mr. Doug Leeper, Senior Environmental Scientis t, Resource Conservation and Development Department, Southwest Florida Water Mana gement District (SWFWMD) (personal communication, January 19, 2006). According to Mr. Leeper, the lake bathymetry data were collected using standard survey equipment (rod/level) or a Global Positioning System (GPS)/Sonar system. The water depths or sediment elevations were measured relative to water level gauges, which are routinely surveyed to check accuracy against known benchmarks within the lake basins. A dditional elevation data for the basins were obtained from SWFWMD aerial photographs in conjunction w ith 0.3048 meter (1.0 foot) contour maps. The elevation data from the ma ps were digitized using ArcMap. The field data and digital elevation data were combined to create TINs using ArcMap. Furthermore, the horizontal tolerance of the su rveyed bathymetry data is reported to be 0.5 meters (1 to 2 feet). The vertical tolera nce of the surveyed data is reportedly within 0.061 meters (0.2 feet) to 0.183 meters (0.6 fe et). The vertical to lerance of the digital elevation data is within 0.152 meters (0.5 f eet). Stage-storage re lationships were later developed from the TINs using ArcMap 3D An alyst, Area and Volume Statistics tool.

PAGE 26

14 Constant interval depths were calculated by dividing the total number of wetland stages ( k ) into the maximum pool depth (Table 2.1) to define the lake stage depths. At each stage, starting at ho, the corresponding planar area and pool volume were calculated using the statistics tool creating the stage-storage profile for the respective lake wetland. The developed stage-storage relationships were then incorporated into this study. 2.2.3. V-h Model Shape Param eter Development The dimensionless wetland fitting or shape parameter ( m ) was calculated using a spreadsheet solver (Microsoft Excel Solver) fo r every wetland in the study. The solver tool was used to minimize the root-mean-squared error (RMSE) between the GISderived, observed or reported volumes ( VGIS) and the V-h Model generated volumes ( VVh) by adjusting the respective wetland shape parameter m [Eq. (2.1)]. The RMSE for the Vh relationship is: k i iVhiGIS VhVV k RMSE1 2)()( 1 (2.2) where RMSEVh [L3] is the RMSE calculated for the V-h Model, i is the wetland or lake pool stage, k is the total number of pool stages, ( VGIS)i [L3] is the wetland or lake volume at stage i as reported in the or iginal articles, and ( VVh)i [L3] is the pool volume produced from the V-h Model [Eq. (2.1)] at stage i For the remainder of the chapter, the RMSE will be reported as a percent of the respective wetland maximum volume ( Vo) to keep the

PAGE 27

15 RMSE comparable with the wide range of wetland and lake volumes: %100*Re oVh lVRMSE RMSE Specific wetland shape parameters were devel oped using two stage-st orage data sets: (1) the upper 80% of the maximum wetland volume (80% Vo), determined by the closest stage data point, and (2) the complete wetland stage-storage data set (100% Vo). The purpose was to evaluate the specific wetland shape parameters developed from the different input data sets. The respective shape parameters were used to reproduce wetland volumes at each stage for all 25 wetlands and 17 lakes discussed in this chapter. 2.3. Methods 2.3.1. Model Evaluation Techniques The RMSE analys is [(Eq. (2.2)] in conjunction with an absolute volumetric error ( VARE) analysis was used to evaluate the predicted wetland storage generated from the V-h Model and general shape parameters. The e rror analyses provide an indication of how well the V-h Model storage prediction matches th e actual wetland storage. It was hypothesized that the V-h Model should perform best on circular bowl shaped wetlands and lakes. However, wetlands and lakes are not generally circular in shape. Even so, these analyses give an overall indication of the fit of the model to the stage-storage characteristics of the wetlands, indirectly ta king into account the devi ation of the wetland shape from a circular bowl. Although the wetl ands and lakes used in this study exhibited a wide variety of shapes, elongated lakes such as oxbows were not included and may need further investigation.

PAGE 28

16 The absolute volumetric error ( VARE) is defined as: %100* )( )()( 11 k i iGIS iGIS iVh AREV VV ABS k V for oVk %20 (2.3) where ABS is the absolute value, and all other parameters are defined in Eq. (2.2). The volumetric error was developed from a range of volumes comprised of the upper 80% Vo, determined by the closest stage data point. The upper 80% Vo was used to calculate the relative volumetric error because the microtopography of the bottom of the wetlands is difficult to know with confidence, particularly when data sets are used from different studies. Further, the relative volumetric error ( VRE) for each wetland was calculated to determine if the V-h Model under or over predicted the actual reported wetland volume. This was accomplished by removing the absolute value term in Eq. (2.3). 2.3.2. General Shape Parameter Development General shape param eters were developed from the specific wetland and lake shape parameters based on the complete we tland stage-storage data set (100% Vo) listed in Table 2.1. The general shape parameters were calculated by combining the specific parameters associated with the various cat egories using three averaging methods: mean, median and volume-weighted. The volume-weighted average ( mV) was calculated as:

PAGE 29

17 n j j o n j j oj t VV Vm m1 1 (2.4) where mV is the volume-weighted average shape parameter value for wetland category t, j is a wetland in wetland category t, mj is the dimensionless shape parameter for wetland j, (Vo)j [L3] is the maximum volume for wetland j, and n is the number of wetlands in the wetland category. These statistical methods were chosen to determine which was the most robust for this application. General shape parameters were developed for each of the four wetland categories and lakes (Table 2.1), and for three different category groupings outlined in Table 2.2. The first group (Case I) consists of the cypress wetlands marsh wetlands and lakes representing the west-central Florida region, the second group (Case II) consists of all 25 wetlands, and the third group (Case III) is comprised of all 25 wetlands and 17 lakes identified in Table 2.1. The groups were chosen to re present west-central Florida regional data, only wetland data, and all we tland and lake data respectively. 2.4. Specific V-h Model Parameters Table 2.1 contains the results of the iterative solver showing the calculated shape parameter for each wetland based on the re spective bathymetry data set (Upper 80% Vo and 100% Vo). The results illustrate that there ar e no significant differences in the shape parameters generated from the two different st age-storage data sets. The average relative

PAGE 30

18 error between the shape parameters produced from each stage-storage data set is only 0.5%. This confirms that minimizing the RMSE between the GIS volumes and V-h Model volumes generate consistent shape parame ters for the stage-storage data sets based on either the upper 80% Vo or the entire data set (100% Vo). The nature of the RMSE analysis weights the larger volumes (upper 80% Vo) more than the smaller wetland volumes (bottom 20% Vo). Since there are minimal differences between the data set shape parameters, the remainder of this study will utilize the shape parameters based on 100% Vo. Furthermore, this analysis shows that the wetland shape factors can be equally developed from the upper 80% Vo potentially eliminating the need to perform expensive and labor intensive detailed bathymetric surveys of the bottom 20% of the wetland. Estimates of the lowest wetland volumes are subject to survey errors and/or noise. These errors are due to small topography variations (micro-topogr aphy) in the bottoms of the wetlands. The respective V-h power-function shape parameter values for each wetland in the study are listed in Table 2.1. The average shape parameter values and corresponding coefficients of variation (CV) for the five wetland categories are: cypress (3.0, 30%), marsh (2.5, 25%), St. Denis NWA (2.1, 9%), pothole (1.5, 11%), and lake (2.5, 35%). The coefficient of variation was calculated for each category from the respective shape parameter mean and standard deviation associated with each category listed in Table 2.1. The average shape parameter and CV for a ll wetlands in the study are 2.3 and 36% respectively. Solving for the shape parameter by minimizing the RMSE between the

PAGE 31

19 GIS-derived or reported volumes and corresponding V-h Model volumes ensured the best overall specific shape parameter and corre sponding wetland stage-storage relationship was found. Additionally, the model statistics (Table 2. 1 Model Evaluation) indicate the normalized RMSE (RMSERel), and the absolute relative volumetric error (VARE) associated with each wetland stage-storage prediction are very clos e. These analyses were performed to evaluate the goodness-of-fit of the V-h Model volumes with the respective GIS-derived wetland volumes. The average RMSERel and VARE developed by the V-h Model for each wetland category are: cypress (2.2%, 5%), marsh (2.5%, 4%), St. Denis NWA (1.1%, 3%), pothole (1.5, 1%), and la ke (2.4%, 6%). The average RMSERel and VARE for the complete wetland data set are only 1.8% a nd 4%. Furthermore, the relative volumetric error analysis (VRE) showed little bias in over or under prediction of the respective wetland volumes. The V-h Model over predicted the wetland storage for 48% of the wetlands and under predicted storage for 52% of the wetlands. Based on these small errors the V-h Model appears robust and adaptive, and was f ound to be accurate for particular shape parameters with relatively small variability. 2.5. V-h Model Parameter Sensitivity Analysis A sensitivity analysis was performed on the respective wetland V-h Model shape parameter (m) and on the maximum pool depth estimate (ho) to better understand the relationships these va riables have on the V-h Model performance. This analysis provides a quantitative measure of how sensitive the V-h Model storage predictions are to the

PAGE 32

20 shape-parameter value-estimate, and how the mo del results are affected by errors in the maximum pool depth. This is important b ecause the shape parameter will need to be estimated when survey data are unavailable, and the maximum pool depth may be determined in a variety of ways (i.e. surveyed or estimated) with varying associated errors. The sensitivity analysis was conducted by adju sting the best-fit shape parameter and the maximum pool depth listed in Table 2.2 by 10% and 20%. Each parameter was adjusted separately to isolate the effect on the V-h Model prediction. The resultant wetland stage-storage relationships were then compared to the specific V-h Model best-fit results. The absolute value of the positive and negative parameter changes were averaged to get a relative sensitivity, a nd evaluated using the corresponding VARE for each wetland. Table 2.2 contains the average VARE generated from the shape parameter (m) and maximum pool depth (ho) estimates for each wetland in the development data set. Adjusting the shape parameters (m) by 10% and 20% increased the average VARE for the data set from 4% (specific m) to 17.5% and 35.7% respectively. Likewise the average VARE increased from 4% (specific m) to 12.8% and 25.8% when the depth parameter (ho) was adjusted by 10% and 20% respectively. This indica tes that the model predicted volumes are more sensitive to the shape para meter (power coefficient) than the maximum depth parameter (linear coefficient); however, as can be seen in Table 2.2, the results indicate strong sensitivity to both values.

PAGE 33

21 Table 2.2. V-h Model shape parameter and maximum pool depth estimate sensitivity analysis. Sensitivity (VARE) ho m m (%) ho (%) Wetland Category (m) 10% 20% 10% 20% West-Central Florida W05 Cypress 0.6 4.317.9 37.3 31.4 70.1 W19 Cypress 0.8 2.517.7 36.9 14.8 30.9 S63 Cypress 0.4 2.719.3 37.7 19.4 35.9 S68 Cypress 0.5 2.016.6 34.6 9.9 20.3 GSC Cypress 0.5 3.518.1 37.7 22.6 48.8 W03 Marsh 1.5 2.816.7 34.5 18.1 37.5 W29 Marsh 0.9 2.017.0 35.3 9.7 20.1 HRSP Marsh 0.7 3.317.2 35.8 21.8 46.8 DP Marsh 2.4 2.515.8 32.8 14.5 30.2 GSM Marsh 0.3 1.816.6 34.4 8.3 16.9 Big Fish Lake 7.3 1.822.8 39.0 17.4 24.2 Bonnie Lake 4.6 2.317.2 36.0 15.3 25.4 Calm Lake 10.02.816.6 34.4 16.6 34.8 Clear Lake 8.8 1.616.6 34.4 7.2 13.2 Garden Lake 8.4 3.416.2 33.8 21.2 45.3 Green Lake 5.6 5.125.4 33.5 35.1 67.1 Jackson Lake 9.8 1.716.4 34.0 8.8 16.0 Letta Lake 6.7 1.517.7 36.9 6.5 12.2 Middle Lake 5.8 1.917.3 36.0 9.4 19.3 Mound Lake 9.7 2.616.6 34.6 14.9 31.2 Mountain Lake 5.1 2.218.7 39.1 14.2 25.4 Neff Lake 7.8 3.117.8 37.1 19.3 41.0 Placid Lake 18.02.617.5 36.4 15.4 32.2 Pretty Lake 8.0 1.716.6 34.6 7.5 15.3 Reinheimer Lake 3.6 2.916.7 34.8 17.1 35.9 Round Lake 8.1 3.216.2 33.7 18.6 39.5 Spring Lake 15.52.216.7 34.8 11.9 24.5

PAGE 34

22 Table 2.2. (Continued). Sensitivity (VARE) ho m m (%) ho (%) Wetland Category (m) 10% 20% 10% 20% St. Denis NWA S92 Pothole 1.2 2.217.7 36.9 11.7 23.5 S104 Pothole 0.7 2.017.5 36.6 10.2 21.0 S109 Pothole 1.2 2.316.9 35.2 12.3 25.5 S120 Pothole 1.1 1.817.7 36.9 8.2 16.9 S125s Pothole 1.0 2.018.2 38.0 9.8 20.2 North Dakota 1 Pothole 2.6 1.517.4 36.2 5.6 10.9 2 Pothole 3.3 1.416.7 34.8 5.0 10.0 3 Pothole 4.3 1.818.1 37.9 8.4 16.2 4 Pothole 3.0 1.417.4 36.2 4.6 9.3 C-1 Pothole 1.9 1.416.4 34.2 4.9 9.8 5 Pothole 2.3 1.416.5 34.3 4.7 9.5 5A Pothole 1.4 1.816.6 34.4 8.5 17.4 6 Pothole 1.5 1.316.8 35.1 4.0 8.2 7 Pothole 1.5 1.618.7 39.0 6.3 12.7 8 Pothole 1.1 1.516.1 33.4 6.8 12.4 Mean 4.3 2.317.5 35.7 12.8 25.8 StD 4.2 0.81.7 1.6 7.0 14.8 Min 0.3 1.315.8 32.8 4.0 8.2 Summary Statistics Max 18.0 5.125.4 39.1 35.1 70.1 2.6. Results and Discussion 2.6.1. General Shape Parameter Analyses The applic ation of this work is to use a single generalized wetland shape parameter in conjunction with V-h Model [Eq. (2.1)] to represent the stage-storage behavior of multiple wetlands and lakes. Table 2.3 lists the generalized shape parameters that were calculated using the three statistic al averages (mean, median and mV) for the four individual wetland categories, the lake category, and for the three wetland groups: Case I cypress wetlands, marsh wetlands and lakes lo cated in west-central Florida, Case II cypress, marsh, St. Denis pothole and North Dakota pothole wetlands, and Case III all

PAGE 35

23 wetlands and lakes identified in Table 2.1. Additionally, the V-h Model performance results produced using the general shape parameters and the wetland category specific shape parameters (specific m) are presented in Table 2.3. The results are listed as summary statistics of the RMSERel and VARE for each wetland category or group. The stage-storage model [Eq. (2.1)] will be referred to as a general V-h Model when general shape parameters are incorporated and as a specific V-h Model when wetland or lake specific shape parameters are incorporated.

PAGE 36

24 Table 2.3. Generalized shape parameter evaluation. Case Data Set General m RMSERel* (%) VARE (%) VOP m StatisticAve StD MinMaxAve StD Min Max (%) Specific m 2.2 1.6 0.6 4.6 4.9 4.0 1.1 10.2 60 3.0 mean 15.814.20.6 36.8 52.954.9 1.3 139.540 2.7 median 14.310.44.6 29.8 45.841.7 10.3 113.460 Cypress 3.0 mV 15.814.20.6 36.8 52.954.9 1.3 139.540 Specific m 2.5 1.9 1.5 5.8 3.6 2.2 1.6 7.1 20 2.5 mean 2.5 median 12.87.2 2.0 19.9 31.620.9 3.7 57.4 40 Marsh 2.5 mV Specific m 2.4 1.7 0.8 6.5 5.7 5.7 1.3 22.8 53 2.5 mean 16.613.92.3 62.3 42.038.1 4.7 173.747 2.3 median 18.016.43.1 74.0 47.047.1 6.7 211.047 Lake 2.2 mV 19.118.21.3 80.8 50.453.3 1.2 232.665 Specific m 1.0 0.5 0.4 1.8 2.7 1.7 0.4 5.2 20 2.1 mean 5.2 2.9 2.3 9.8 12.17.8 5.3 24.9 40 2.0 median 5.2 3.6 1.2 9.5 11.89.0 2.3 22.4 60 St. Denis Pothole 2.1 mV 5.2 2.9 2.3 9.8 12.17.8 5.3 24.9 40 Specific m 0.8 0.3 0.4 1.3 1.6 1.0 0.4 3.6 60 1.5 mean 5.9 3.9 1.7 14.1 15.410.5 3.9 35.7 40 1.4 median 6.1 6.4 1.1 20.0 16.418.3 2.1 50.9 60 N. Dakota Pothole 1.5 mV 5.9 3.9 1.7 14.1 15.410.5 3.9 35.7 40 Specific m 2.4 1.6 0.6 6.5 5.2 4.9 1.1 22.8 48 2.6 mean 15.411.71.2 57.2 40.234.7 1.8 157.544 2.5 median 15.812.60.6 62.3 42.138.3 1.3 173.748 I C,M,L+ 2.2 mV 19.216.61.3 80.8 53.653.5 1.2 232.667 Specific m 1.5 1.3 0.4 5.8 2.9 2.4 0.4 10.2 44 2.1 mean 17.012.62.3 55.6 46.644.6 5.3 208.236 2.0 median 17.214.61.2 61.7 47.751.4 2.3 230.140 II C,M,St.D, N.D+ 1.6 mV 25.325.02.1 94.9 72.284.9 4.1 347.268 Specific m 1.8 1.5 0.4 6.5 4.0 4.3 0.4 22.8 48 2.3 mean 17.512.91.4 74.0 46.439.5 3.4 211.036 2.2 median 17.814.31.380.847.844.8 1.25 232.645 III All 2.0 mV 19.518.11.296.753.857.9 2.3 283.350 + C, M, St.D, N.D, L = Cypress, Marsh, St. Denis NWA and North Dakota wetlands, and Lakes. RMSERel = RMSEV-h normalized by the maximum wetland volume ( Vo). #.# = Optimal general shape parameter ( m ).

PAGE 37

25 The general V-h Model results were obtained by pr edicting the wetland stage-storage relationships for each wetland in a category or group using the general shape parameter developed from each of the three averages. Likewise, the specific V-h Model results (Table 2.3 Specific m) were obtained using the specific wetland shape parameters listed in Table 2.1 (100% Vo). These parameters were used to predict the stage-storage relationships of the individual wetlands and lakes in each category and group. Summary statistics were then calcula ted for the respective general V-h Model predictions and the specific V-h Model predictions for all wetlands in the particular category or group. The general V-h Model performance was evaluated by comparing the respective RMSERel and VARE results (Table 2.3) to that of the specific V-h Model RMSERel and VARE results (Table 2.3 Specific m). The optimal general shape parameter for each wetland category and group was determined by identifying the general V-h Model with the least deviation from the specific V-h Model RMSERel and VARE results. 2.6.1.1. Individual Wetland Categories The general shape parameters developed fr om the three statistical measures (mean, median and mV) had little variation within the indivi dual wetland categories. The largest range of shape parameter values for the individual wetland categories was 0.3 found in the cypress and lake wetland categories (T able 2.3). The cypress general shape parameter ranged from 2.7 (median) to 3.0 (mean and mV), and the lake general shape parameter ranged from 2.2 (mV) to 2.5 (mean). The range of parameter values for the St. Denis and North Dakota pothole wetlands was only 0.1, and th ere was no variation in the general parameters calculated for the marsh wetlands (m = 2.5). Overall, the general

PAGE 38

26 parameter values calculated for the indivi dual wetland categories ranged from 1.4 (North Dakota pothole) to 3.0 (cypress) In order to gain some perspective of the errors associated with this shape parameter range, the RMSERel and VARE were calculated for all the wetlands in this stud y (Case III). The average RMSERel and VARE associated with each general shape parameter are: 42.2% and 120.4% (m = 1.4) and 19.9% and 48.4% (m = 3.0). Another analysis was performed to determine the number of wetlands needed to develop an effective general shape parameter for use in the V-h Model. For this exercise, the lake and pothole wetland categories (St. Denis and North Dakota) were chosen because they had the most entries, 17 lakes and 15 pothol e wetlands (Table 2.1). The number of entries used to calculate the general shape parameter was incrementally increased from two to the maximum number of entries in each data set. The order of the lakes and pothole wetlands were randomly chosen before the calculations were performed. This procedure was repeated several times for each data set. The mean shape parameter for the lakes became constant, m 2.5, when the data set was comprised of five to 10 lakes. Also, the mean shape parameter for the pothole wetlands became constant, m 1.7, when the data set was comprised of five to 10 wetlands. The specific number of entries required to stabilize the shape parameter varied based on the order of selection. If the first entries had a shape para meter near the mean, fewer en tries were needed; conversely if the first entries had a shape parameter far from the mean, more were needed to reach a stable value. This analysis suggests that a minimum of five to 10 wetlands might be

PAGE 39

27 needed to develop a general shape parameter(s ) that can be used to describe the stagestorage characteristics of multiple wetlands and lakes. 2.6.1.2. Wetland Groups The generalized shape param eters for the Case I, Case II and Case III wetland groups are listed in Table 2.3. The Case I general shape parameters ranged from 2.2 (mV) to 2.6 (mean), the Case II shape parameters ranged from 1.6 (mV) to 2.1 (mean), and the Case III shape parameters ranged from 2.0 (mV) to 2.3 (mean). The largest general shape parameter range was 0.5 found in Case II, wh ich was comprised of all wetlands except lakes. However, the Case III scenario, consisting of all 42 wetland and lakes, had a reduced range of general shape parameters (0.3). 2.6.2. General V-h Model Performance 2.6.2.1. Individual Wetland Categories The RMSERel and VARE analysis results generated from the general V-h Model predictions and the specific V-h Model predictions are presented in Table 2.3. The optimal general shape parameters found for each wetland category are highlighted in grey. The optimal general shape parameters th at produced the smallest RMSERel and VARE difference from the specific V-h Model results for each wetland category were: cypress (2.7), marsh (2.5), lake (2.5), St. Denis pothole (2.0), and Nort h Dakota pothole (1.5). The mean statistic produced the best stage-storage results for the marsh, lake and pothole wetland categories; while the median statistic produced the best storage results for the cypress and St. Denis pothole wetland categories. The respective average RMSERel and VARE

PAGE 40

28 associated with each identified optimal shape parameter, hence the optimal V-h Models, were: cypress (14.3% and 45.8%), marsh (12. 8% and 31.6%), lake (16.6% and 42.0%), St. Denis pothole (5.2% and 11.8%), and North Dakota pothole (5.9% and 15.4%). Comparatively, the average RMSERel and VARE baseline (Table 2.3 Specific m) values generated from the specific shape parameters were: cypress (2.2% and 4.9%), marsh (2.5% and 3.6%), lake (2.4% and 5.7%), St Denis pothole (1.0% and 2.7%), and North Dakota pothole (0.8% and 1.6%). The optimal shape parameters for the cypr ess wetlands, marsh wetlands and lakes of west-central Florida are very similar, m 2.5. It is interesting to note that the lake storage behavior is similar to the cypre ss and marsh behavior, although the lakes are much deeper and larger than the cypress and marsh wetlands and exhibit a larger diversity in depth and storag e (Table 2.1). Further, the St. Denis pothole and North Dakota pothole wetlands differ from the we st-central Florida wetlands. The optimal shape parameter for the St. Denis pothol e and North Dakota pothole wetlands were m = 2.0 and m = 1.5 respectively. Because the shap e parameter describes the stage-storage relationship of a wetland, it is reasonable to expect different shape parameters for wetlands that are formed by different mechan isms. Hence, on average the St. Denis and North Dakota pothole wetlands appear to have steeper storage profiles than the westcentral Florida wetlands and lakes. The cypress wetlands appear to have the shallowest stage-storage profile of all the wetlands in this study. This is to be expected since cypress wetlands need to be somewhat shallow for the cypress seeds to germinate, and to support cypress knob root structures (Mitsch and Ewel 1979).

PAGE 41

29 Overall, the general shape parameters calcula ted for the individual wetland categories did a reasonable job predicting the individual wetland stage-storage behavior. The optimal V-h Models developed for the St. Denis pothole wetlands (m = 2.0) and North Dakota pothole wetlands (m = 1.5) produced the best resu lts for the individual wetland categories. The respective average RMSERel and VARE for both categories were: 5.2% and 11.8% (St. Denis), and 5.9% and 15.4% (North Dakota) (Table 2.3). This is due, in part, to the small shape parameter distribution in each wetland category (Table 2.1). The shape parameter variance expressed as the coe fficient of variation (CV) for the St. Denis and North Dakota pothole wetland categories are: 9% and 11%, respectively. The small parameter variance could be attributed to th e similar wetland topograp hy in the respective category. The cypress wetland, marsh wetland and lake optimal V-h Model storage predictions were not as good. The average RMSERel and VARE for each category were: 14.3% and 45.8% (cypress), 12.8% and 31.6% (marsh), and 16.6% and 42.0% (lakes) (Table 2.3). Again, this is due in part to the shape parameter variability associated with each category. The variability (CV) in the individual wetland shape parameters was: 30% (cypress), 25% (marsh), and 35% (lake). 2.6.2.2. Wetland Groups Table 2.3 also lis ts the RMSERel and VARE general V-h Model analysis results for each wetland group (Cases I, II, III). Again, the RMSERel and VARE were calculated from the respective general shape parameters and compared to the specific V-h Model results for each wetland group. The optimal general shape parameters found for each wetland group are highlighted in grey. The optimal general shape parameters that produced the smallest

PAGE 42

30RMSERel and VARE were: 2.6 (Case I), 2.1 (Case II), and 2.3 (Case III). The average RMSERel and VARE produced from the optimal V-h Model for each case were: 15.4% and 40.2% (Case I), 17.0% and 46.6% (Case II ), and 17.5% and 46.4% (Case III). Comparatively, the average RMSERel and VARE produced from the specific V-h Models (Specific m) were: 2.4% and 5.2% (Case I), 1.5% a nd 2.9% (Case II), and 1.8% and 4.0% (Case III) (Table 2.3). Again, the elevated e rrors can be attributed to the diverse nature of the wetlands incorporated in each group. The variability (CV) in the individual wetland shape parameters for each group was: 32% (Case I), 35% (Case II) and 36% (Case III). Figure 2.1 contains a series of panels s howing representative volume reproductions for the west-central Florida wetlands, St. Deni s pothole wetlands, North Dakota pothole wetlands, and lakes analyzed in this study. The graphs co mpare the reported volumes (Actual), and the V-h power-function model stage-storage reproduc tions based on the specific wetland shape parameter found in Table 2.1 (100% Vo), and on the Case III (all wetlands and lakes) optimal ge neralized shape parameter, m = 2.3. Figure 2.1 illustrates the goodness-of-fit of the general V-h Model predictions (Case III) and the specific V-h Model storage predictions to the actual wetland volumes. Overall the general V-h Model predicted the individual wetland st orage behavior rather well. The largest deviation from the actual storage occurred with the cypress wetland (W05). According to Haag et al. (2005) this cypress wetland has a large surface area (35,500 square meters) but an intermediate maximum depth of 0.6 meters. The re latively shallow depth was due in part to a thick layer of organics and flocculent se diment that accumulated on the floor of the

PAGE 43

31 wetland. The organic rich material may have filled in the deeper parts of the wetland basin, impacting the shape factor.

PAGE 44

32 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 2000 4000 6000 8000 10000 h (m)Volume (m3)a) Cypress Wetland (W05) Actual Specific m General m 0 2 4 6 8 10 0 0.5 1 1.5 2 2.5 3 x 106 h (m)Volume (m3)d) Lake (Calm) Actual Specific m General m 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 500 1000 1500 2000 h (m)Volume (m3)b) St. Denis Wetland (S92) Actual Specific m General m 0 0.5 1 1.5 0 2 4 6 8 10 12 x 104 h (m)Volume (m3)c) Pothole Wetland (P7) Actual Specific m General m Figure 2.1 Representative model prediction comparisons for a) West-central Florida wetlands, b) St. Denis NWA wetlands, c) Pothole wetlands, and d) Lakes.

PAGE 45

33 2.6.2.3. Volume Over Prediction (VOP) Analysis An analysis identifying the number of wetlands in which the general V-h Models over predicted the wetland storage (VOP) in each category and group is shown in Table 2.3. The VOP is defined as the ratio of the number of wetlands the V-h Model over predicted the storage to the total number of wetlands in each category expressed as a percent (Table 2.3 VOP). The VOP for the specific V-h Model predictions a nd the optimal general V-h Model predictions for each wetland category we re: cypress (60% vs. 60%), marsh (20% vs. 40%), lake (53% vs. 47%), St. Denis pothole (20% vs. 60 %), and North Dakota pothole (60% vs. 40%). Additionally, the VOP for the specific V-h Model predictions and the optimal general V-h Model predictions for the wetla nd groups were: Case I (48% vs. 44%), Case II (44% vs. 36%), and Case III (48% vs. 36%). On average, both the specific and optimal general V-h Models showed little bias in over or under prediction of the respective wetland volumes. 2.7. Validation Data Set Analysis An application of the V-h Model and method is presented here. The model was used to predict the storage behavior of an independent data set comprised of 21 lakes in westcentral Florida (Table 2.4). Based on the method, two parameters for each lake must be estimated or calculated. Fi rst the maximum pool depth (ho) must be determined. This can be accomplished via a survey or estimation. Second the shape parameter (m) describing the stage-storage characteristics of the lake must be determined. The shape parameters can be calculated using the pro cedure outlined in Nil sson et al. (2008) if survey data is available, or from the parameters developed and pr esented in this study

PAGE 46

34 (Tables 2.1 and 2.3). Once the maximum pool depth and shape parameter for each lake are determined they can be incorporated into the V-h Model [Eq. (2.1)] to calculate the respective lake stage-storage characteristics.

PAGE 47

35 Table 2.4. Lake validation data set for gene ral shape parameter evaluation. Characteristics General V-h Model Evaluation (%) Ao ho Vo m = 2.3 m = 2.5 Wetland Category X103 m2 m x103 m3 k RMSERel* VARE VRE hARE hRE RMSERel* VARE VRE hARE hRE Armistead Lake 145 7.9302 1437.3 112.894.0 27.1 -22.6 29.2 88.1 73.4 21.7 -18.0 Brant Lake 256 5.5571 106.9 21.4 21.4 7.5 -7.5 3.6 8.0 5.2 2.8 -1.7 Carlton Lake 332 3.0488 6 7.1 19.4 19.4 10.0 10.0 11.6 31.1 -31.1 16.8 16.8 Chapman 2 Lake 52 1.846 4 7.6 26.6 -26.6 16.5 16.5 12.1 37.4 -37.4 24.3 24.3 Chapman Lake 165 5.5345 1010.5 37.5 37.5 12.2 -12.2 5.0 19.1 19.1 6.4 -6.4 Commiston Lake 63.8 7.3173 1312.4 39.6 39.6 12.8 -12.8 6.9 21.5 21.5 7.1 -7.1 Deer Lake 142 9.8479 1718.6 63.9 63.9 17.9 -17.9 12.6 43.3 43.3 12.4 -12.4 Elaine Lake 9.41 2.113.5 8 23.6 50.8 -42.4 38.2 31.9 26.8 57.5 -47.9 43.9 36.5 Elizabeth Lake 77.0 7.3251 131.7 2.7 -1.5 1.2 0.7 5.9 14.9 -14.9 6.7 6.7 Fleur Lake 12.1 5.515.4 1046.5 136.2113.5 30.8 -25.7 37.4 105.387.7 24.8 -20.6 George Lake 109 7.3350 131.4 4.0 -3 .5 1.8 1.6 6.3 16.9 -14.8 7.8 6.8 Glass Lake 85.2 6.7146 1239.9 142.8142.8 30.3 -30.3 31.5 112.2112.224.7 -24.7 Grace Lake 65.1 6.1185 114.8 12.5 10.7 6.0 5.2 9.4 24.1 -20.7 11.9 10.2 Joseph Lake 198 6.7371 1235.7 112.389.9 26.5 -21.2 28.0 88.4 70.7 21.3 -17.0 Little Deer Lake 40.1 6.183.4 1115.6 43. 3 36.1 14.3 -11.9 9.3 25.4 21.2 8.6 -7.2 Lutz Lake 26.9 5.553.3 1013.0 42.7 42. 7 13.6 -13.6 7.1 23.6 23.6 7.7 -7.7 Mead Lake 59.9 2.462.1 5 4.0 11.1 -7.3 5.6 3.7 7.7 21.5 -16.1 11.1 8.3 Platt Lake 256 4.3427 8 8.0 22.2 17.8 8.0 -6.4 3.4 6.3 5.1 2.3 -1.8 Rocket Lake 13.2 4.321.8 8 12.0 55.8 55.8 14.6 -14.6 7.3 34.2 34.2 8.9 -8.9 Starvation Lake 66.3 2.469.9 9 1.7 6.8 6.6 2.6 -2.5 4.5 8.2 -8.2 3.6 3.6 Wastena Lake 74.9 7.3218 138.5 26.3 26.3 8.8 -8.8 4.4 10.4 9.9 3.4 -3.2 Mean 107 5.5222 10 15.1 47.2 32.2 14.6 -6.6 12.9 38.0 16.0 13.2 -1.1 StD 91.6 2.2181 3 13.9 43.9 49.3 10.6 15.2 10.7 33.0 43.6 10.3 15.5 Min 9.41 1.813.5 4 1.4 2.7 -42.4 1. 2 -30.3 3.4 6.3 -47.9 2.3 -24.7 Summary Statistics Max 332 9.8571 1746.5 142.8142.8 38.2 31.9 37.4 112.2112.243.9 36.5 RMSERel = RMSEV-h normalized by the maximum wetland volume ( Vo). k represents the total number of pool stages or pool volume increments.

PAGE 48

36 In this application, general shape parameters presented in Table 2.3 were used to predict the storage behavior of the 21 lakes in west -central Florida. The maximum pool depths were obtained from surveys. The validation data set (Table 2.4) was used to evaluate the general V-h Model predictive capabilities on a unique lake group. Two measures were chosen to evalua te the model: (1) storage prediction (V) and (2) stage prediction (h). The storage predictions were performed and evaluated using the techniques outlined in the Theory and Me thodology section of the manuscript. Two general shape parameters, m = 2.3 and m = 2.5, were incorporated into the V-h Model [Eq. (2.1)] and used to develop the storage ch aracteristics of each lake. The first was the optimal shape parameter for the 42 wetlands and lakes in the development data set (Table 2.3 Case III), and the second was the optim al shape parameter for the west-central Florida lake category (Table 2.3 Lake). Th e lake general parameter was chosen to see if the optimal parameter specific to the lake category performed better than the overall general parameter developed fr om the entire data set. The stage prediction was added to the evaluation because hydrologic models such as HSPF require a defined relationship between stage, storage and discharge. This technique could be used to popul ate a table relating stage to storage volumes. The stage predictions were performed by rearrangi ng Eq. (2.1), solving for the stage (h) given a specific volume (V):

PAGE 49

37 m oo om hA V hVh1)( (2.5) Again the general shape parameters, m = 2.3 and m = 2.5, were used in the analysis. Once the stage profile correspond ing to given volumes was obtained, the absolute relative stage error ( hARE) was calculated and used to evalua te the model performance. The hARE was calculated by replacing the volume parameter ( V ) in Eq. (2.3) with a stage parameter ( h): %100* )( )()( 11 k i iGIS iGIS iVh AREh hh ABS k h for oVk %20 (2.6) where ABS is the absolute value, i is the lake pool volume increment, k is the total number of pool volume increments, ( hGIS)i [L] is the known lake GIS-derived stage at lake volume i and ( hVh)i [L] is the lake stage produced from the rearranged V-h Model [Eq. (2.5)] at lake volume i Furthermore, the relative stage error ( hRE) for each lake was calculated to determine if the V-h Model under or over predicte d the actual reported lake stage, which was accomplished by removing the absolute value term in Eq. (2.6).

PAGE 50

38 2.7.1. Validation Data Set The 21 lake validation d ata set was devel oped out of 100 lakes found on the Hillsborough County Watershed Atlas (Florida Center for Community Design and Research 2007). The lakes were grouped into three size cat egories based on maximum volumes and seven lakes were randomly selected from each size category for the valida tion data set. The lake bathymetry data were collected using a SONAR dept h finder along with a Global Positioning System (DGPS). Triangulated Irregul ar Networks (TINs) were created from the digital survey data using ArcMap (Envi ronmental Systems Research Institute Inc. (ESRI) 2007), which were subsequently used to create 0.6096 meter (2.0 foot) contour maps. The contour maps were downloaded from the Hillsborough County Watershed Atlas web site and converted to raster images, cell size of 0.061 meters (0.2 feet). Stagestorage relationships were developed from the raster images using the ArcMap 3D Analyst, Area and Volume Statistics tool. C onstant interval depths were calculated by dividing the total number of lake stage values ( k ) into the maximum pool depth (Table 2.4) to define the lake stage dept hs. At each stage, starting at ho, the corresponding open water surface area and pool volume were calcu lated using the GIS 3D Analyst tool. The developed stage-storage relationships were th en used as the reference values for error analyses. Table 2.4 identifies the lakes in the validation data se t and contains the maximum pool areas ( Ao), pool depths ( ho) and pool volumes ( Vo) for each lake. The summary statistics of Ao, ho and Vo for the validation data set are listed at the bottom of Table 2.4.

PAGE 51

39 2.7.2. General V-h Model Performance with Validation Data 2.7.2.1. Storage Predictions ( V ) The average RMSERel and VARE corresponding to the storage predictions for the general shape parameter m = 2.3 were: 15.1% and 47.2% resp ectively (Table 2.4). The VARE ranged from 2.7% to 142.8% with a standard deviation of 44%. The average RMSERel and VARE corresponding to the storage predicti ons for the general shape parameter m = 2.5 were: 12.9% and 38.0% respectively (Table 2.4). The VARE ranged from 6.3% to 112.2% with a standard deviation of 33%. Additionally, the general V-h Model over predicted (positive VRE) 67% of the lake volumes with m = 2.3, and 62% of the lake volumes with m = 2.5 (Table 2.4 VRE). 2.7.2.2. Stage Predictions (h) The average absolute relative stage error ( hARE) corresponding to the stage predictions for m = 2.3 was 14.6% (Table 2.4). The hARE ranged from 1.2% to 38.2% with a standard deviation of 10.6%. The average hARE corresponding to the ge neral shape parameter storage predictions for m = 2.5 was 13.2%. The hARE ranged from 2.3% to 43.9% with a standard deviation of 10.3%. Additionally, the general V-h Model over predicted (positive hRE) 33% of the lake stages with m = 2.3, and 38% of the lake stages with m = 2.5 (Table 2.4 hRE). In general, the error in predicting stages was much lower than that for the predicted volumes.

PAGE 52

40 2.7.3. Discussion of the Validation Application The use of a general shape param eter and the V-h Model to predict the storage characteristics of an independent lake data set was very promising even though some relatively high errors were observed for specific lakes. The average VARE produced by the optimal lake general parameter ( m = 2.5) and the Case III optimal general parameter ( m = 2.3) might seem large to some; however, th is error may be acceptable since hydrologic modelers often do not have any wetland storage data to incorporate into a model, and consequently, no useful way to estimate the error introduced fr om a set of wetland assumptions. Models are often made of areas where significant la kes, reservoirs and wetlands are present with inadequate stagestorage survey information. Given this consideration, the average stage error hARE produced by these shape parameters might be very reasonable. Modelers do need to be aware of the large variation in the VARE and hARE when utilizing this technique. The variability expressed as the coefficient of variation (CV) in the VARE was 93% ( m = 2.3) and 87% ( m = 2.5). The variability (CV) in the hARE was slightly lower with 73% ( m = 2.3) and 78% ( m = 2.5) (Table 2.4). Last, the results for lake storage prod uced by the general lake parameter ( m = 2.5) had slightly less variation an d an overall lower average VARE and hARE than the results produced by the more general Case III parameter, m = 2.3 (Table 2.4). This is an indication that specific we tland category or lake genera l shape parameters can be developed that will predict, within an im proved and possibly acceptable error level, the storage characteristics of wetlands and la kes that have not been surveyed.

PAGE 53

41 2.8. Conclusions The goal of this chapter was to provide water resource en gineers and hydrologists the analytical tools needed to estimate wetland and lake water storage where data does not exist. An analytical technique utilizing a power-function model that is based on a single fitting parameter and two physically-derived parameters was developed to describe wetland and lake storage in th e absence of detailed bat hymetry for use in hydrologic modeling studies. The model u tilizes two readily obtainable physical characteristics: the maximum or representative we tland or lake planar area Ao, and the corresponding maximum pool depth ho. Best-fit dimensionless shape pa rameters describing wetland and lake stage-storage relationships were deve loped using an iterative procedure for known bathymetry data sets. There was little diffe rence between the specific shape parameters generated from the data sets based on the higher end subset, 80% of the maximum pool ( Vo), and the complete data set, 100% Vo. Hence, the V-h power-function model (V-h Model) could be based on the bulk of the storage data, i.e. the upper 80% Vo, potentially eliminating the expensive and labor intensive effort required to precisely survey the intricate and perhaps inaccessible wetland or la ke bottom. Furthermore, this approach produced the optimal shape parameter (minimum RMSE) for each wetland which accurately described the individual stage-stor age relationships with an average absolute relative volumetric error less than 5.0 % for all wetlands in this study. The application of the aforementioned V-h power-function model was to predict the storage behavior of multiple wetlands and lake s in the absence of detailed survey data. General shape parameters describing wetland and lake stage-storage relationships were

PAGE 54

42 developed and evaluated for four wetland categories, 17 lakes, and three different category groupings. The general shape para meters were incorporated in to the V-h Model and used to predict individual wetland and lake storage characteristics. Furthermore, the predictive capabilities of two general V-h Models were evaluated on an independent validation data set consisting of 21 lakes. Last, the study provided insight into the magnitude of error associated with this shape parameter and method. Overall the general V-h Models predicted the indivi dual wetland and lake storage behavior well. The average relative volum etric predictive errors ranged from 11.8% (St. Denis pothole wetlands) to 46.4% for the category group comprised of all 42 wetlands and lakes in the study. These results were s ubstantiated by the valid ation data set stagestorage predictions. The average relative vo lumetric error and average relative stage error produced by the general V-h Models on the lake validat ion data set were 43% and 14% respectively. It should be noted there was high variability in all of the prediction results. These errors may be acceptable for estimating storage volumes considering hydrologic study areas might be comprised of many wetlands and lakes for which no storage data or representative parametric models are available. One of the benefits of this method is that errors associat ed with the storage model have been quantified. Hydrologic modelers can utilize these errors as they evaluate their confid ence in the model results. In lieu of an analytical met hod, many hydrologic modelers are forced to quantify wetland storage with a guess.

PAGE 55

43 The predictive capability of the general V-h models, i.e. general shape parameters, was affected by the diversity of the wetland or lake topography asso ciated with an individual wetland category, lake category, or category group. The differences in the wetland or lake topography were due to diverse mechanisms of formation. In general, it appears that five to 10 wetlands are necessary to produ ce a useful general shape parameter. This work demonstrates that a single wetland shape parameter could be used to represent the storage of multiple wetlands and/or lake s with acceptable and quantifiable error in field, theoretical and modeling studies. A specific example is the stage-volume relationship needed for an HSPF f-table as the errors associat ed with stage estimates are low. Additionally, the V-h power-function model shape parameter(s) could be used by modelers as a calibration f actor in hydrologic models; as opposed to individually adjusting rating relationship te rms thereby easing calibration difficulty and reducing over parameterization. There are many other types of wetlands such as bogs, mangrove swamps, estuarine and tidal wetlands, arctic wetland areas, playa lakes, vernal pools, and riparian areas, to name a few that were not covered in this study. Shape factors could be developed for each of these wetland types and used to develop storag e models. The expansion of the data set to include any of these wetlands would help in the development of specific shape parameters that could be used to define th e stage-storage relations hip for a particular wetland category.

PAGE 56

44 2.9. Acknowledgements I graciously acknowledge the contributions of Terrie M. Lee, Unite d States Geological Survey, W ater Resources Division, who provide d the detailed wetland bathymetry data for the cypress and marsh wetlands and Doug Leeper, Senior Environmental Scientist, Resource Conservation and Development Department, Southwest Florida Water Management District, who provided the TIN da ta for the lakes. Additionally, we would like to thank Dr. Ahmed Said, ECT, and Dr Jeff Geurink, Water Resource Engineer, Tampa Bay Water, for providing valuable ad vice throughout the development of the two papers emanating from this chapter (Nilss on et al. 2008; Nilsson et al. In Press).

PAGE 57

45 CHAPTER 3 HYDROLOGIC CHARACTERIZATION OF 56 WEST-CENTRAL FLORIDA ISOLATED WETLANDS USING A PROBABILISTIC METHOD 3.1. Introduction The hydrologic characterization of wetlands re quires the use of long-term data records and the characterization of climatic variability to describe the durat ion of different waterlevel elevations in the wetland when it is flooded, and the duration of different watertable elevations below the wetland when th ey are dry. Long-term data records are important because short-term records (months year) might not adequately represent the hydrologic trends of the wetland, nor adequately explain the variabil ity in the wetland surface and subsurface wate r-level fluctuations. This variability is attributed to the inherent randomness in the driving variables, i.e. climatic variability (precipitation and evapotranspiration), the hydrologic system (topogr aphic, aquifer, and soil characteristics) (Bras and Rodrguez-Iturbe 1993; Maidment 1993), and anthropogenic stresses such as groundwater pumping and surface water augm entation associated with the wetland. Further, short-term records may be indicativ e of a purely transient response to the shortterm climatic variability.

PAGE 58

46 Monitoring the pooled (surface) water as well as the subsurface water-table fluctuations associated with wetlands is critical in understanding and eval uating the hydrologic response of the wetland. High variability in the surface and subsurface water level fluctuations are indicative and possibly respons ible for the distinct differences exhibited by different wetland types in shallow water-table environments, such as Florida. In westcentral Florida, the water-table is typically within a few of meters of land surface for a good portion of the year creating a situation where this shal low water-table environment can influence surface water bodies such as rivers, lakes and wetlands. Additionally, almost all of the wetlands in Florida spend part of the year with no standing surface water (dry bed) due to the seasonal precipitati on patterns, and in some cases due to anthropogenic influences such as groundwater pumping (Dahl 2005). Even when there is no standing water in the wetlands, existing phreatophyte plant communities continue to draw water from the proximal groundwater system supporting evapotranspiration. Water levels in and around we tlands control the plant species present in wetlands, and may be crucial in determining the sp eciation of the different wetland types (Hammersmark et al. 2009). The distribution of plant species associated with wetlands are assumed to be a function of environmen tal gradients, such as groundwater levels (Rains et al. 2004). Rains et al. (2004) simulated the mean depth to groundwater for several plant communities: grassland (-1.06 m), riverine forest (-0.77 m), sedge meadow (-0.48 m), willow forest (-0.27 m), and em ergent marsh (0.07 m above the ground surface). The study provides examples that diverse plant communities can be associated with small differences in groundwater depths.

PAGE 59

47 Wetland stage and groundwater levels have been collected over time in numerous wetlands for various studies, e.g. Bradley (2002), Hayashi et al (1998), Johnson et al (2004), and Wise et al. (2000). However, th ese studies and others mainly focused on a single or particular t ype of wetland behavior. Generall y, the emphasis was to provide a detailed description of the water-level fl uctuations and surface-water storage for a particular wetland, and did not attempt to ge neralize these descriptions across a broad range of wetland types or categories. Lee et al. (2009) conducted an extensive four-year study comparing the hydrology, water quality and ecology of 10 isolated wetlands in west-central Florida. They recognized the importance of studying multiple wetlands to compare and contrast the hydrolog ic character. This study, in effect, builds on Lee et al. by utilizing a unique and extensiv e set of water elevations in 56 various isolated wetlands located in west-central Florida. The seve n year study utilized water elevation data associated with paired wetland and upland monitoring wells associated with each wetland. The hydrologic analysis and understanding of long-term wetland water levels requires reducing the data into speci fic patterns. According to Nestler and Long (1997) hydrologic patterns can be used to understand nearly all significant wetland processes. Various statistical techniques can be used to reduce wetland water level data into patterns for analysis. For instance, simple indices a nd summary variables such as the statistical mean, median and ranges, describe measures of central tendency and dispersion, however they provide a low resolution description summary of complex hydrological patterns (Nestler and Long 1997). Advanced statistics and probabilistic methods can provide a

PAGE 60

48 more robust means for analyzing the comple x hydrological patterns associated with wetlands. Empirical distribution functions representing frequencies of occurrence are one of the advanced statistical technique s used to analyze hydrologi c data (Maidment 1993). The distributions portray ordered or ranked sample data as relative frequencies. In short, the empirical distribution function is a graphical display that provides information about how data values are distributed in relation to other values (Maidment 1993). For example, frequencies and probabilities of occurrences over time of sample data can be used to determine the probability of a response being less than a ce rtain value, greater than a certain value, or between two values (Hogg and Ledolte r 1987; Weisstein 2009a). Furthermore, empirical distribution functions are used in hydrology to: 1) compare two or more data distributions, 2) compare data to a theoretical distribution such as the normal distribution, and 3) calculate the fr equencies of exceedance (Maidment 1993). The objective of this chapter is to char acterize wetland surface and subsurface waterlevels based on the probability of inundation and the freque ncy distribution of the depth to the water-table. Specifically, empirical distribution functions (E DFs) were developed using water elevations associated with 56 various isolated wetlands in west-central Florida. The water elevations were record ed over a seven year period from January 2001 through December 2007. The empirical distri bution functions were used to identify representative wetland non-exceedance probabil ity percentiles for the direct hydrologic comparisons of various wetland types, to identify the period of wetland inundation, and

PAGE 61

49 to identify the percent of time that high wate r marks in wetlands were experienced or exceeded. Further, the empirical distribut ion functions were used to compare the different wetland categories for unique hydrol ogic characteristics or responses, and possibly to identify individual wetlands that may be under adverse hydrologic stresses. 3.2. Description of Study Area The wetlands used in this study are located in west-central Florid a within the boundaries of the South W est Florida Water Management District (SWFWMD) and spread across parts of six counties (Figure 3.1). The climat e of the study area is humid and subtropical. The regional geology is comprised of a mantle d karst terrain which is characterized by numerous sinkholes brought about by the disso lution of the underlying limestone (Lee et al. 2009). The wetlands are located within three physiographic regions: Northern Gulf Coastal Lowlands (NGCL), Lake Upland (L U) and Western Valley (WV) outlined in Table 3.1. These regions are characterized by a relatively high water-table and are underlain by the Upper Floridan aquifer (Haag et al. 2005; Lee et al. 2009).

PAGE 62

50 ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !# # # # # # # # #" " "X W X W X W [ [ [CROSS BAR RANCH STARKEY SOUTH PASCO ELDRIDGE WILDE MORRIS BRIDGEPASCO POLK HILLSBOROUGH HERNANDO SUMTER LAKE PINELLAS WellFields GreenSwamp Physiographic Region LAKE UPLAND NORTHERN GULF COASTAL LOWLANDS WESTERN VALLEY Cypress wtlds#Marsh wtlds"CypM wtldsX WHardwood wtlds[Wet Prairie wtlds 01 02 0 5 Kilometers4 Figure 3.1 Study wetland locations in the northern Tampa Bay regi on of west-central Florida.

PAGE 63

51 ! ! !! ! ! ! ! ! ! !# # # # # #" " "X W X W [STARKEY SOUTH PASCO ELDRIDGE WILDE MORRIS BRIDGEPASCO HILLSBOROUGH PINELLAS 89 70 51 544 505 493 489 320 301 276 261 252 215 201 165 143 136 112 3713 4352 21 407 384 304 183 3715 20 501 379 331 154 1332 1316 1317 WellFields GreenSwamp Physiographic Region LAKE UPLAND NORTHERN GULF COASTAL LOWLANDS WESTERN VALLEY Cypress wtlds#Marsh wtlds"CypM wtldsX WHardwood wtlds[Wet Prairie wtlds 061 2 3 Kilometers4 Figure 3.1 (Continued).

PAGE 64

52 ! ! ! ! ! ! ! !# # #X W [ [POLK PASCO LAKE SUMTER HILLSBOROUGH HERNANDO 84 388 541 295 196 170 4187 4184 3344 1335 1327 1326 1324 1323 1319 81 605 1320 1329 1325 WellFields GreenSwamp LAKE UPLAND NORTHERN GULF COASTAL LOWLANDS WESTERN VALLEY Cypress wtlds#Marsh wtlds"CypM wtldsX WHardwood wtlds[Wet Prairie wtlds 048 2 Kilometers4 Figure 3.1 (Continued).

PAGE 65

53 Table 3.1. Wetland identification and physical characteristics. Wetland Wetland Physical Characteristics Location Physiographic ID DB (m) NP (m) A (x103 m2) P (x102m) Type Region 20 12.4 13.1 59.7 14.8 CM MB WV 21 8.0 9.4 392.9 31.2 M S NGCL 51 7.9 8.8 6.9 3.1 C EW NGCL 70 13.4 14.1 31.1 8.7 C S NGCL 81 22.0 24.4 25.1 7.4 M UHFDA WV 84 30.9 31.3 4.6 2.5 C GS LU 89 13.6 13.9 188.9 38.8 C S NGCL 112 11.9 12.7 18.2 6.1 C S NGCL 136 13.9 14.3 143.5 18.3 C S NGCL 143 12.3 12.9 80.6 18.4 C S NGCL 154 21.9 23.3 52.3 9.5 CM CBR NGCL 165 11.7 12.2 34.0 9.4 C MB WV 170 30.3 30.6 7.8 3.3 C GS WV 183 10.0 10.9 9.5 3.8 M MB WV 196 29.4 29.9 4.0 2.4 C GS WV 201 11.0 11.6 34.9 7.9 C MB WV 215 13.5 14.3 32.1 7.0 C S NGCL 252 10.9 11.6 8.6 3.4 C S NGCL 261 3.5 4.6 12.0 4.1 C NGCL 276 17.0 18.1 7.3 3.2 C SP NGCL 295 30.9 31.4 10.4 3.9 C GS LU 301 4.0 4.5 3.3 2.1 C EW NGCL 304 13.5 14.3 1.1 1.5 M S NGCL 320 12.4 12.9 25.3 6.0 C MB WV 331 12.3 13.7 15.5 4.6 CM S NGCL 379 9.8 10.5 28.2 6.4 CM MB WV 384 12.3 12.9 13.3 4.2 M MB WV 388 30.0 30.6 3.0 2.0 C GS WV DB = Dry-bed elevation GS = Green Swamp Wildlife Management Area NP = Normal Pool elevation UHFDA = Upper Hillsborough Flood Detention Area Wetland Identification Well Field Identification C = Cypress wetland CB = Cypress Bridge CM = Cypress-Marsh wetland CBR = Cross Bar Ranch H = Hardwood wetland EW = Eldridge Wild M = Marsh wetland MB = Morris Bridge WP = Wet Prairie wetland S = Starkey Physiographic Region SP = South Pasco LU = Lake Upland NGCL =. Northern Gulf Coastal Lowland WV = Western Valley

PAGE 66

54 Table 3.1. (Continued). Wetland Wetland Physical Characteristics Location Physiographic ID DB (m) NP (m) A (x103 m2) P (x102m) Type Region 407 10.5 11.2 4.5 2.5 M MB WV 489 11.7 13.2 34.8 9.4 C S NGCL 493 6.7 7.3 21.0 5.5 C EW NGCL 501 8.8 11.3 44.4 7.9 CM S NGCL 505 10.1 10.7 25.4 7.1 C MB WV 541 29.4 30.1 7.1 3.2 C GS WV 544 14.2 14.8 42.9 11.1 C S NGCL 605 28.2 28.6 8.3 3.4 M GS WV 1316 9.0 10.0 237.9 27.7 H S NGCL 1317 14.1 14.5 2.9 2.7 WP S NGCL 1319 14.1 14.6 23.4 6.8 C CB WV 1320 12.9 13.5 9.9 3.8 M CB WV 1322 23.5 24.1 6.9 3.0 C UHFDA WV 1323 21.9 23.0 35.2 8.2 C UHFDA WV 1324 23.2 23.7 35.9 8.4 C UHFDA WV 1325 23.3 24.2 89.0 17.4 WP UHFDA WV 1326 27.4 28.0 25.4 8.5 C UHFDA WV 1327 29.2 29.6 23.5 7.5 C UHFDA WV 1329 30.5 31.0 12.1 5.1 WP UHFDA WV 1332 8.9 9.8 159.3 22.9 H S NGCL 1335 26.6 27.0 32.5 11.1 C GS WV 1337 30.3 30.9 272.6 23.4 H GS WV 3344 23.7 24.2 354.0 50.4 C UHFDA WV 3713 12.4 12.8 29.4 7.7 C WV 3715 11.6 12.1 9.8 3.8 M WV 4184 29.3 29.6 30.3 6.5 C WV 4187 29.3 29.8 38.5 7.3 C WV 4352 8.9 9.4 23.7 6.4 C S NGCL DB = Dry-bed elevation GS = Green Swamp Wildlife Management Area NP = Normal Pool elevation UHFDA = Upper Hillsborough Flood Detention Area Wetland Identification Well Field Identification C = Cypress wetland CB = Cypress Bridge CM = Cypress-Marsh wetland CBR = Cross Bar Ranch H = Hardwood wetland EW = Eldridge Wild M = Marsh wetland MB = Morris Bridge WP = Wet Prairie wetland S = Starkey Physiographic Region SP = South Pasco LU = Lake Upland NGCL =. Northern Gulf Coastal Lowland WV = Western Valley

PAGE 67

55 3.2.1. Precipitation Patterns Precip itation patterns and climate variability are important factors that influence the hydrologic conditions of wetlands. Overall, the annual precipitati on in west-central Florida is approximately 132 cm (52 in) per ye ar based on 110 years of record (Lee et al. 2009), with approximately 60% falling dur ing the months of June September (Southeast Regional Climate Center 2010). B ecause precipitation in the region varies strongly both temporally and spatially, it shoul d be measured at multiple locations around a study area to get a true indication of the precipitation and hydrologic response for the area. Annual precipitation r ecords from two weather statio ns, 1) Tampa International Airport (T.I.A.) located to the south side of the study area and 2) Hillsborough River State Park (HRSP) located central to the study area, are presented in Figure 3.2. The mean annual precipitation duri ng the seven year study period was 127.2 cm (50 in) at the Tampa International Airport, and 137.4 cm (54 in) at the HRSP (N ational Oceanic and Atmospheric Administration 2009). Typically the airport receives less rain than other areas of the region. The long-term (119 yea r) mean annual precipitation recorded at the airport is 113.4 cm (45 in), compared to 139.0 cm (55 in), the 64 year mean annual precipitation recorded at the HRSP site.

PAGE 68

56 0 20 40 60 80 100 120 140 160 180 2001200220032004200520062007Annual Rainfall (cm) T.I.A. HRSP wc-Florida 110yr average Figure 3.2 Annual regional ra infall measured at the Tampa International Airport (T.I.A.) and at the Hillsborough River State Park (HRSP). 3.2.2. Wetland Classifications The wetlands studied are com prised of 36 cypress (C), 5 cypress-marsh (CM), 3 hardwood (H), 9 marsh (M), and 3 wet prairi e (WP) as classified by SWFWMD (Table 3.1). Nine wetlands are located in the Gr een Swamp Wildlife Management Area (GS), which encompasses 194 km2 (75 square miles) of the region (Figure 3.1). The associated surface water levels in the Green Swamp are considered largely unaffected by human activities, and there has been little development of the gr oundwater resource in the region (Haag et al. 2005). Additionally, 33 wetlands ar e located within the extents or in the immediate vicinity of six public water-suppl y well fields in the northern Tampa Bay region: Cypress Bridge (CB) not shown, Cross Bar Ranch (CBR), Eldridge Wilde (EW), Morris Bridge (MB), South Pasco (SP) and St arkey (S) (Figure 3.1). The rest of the

PAGE 69

57 wetlands are located throughout Hillsborough, Pasc o, Pinellas, and Polk counties. It is believed that these wetlands represent the range of hydrologic conditions across westcentral Florida. Wetland planar area (A) and perimeter (P) va lues corresponding to the wetland extents and spatial coverage were calculated from the Geographic Information System (GIS) Land Use 1999 and National Wetland Inventory (NWI) shapefile records (Table 3.1) (Southwest Florida Water Management Distri ct 2007; U.S. Fish and Wildlife Service 2007a). The shapefiles contain approximati ons of the wetland extents based on aerial photos (Figure 3.3). These files were used to calculate the wetland areas and perimeters because physical measurements for the 56 wetland areas and perimeters were unavailable. The isolated wetlands range in size (spatial coverage) from 1,100 km2 to 392,900 km2 with an average wetland area and perimeter of 51,700 km2 and 9,300 km respectively (Table 3.1). In general, th e hardwood wetlands have the largest spatial coverage (223,300 km2), while the wet prairie wetlands have the smallest spatial coverage (34,700 km2).

PAGE 70

58 ) ( ( Instrumentation! (Upland Well! (Wetland Well" )Staff Gauge GIS Wetland Extents 4 010203040 5 Meters Figure 3.3 Example of SWFW MD well and staff gauge locations for a typical wetland. Additionally, wetland dry bed (DB) and normal pool (NP) elevations have been previously identified by SWFW MD staff (Table 3.1). A ccording to Mike Hancock, Senior Professional Engineer, Resource Proj ects Department, Southwest Florida Water Management District the wetland dry bed elevation was identified as the dry reading

PAGE 71

59 mark on the staff gauge and assumed to be the representative low point in the wetland (personal communication, February 4, 2010). This datum was subject to the ability of the staff gauge installation crew to get to the d eepest point in the wetland as well as a visual inspection of the wetland, so some variability exists. Also wetland bottom elevations are not consistent throughout the wetland extents (Haag et al. 2005; Nilsson et al. 2008). Therefore, the dry-bed elevation may not be representative for larger wetlands. The wetland normal pool elevation corresponds to physical and vegetative markers, i.e. inflection points on the buttre sses of cypress trees, stain marks, moss collars, and the elevation of root crowns, that identify a high water mark for the respective wetland (Carr and Rochow 2004; Haag et al. 2005). The normal pool elevation is typically a more consistent measure across the entire wetland than the dry-bed elevation. The mean dry bed and normal pool elevations based on the National Geodetic Vertical Datum of 1929 (NGVD) for all wetlands and each wetland category are: 17.1 m and 17.9 m (all), 18.1 m and 18.7 m (cypress), 14.3m and 15.3 (marsh), 13.0m and 14.4m (cypress-marsh), 16.1 m and 16.9 m (hardwood) and 22.7 m and 23.2 m (wet prairie) (Table 3.1). The wetland dry-bed elevatio ns range from 3.5 m NGVD (NGCL region) to 30.9 m NGVD (LU region). The wetland maximu m depth estimate, calculated as the difference between the normal pool elevation and dry-bed elevation, range from 0.4 m to 2.5 m with an average depth of 0.8 m. On av erage, the cypress and wet prairie wetlands are the shallowest at 0.6 m and the cypressmarsh wetlands are the deepest at 1.4 m.

PAGE 72

60 3.2.3. Wetland Hydrogeologic Setting The study wetlands are underlain by three hydr ogeological units: th e surficial aquifer system intermediate confining unit, and U pper Floridan aquifer. Figure 3.4 shows a typical wetland in relation to the hydrogeolog ic units. The surficial unit borders the land surface and primarily consists of unconsolid ated sand and clayey sand deposits a few meters to 10s of meters thick (Lee et al. 2009). The intermediate conf ining unit is rich in clay and generally impedes the flow of wate r from the surficial aquifer to the Upper Floridan aquifer (Lee and Swancar 1997). The de epest of the units is the Upper Floridan aquifer. The unit is the primary source of lo cal water supplies in th e study area, and is confined or semi-confined depending on the inte grity of the intermediate confining unit. The Upper Floridan unit is 100s to 1000s of meters thick depending on location, comprised of limestone and dolomites, and is hi ghly transmissive due to large fractures.

PAGE 73

61 Figure 3.4 Generalized hydroge ologic section and vert ical head distribution (modified from Lee et al., 2009). 3.3. Wetland and Upland Water Elevation Data Surface water and groun dwater elevations as sociated with each of the 56 isolated wetlands were monitored and recorded by SWFWMD over a period of seven years, January 1, 2001 thru December 31, 2007. This period was chosen due to data

PAGE 74

62 availability, and to ensure representative hydrologic and meteorologic conditions were covered. Two monitoring wells associated w ith each wetland were used to record the groundwater levels in the surfic ial aquifer system (Figure 3.3 ). One monitoring well is located within the wetland close to the st aff gauge, hence forth referred to as the wetland well and the other is located in the near by upland area/vegeta tion surrounding the wetland, hence forth referred to as the upland well The wetland well data represent standing water levels measured inside th e wetland when the wetland was flooded, and groundwater levels below the wetland during dr y periods. The standing water elevations were verified and are generally very c onsistent with the corresponding staff gauge (Figure 3.3) water elevations for each we tland to ensure accurate surface water representation (Appendix A). The upland well data represent groundwater levels adjacent to the wetlands. Physical properties of the 56 paired wetland wells and upland wells are listed in Table 3.2. The wells range in diameter (Dia) from 5.1 cm (2.0 in) to 15.2 cm (6.0 in), and range in total depth from 1.1 m (3.5 ft) to 10.2 m (33.5 ft) below the respective ground elevation listed in Table 3.4. Further, th e linear distance (Ld) between each paired wetland well and upland well is listed. The linear distances were calculated using the respective well GIS latitude and longitude co ordinates. The mean separation distance between the paired wells is 56 m, ranging from 8 m to 149 m.

PAGE 75

63 Table 3.2. Monitoring well identification and physical properties. Wetland Wells Upland Wells Wetland Well Dia Depth (m) Well Dia Depth (m) Ld ID ID (cm) TotalCasingID (cm) Total Casing (m) 20 1959 15 4.3 0.6 1792 5 4.6 1.5 23 21 1943 5 3.8 0.8 17542 5 6.2 3.2 149 51 1924 5 6.1 1.5 1923 5 4.3 1.2 79 70 1932 15 9.1 1.5 1741 5 4.0 0.9 29 81 1971 5 2.6 1.1 1818 5 3.8 0.8 98 84 1989 15 1.4 0.3 17400 5 5.8 2.7 49 89 1985 5 4.1 1.1 1928 5 5.2 0.9 94 112 1937 5 8.7 2.6 1718 5 4.6 0.9 35 136 1933 5 8.8 1.2 1724 5 3.0 0.6 49 143 1936 5 10.2 1.1 1719 5 4.6 0.9 63 154 1948 5 5.5 0.9 1947 5 4.6 1.5 64 165 1953 5 8.5 0.9 17456 5 5.8 2.7 78 170 1987 15 3.0 1.5 17398 5 3.0 0.5 57 183 1954 15 4.6 1.5 17485 5 5.7 2.7 59 196 1992 15 1.2 0.3 17403 5 2.7 0.3 46 201 1952 5 3.7 0.6 17457 5 5.7 2.7 59 215 1929 15 4.7 1.7 1930 5 3.0 0.9 57 252 1939 5 4.0 1.2 1736 5 3.0 0.6 78 261 2076 5 3.0 0.6 2077 5 4.0 0.9 67 276 1984 5 6.7 0.6 2120 5 6.7 1.1 35 295 1990 15 1.2 0.3 17401 5 2.9 0.5 74 301 1918 15 8.5 2.4 1745 5 3.0 0.9 31 304 17427 5 3.8 2.9 17478 5 5.2 2.1 73 320 1958 5 4.3 1.2 1791 5 7.6 1.5 77 331 2157 5 4.0 0.6 2158 5 5.2 0.6 38 379 1951 5 5.8 1.2 17458 5 5.6 2.5 84 384 1793 5 3.0 1.2 17461 5 3.9 0.8 37 388 1988 15 2.4 0.6 17399 5 2.8 0.4 53

PAGE 76

64 Table 3.2. (Continued). Wetland Wells Upland Wells Wetland Well Dia Depth (m) Well Dia Depth (m) Ld ID ID (cm) TotalCasingID (cm) Total Casing (m) 407 1790 5 6.1 1.5 17460 5 3.3 0.8 59 489 1927 5 5.8 1.2 1926 5 7.6 0.9 58 493 17487 5 2.4 1.5 1925 5 11.0 1.2 54 501 1941 5 7.5 1.4 17488 5 5.8 2.8 24 505 1950 5 5.2 0.6 17459 5 5.7 2.6 62 541 1991 15 4.6 0.9 17402 5 4.2 1.2 51 544 1931 5 5.6 1.1 1744 5 4.0 0.9 31 605 1966 15 1.8 0.3 17397 5 2.8 0.4 58 1316 1942 5 7.0 0.9 1743 5 4.6 1.5 36 1317 1934 5 6.1 1.5 1717 5 4.6 0.6 31 1319 1961 15 2.4 0.5 1786 5 4.6 1.5 43 1320 1960 15 1.2 0.3 1785 5 6.1 1.5 39 1322 1974 5 2.6 0.5 1804 5 2.1 0.6 43 1323 1975 5 1.5 0.3 1802 5 2.1 0.6 103 1324 1976 5 6.1 1.5 1811 5 7.6 1.5 41 1325 1977 15 2.1 0.6 1803 5 3.0 0.6 67 1326 1978 15 1.5 0.6 1800 5 4.6 0.9 8 1327 1979 5 4.9 0.3 1799 5 2.1 0.6 42 1329 1981 15 3.0 0.6 1808 5 2.1 0.6 83 1332 1940 5 5.9 0.6 1742 5 4.6 1.5 27 1335 1965 5 1.5 0.6 1739 5 1.5 0.3 82 1337 1995 15 1.8 0.3 1851 5 3.0 0.9 75 3344 1972 5 2.1 0.6 1973 5 2.3 0.8 25 3713 2064 15 4.1 0.8 2063 5 3.5 0.5 83 3715 2060 15 1.1 0.3 2059 5 2.9 0.6 79 4184 2260 5 1.5 0.5 2259 5 1.5 0.5 29 4187 2251 5 1.1 0.3 2250 5 1.5 0.3 58 4352 2253 5 3.0 0.3 2252 5 3.7 0.6 25 The wetland water levels and upland groundwat er levels were measured continuously using electronic water-level reco rders and/or periodically with a graduated tape. Over the course of the study, between 52 and 2,153 water level measurements were collected at each wetland well, and between 52 and 137 groundwater level measurements were collected at each upland well (Table 3.3). A couple of factors influenced the number of water-level records available at each monitori ng well: 1) the start of the data collection

PAGE 77

65 and 2) the frequency at which the data wa s collected. The wetland water level data collection began in 2001 for all but four of the wells, which were installed in 2002 and 2003. The wetland water-level data were collect ed continuously (daily), semi-monthly or on a monthly basis. The upland groundwater data collection began in 2001 for all but five of the wells, which were installed in 2002 and 2003. The upland water-level data were collected on a semi-monthly or monthly basis. The water-elevation data recorded at each wetland well and upland well corresponding to the 56 wetlands are summarized in Table 3.4. In addition, the ground elevation (GE) associated with each well is listed. Th e ground elevation was determined from a land survey adjacent to the well. The wetl and well ground elevations range from 3.5 m to 30.9 m NGVD, and the upland well ground el evations range from 4.8 m to 31.6 m NGVD.

PAGE 78

66 Table 3.3. Wetland well and upland well data collection schedule. Wetland Wells Upland Wells Wetland Well Data Collection Well Data Collection ID ID Start Frequency RecordsID Start Frequency Records 20 1959 2001 continuous 1837 1792 2001 monthly 137 21 1943 2001 monthly 122 17542 2001 monthly 131 51 1924 2001 monthly 66 1923 2001 monthly 64 70 1932 2001 continuous 1464 1741 2001 monthly 83 81 1971 2001 monthly 71 1818 2001 monthly 84 84 1989 2001 continuous 1421 17400 2001 monthly 74 89 1985 2001 monthly 125 1928 2001 monthly 125 112 1937 2001 monthly 125 1718 2001 monthly 132 136 1933 2001 monthly 125 1724 2001 monthly 130 143 1936 2001 monthly 122 1719 2001 monthly 130 154 1948 2001 monthly 67 1947 2001 monthly 71 165 1953 2001 monthly 128 17456 2001 monthly 134 170 1987 2001 continuous 1354 17398 2001 monthly 75 183 1954 2001 continuous 1757 17485 2001 monthly 136 196 1992 2001 continuous 1385 17403 2001 monthly 75 201 1952 2001 monthly 128 17457 2001 monthly 134 215 1929 2001 continuous 1811 1930 2001 monthly 128 252 1939 2001 monthly 123 1736 2001 monthly 132 261 2076 2001 monthly 62 2077 2001 monthly 65 276 1984 2001 monthly 67 2120 2002 monthly 61 295 1990 2001 continuous 1326 17401 2001 monthly 73 301 1918 2001 continuous 1849 1745 2001 monthly 73 304 17427 2001 monthly 131 17478 2001 monthly 133 320 1958 2001 monthly 126 1791 2001 monthly 132 331 2157 2002 monthly 115 2158 2002 monthly 116 379 1951 2001 monthly 126 17458 2001 monthly 134 384 1793 2001 monthly 135 17461 2001 monthly 135 388 1988 2001 continuous 1424 17399 2001 monthly 73

PAGE 79

67 Table 3.3. (Continued). Wetland Wells Upland Wells Wetland Well Data Collection Well Data Collection ID ID Start Frequency RecordsID Start Frequency Records 407 1790 2001 monthly 134 17460 2001 monthly 135 489 1927 2001 monthly 124 1926 2001 monthly 125 493 17487 2001 monthly 72 1925 2001 monthly 65 501 1941 2001 monthly 124 17488 2001 monthly 132 505 1950 2001 monthly 128 17459 2001 monthly 135 541 1991 2001 continuous 1205 17402 2001 monthly 75 544 1931 2001 monthly 126 1744 2001 monthly 130 605 1966 2001 continuous 1197 17397 2001 monthly 74 1316 1942 2001 monthly 125 1743 2001 monthly 130 1317 1934 2001 monthly 125 1717 2001 monthly 133 1319 1961 2001 continuous 2153 1786 2001 monthly 82 1320 1960 2001 continuous 2140 1785 2001 monthly 81 1322 1974 2001 monthly 77 1804 2001 monthly 82 1323 1975 2001 monthly 79 1802 2001 monthly 79 1324 1976 2001 monthly 79 1811 2001 monthly 82 1325 1977 2001 continuous 1965 1803 2001 monthly 81 1326 1978 2001 continuous 1688 1800 2001 monthly 76 1327 1979 2001 monthly 71 1799 2001 monthly 76 1329 1981 2001 continuous 1826 1808 2001 monthly 76 1332 1940 2001 monthly 124 1742 2001 monthly 130 1335 1965 2001 monthly 77 1739 2001 monthly 77 1337 1995 2001 continuous 1207 1851 2001 monthly 74 3344 1972 2001 monthly 78 1973 2001 monthly 79 3713 2064 2001 continuous 1964 2063 2001 monthly 70 3715 2060 2001 continuous 1967 2059 2001 monthly 70 4184 2260 2003 bi-monthly 56 2259 2003 bi-monthly 55 4187 2251 2003 bi-monthly 55 2250 2003 bi-monthly 56 4352 2253 2003 bi-monthly 52 2252 2003 bi-monthly 52

PAGE 80

68 Table 3.4. Wetland and upland water elev ation summary statistics (NGVD 29). Wetland Water Elevations Upland Water Elevations Wetland Well GE Water Level Statistics (m) Well GE Water Level Statistics (m) ID ID (m) MeanStD MedianMin MaxID (m) MeanStD MedianMin Max 20 1959 12.5 12.2 0.7 12.3 9.8 13.11792 13.1 11.7 1.0 11.9 9.0 13.2 21 1943 8.0 8.3 0.7 8.4 6.9 9.8 1754210.1 8.1 0.8 8.0 6.6 9.7 51 1924 7.9 7.7 0.7 7.8 6.4 8.9 1923 8.7 7.7 0.8 7.8 5.2 8.9 70 1932 13.5 13.5 0.5 13.7 12.214.21741 14.4 13.5 0.5 13.6 12.214.3 81 1971 22.1 22.8 0.8 23.0 20.823.91818 24.4 23.4 0.6 23.5 21.824.4 84 1989 30.9 31.0 0.3 31.1 29.631.51740031.6 30.9 0.4 31.0 29.931.5 89 1985 13.5 13.5 0.5 13.7 11.914.01928 14.2 13.4 0.5 13.6 12.214.1 112 1937 12.3 12.3 0.4 12.5 11.212.91718 12.9 12.4 0.4 12.5 11.012.9 136 1933 14.0 13.8 0.4 14.0 12.714.51724 14.5 13.8 0.5 13.9 12.514.5 143 1936 12.4 12.3 0.5 12.4 10.913.21719 13.0 12.2 0.6 12.4 10.713.1 154 1948 21.9 22.2 0.7 22.5 20.323.01947 23.3 22.3 0.7 22.6 20.423.2 165 1953 11.7 11.6 0.5 11.8 10.012.21745612.4 11.6 0.5 11.6 10.112.4 170 1987 30.3 30.3 0.3 30.4 28.830.71739830.8 30.1 0.3 30.2 29.130.8 183 1954 10.1 9.9 0.9 10.1 7.5 11.11748511.4 9.3 1.0 9.4 7.4 11.3 196 1992 29.5 29.3 0.5 29.5 28.330.01740330.1 29.2 0.6 29.4 27.630.0 201 1952 11.1 11.0 0.7 11.4 8.9 11.71745712.0 10.6 1.0 10.9 8.7 12.0 215 1929 13.6 13.8 0.4 14.0 12.214.21930 14.5 13.9 0.4 14.0 12.614.5 252 1939 10.9 10.7 0.7 10.8 9.1 11.61736 11.8 10.7 0.7 10.8 9.0 11.7 261 2076 3.5 4.3 0.2 4.3 3.5 4.6 2077 4.9 4.3 0.2 4.4 3.6 4.7 276 1984 17.1 17.0 0.9 17.3 14.818.12120 17.9 17.2 0.8 17.3 15.518.1 295 1990 30.9 30.9 0.4 31.0 29.731.51740131.6 30.8 0.4 30.9 29.631.5 301 1918 4.0 4.1 0.3 4.2 3.1 4.6 1745 4.8 4.1 0.3 4.2 3.3 4.6 304 17427 13.7 13.4 0.6 13.4 11.914.61747815.1 13.5 0.6 13.5 12.114.8 320 1958 12.5 12.4 0.6 12.6 10.813.01791 13.0 12.0 0.6 12.0 10.513.0 331 2157 12.8 13.3 0.3 13.4 11.913.62158 13.7 13.2 0.4 13.3 12.013.7 379 1951 9.8 9.2 1.1 9.7 5.9 10.61745810.3 9.2 0.9 9.3 6.0 10.6 384 1793 12.3 12.0 0.7 12.1 10.312.91746113.1 11.7 0.7 11.8 9.9 13.1 388 1988 30.0 30.1 0.5 30.2 28.330.81739930.8 30.1 0.4 30.1 28.830.8

PAGE 81

69 Table 3.4. (Continued). Wetland Water Elevations Upland Water Elevations Wetland Well GE Water Level statistics (m) Well GE Water Level statistics (m) ID ID (m) MeanStD MedianMin MaxID (m) MeanStD MedianMin Max 407 1790 10.5 10.3 0.7 10.6 8.4 11.41746011.5 10.1 0.8 10.1 8.4 11.4 489 1927 11.8 12.4 0.5 12.6 10.813.01926 13.3 12.5 0.5 12.8 11.013.2 493 17487 6.7 6.8 0.6 7.0 5.3 7.6 1925 7.3 6.9 0.5 7.0 5.8 7.5 501 1941 8.9 9.7 0.9 9.9 7.8 11.31748811.7 9.9 1.0 10.0 7.9 11.6 505 1950 10.1 9.8 0.9 10.2 7.7 11.61745911.1 9.3 1.2 9.6 6.4 11.0 541 1991 29.5 29.6 0.5 29.7 28.230.31740230.2 29.5 0.5 29.6 28.230.1 544 1931 14.2 14.3 0.5 14.5 12.814.81744 14.8 14.2 0.4 14.3 13.014.9 605 1966 28.1 28.3 0.2 28.3 26.928.71739728.8 28.0 0.5 28.0 26.628.8 1316 1942 9.2 9.0 0.6 9.1 7.7 10.01743 10.4 9.1 0.6 9.1 7.7 10.2 1317 1934 14.1 13.5 0.6 13.5 12.114.61717 14.6 13.5 0.6 13.5 12.114.6 1319 1961 14.1 14.2 0.4 14.3 11.614.91786 14.8 12.9 0.5 13.0 11.514.2 1320 1960 12.9 13.2 0.3 13.3 11.813.81785 13.6 12.4 0.7 12.4 10.713.6 1322 1974 23.6 22.8 0.7 22.8 21.024.01804 24.2 22.6 0.6 22.6 21.624.0 1323 1975 21.1 21.5 0.6 21.4 20.522.91802 23.1 22.1 0.5 22.1 21.023.1 1324 1976 23.2 22.2 1.3 22.2 18.324.21811 23.7 21.7 1.6 21.9 18.223.7 1325 1977 23.4 22.9 0.8 23.0 21.324.51803 24.6 22.8 0.8 22.7 21.624.4 1326 1978 27.4 27.4 0.6 27.7 26.028.11800 28.0 26.8 1.0 27.2 24.328.0 1327 1979 29.2 28.9 0.6 29.2 27.529.51799 29.7 28.6 0.7 28.7 27.529.6 1329 1981 30.5 30.5 0.4 30.6 29.031.11808 31.0 30.1 0.5 30.1 28.931.0 1332 1940 8.9 9.1 0.7 9.2 7.5 9.9 1742 10.2 9.1 0.7 9.3 7.5 10.1 1335 1965 26.6 26.4 0.5 26.6 25.127.01739 26.9 26.4 0.5 26.6 24.927.0 1337 1995 30.3 30.4 0.3 30.5 29.030.91851 31.2 30.3 0.6 30.5 28.531.2 3344 1972 23.7 23.6 0.7 23.9 21.724.51973 24.3 23.6 0.7 23.8 21.824.5 3713 2064 12.4 12.0 0.7 12.3 10.012.82063 13.1 12.2 0.5 12.1 11.113.1 3715 2060 11.6 11.8 0.4 11.9 10.612.32059 12.1 11.2 0.7 11.2 9.5 12.1 4184 2260 29.2 29.3 0.4 29.4 27.829.62259 29.7 29.1 0.4 29.1 28.229.6 4187 2251 29.3 29.5 0.4 29.6 28.529.92250 29.9 29.1 0.5 29.1 28.029.9 4352 2253 9.2 9.1 0.1 9.1 8.5 9.3 2252 9.8 9.0 0.2 9.1 8.4 9.4

PAGE 82

70 3.4. Methods 3.4.1. Hydrologic Evaluation 3.4.1.1. Empirical Distribution (Frequency) Development Em pirical distribution functions (EDFs) were developed from the wetland water elevation and upland groundwater elevati on data sets summarized in Table 3.4. The functions represent the discrete frequency distribution of surface and groundwater elevations within the wetland extents, and the frequency dist ribution of groundwater elevations in the adjacent/surrounding upland. The resulting ir regular distributions were compared between the different study wetlands and wetland categories. The empirical distribution function is a disc rete step function that is an unbiased estimator of the cumulative or probability distribution function (Chow et al. 1988). The empirical distribution fu nction is defined as: x m ms xspfpF1)() ( (3.1) where xspF is the sum of the values of the relative frequencies mspf up to a given observation x m is the rank of the observation, and the subscript s denotes the function is calculated from sample data (C how et al. 1988; Maidment 1993).

PAGE 83

71 Relative frequencies (probabil ities) were computed based on ordered or ranked data (Maidment 1993; Weisstein 2009c). The ge neral expression for computing relative frequencies is: 12 aN am pm (3.2a) 1 N m pm (3.2b) where pm is the relative frequency, i.e. the pr obability of a value being less than the mth smallest observation in the data set, m is the rank of the observation, N is the sample size, and a is a constant associated with a probability model, e.g. for a uniform distribution, a = 0. For this application a uniform distri bution is assumed, therefore the relative frequencies were calculated using Eq. (3.2b). The procedure used to develop empirical distributions to represent the wetland and upland water elevations associated with an i ndividual wetland or cate gory is: 1) sort the water elevations from the smallest value to the largest value, 2) assign a rank ( m ) to each water elevation, and 3) calculate the rela tive frequency (probability of exceedance) associated with each water elevation using Eq. (3.2b). The cumulative probability for any water elevation is the sum of all relativ e frequencies up to the respective value.

PAGE 84

72 The empirical distributions developed in this chapter are presented as percentiles (Altman and Bland 1994). In general, the percentile of a distribution repres ents the fraction of data that are less than or e qual to a specific observation or value (StatSoft Inc. 2010). For instance, the 50th percentile indicates 50% of the data is equal to or less than the corresponding value. In this work, the percen tiles represent the cumulative probabilities, i.e. frequencies of occurrence, and the am ount of time the wetland or upland recorded water elevations were at or be low a specific elevation. In order to glean insight into represen tative wetland water elevation and upland groundwater elevation characte ristics associated with th e 56 wetlands, representative percentiles needed to be identified. Three percentiles were chosen due to the skewed (non-normal) nature of the wetland and upla nd water elevation hist ograms (Appendix B). Altman and Bland (1994) recommend using the median (50th percentile) and two outer percentiles to summarize skewed distributions. Therefore, the 10th, 50th, and 90th percentiles, hereafter referred to as target percentiles, were chosen to represent the respective frequency distributions. The 10th percentile represents low water elevations, the 50th percentile represents median water elevations, and the 90th percentile represents high water elevations. 3.4.1.2. Relative Water Level Development The individual wetland and upland well water-e levation data were norm alized so as to provide a means of comparing one data set to another. Relative water levels (RWLs) were developed with respect to two datu ms: 1) the ground elevation at each well ( GE ),

PAGE 85

73 and 2) the associated we tland dry-bed elevation ( DB ). The normalized data are presented as centimeters above or below the respective da tum. The datums were chosen to provide different perspectives on the water-elevation distributions. Further, the datums provide a means to evaluate the overall trend of the wetland and upland water-elevation distributions, and evaluate the variability in the water levels associated with all the wetlands at the 10th, 50th, and 90th percentiles. 3.4.1.3. Frequency of Water Levels at Dry Bed (DB ) and Norm al Pool ( NP ) Additionally, the empirical dist ribution functions were used to identify the cumulative probability (percentile) corresponding to the wetland dry-bed elevation, F ( DB ), and the wetland normal-pool elevation, F ( NP ). The percentiles were used to identify the duration of time standing water was present within the wetlands, and the amount of time water was above the normal pool vegetative markers over the seven year study. Monthly data, based on a single measurement or the average of all water elevations recorded during the respective month, were used in this analysis. Monthly values were used to eliminate any bias brought about by combining data sets comprised of daily water elevations and monthly elevations. Further, the use of monthly data did no t influence the results because data sets comprised of monthly data are no t statistically different from data sets comprised of daily data (Appendix C). 3.4.2. Wetland Category and Group Comp arisons Kolm ogorov-Smirnov Tests The two-sample Kolmogorov-Smirnov test (KStest) was used to compare the wetland categories and various groups of wetlands. The KS-test is a form of minimum distance

PAGE 86

74 estimation used to compare empirical distribution functions to determine if two datasets are statistically similar (Massey 1951; StatSoft Inc. 2010). The KS-test makes no assumption about the data distribution, i.e. it is a non-parametric and distribution free test. The test was used due to the non-parametric nature of the water-elevation data (Appendix B). The KS-test statistic ( D statistic) quantifies the maximum distance expressed as a probability betw een the empirical distribution functions of two samples. The null hypothesis for the test is that the two data sets are from the same continuous distribution. The result of h = 1 is returned if the test re jects the null hypothesis at the 5% significance level, otherwise the result of h = 0 is returned indicating a failure to reject the null hypothesis. 3.4.2.1. Wetland Category Comparisons Em pirical distribution functions representing the various we tland categories (Table 3.1) were developed and evaluated to determine if the observed water levels associated with each wetland category are significantly different. As indicated previously, monthly data records were used in this analysis to e liminate potential bias brought about by various data collection frequencies. The distributi ons were developed by combining individual wetland water elevation data, normalized by subtracting the individual wetland dry bed elevation, within the specific wetland category. For example, the wetland relative water levels associated with each of the 36 cypre ss wetlands were combined into one aggregate or representative data set. Using the procedure outlined in Section 3.4.1.1, relative frequencies were then calculated for each of the water levels in the data set, and a characteristic empirical distri bution was developed for the cy press category. Archetypal

PAGE 87

75 empirical distributions were developed for each of the five wetland categories. Each wetland category distribution was compared to every other wetland category distribution using the KS-test to determine if the respective hydrologic data were statistically similar. 3.4.2.2. Wetland Group Comparisons Four wetland groups were identified based on regional location and the associated hydrogeology. The first group is comprised of the nine wetlands located in the Green Swa mp, the second and third groups are comp rised of wetlands located in the Morris Bridge (9 ea.) and the Starkey (17 ea.) well fields, and the fourth group is comprised of nine wetlands in the UHFDA area located south-southeast of the Green Swamp (Figure 3.1). Individual wetlands w ithin a particular group are id entified in Table 3.1. The wetland groups were compared to determine if the observed water levels are statistically different. As before, empirical distribution functions were developed by aggregating the wetland relative water levels within the specific wetland group. Each aggregated wetland group distribution was compared to every ot her wetland group distri bution using the KStest. Monthly data were also used for this comparison. 3.5. Results 3.5.1. Wetland Hydrologic Evaluation (Frequency Analysis) 3.5.1.1. Well Ground Elevation Datum ( GE ) The em pirical distribution functions shown in Figure 3.5 depict the re lative water levels, based on the respective ground elevations at the well ( GE ), associated with all 56 wetland wells (Chart A) and upland wells (Chart B). The charts provide a direct comparison of

PAGE 88

76 the wetland and upland relative water levels, and illustrate the variability in the water levels at each cumulative probability (percent ile). Further, the charts show no distinct patterns that could be used to identify the various wetland types. The overlaid water-level distri butions presented in Charts A and B are a bit hard to decipher; therefore, mean-relative-water-level plots representing the wetland water levels (Chart C) and upland groundwater levels (Chart D) were developed from the 56 individual wetland and upland water-level di stributions. The charts show the mean relative water level one standard deviation at each percentile, and show the variability of the water levels over the em pirical distribution range. Wetland water levels and upland groundw ater levels corresponding to the 10th, 50th, and 90th target percentiles on Figure 3.5 are summ arized in Table 3.5 to provide additional insights into the wetland hydrologi c characteristics. On average, the wetland water levels were 90.8 cm below the ground elevation at the wetland well for the 10th percentile, and 13.0 cm and 57.5 cm above the ground el evation at the wetland well for the 50th and 90th percentiles respectively (Table 3.5 Regi onal Wetlands). In general, the upland groundwater levels were below the ground elevat ion at the well for each target percentile: 196.5 cm (10th percentile), 94.3 cm (50th percentile) and 28.6 cm (90th percentile) (Table 3.5 Regional Wetlands).

PAGE 89

77 The variability, indicated by the standard deviation (StD), of the wetland water levels and upland groundwater levels was generally highest at the 10th percentile (57.6 cm and 80.7 cm respectively), and lowest at the 90th percentile (35.6 cm and 24.6 cm respectively). This can be seen on Figure 3.5, Charts C a nd D, as well. Also, Charts A and B on Figure 3.5 show the wetland and upland water levels associated with the various wetlands crisscross at the target percentiles, which is substantiated in Table 3.5. For instance, water levels at the 50th percentile for the cypress wetlands ranged from 103.6 cm below the well ground elevation (min) to 81.4 cm above the well ground elevation (max), and for the marsh wetlands ranged from 28.7 cm below the well ground elevation (min) to 89.3 cm above the well ground elevation (max). The interdecile range is th e difference between the 10th and 90th percentiles, and is a measure of dispersion of the values in the data set. The range is the width about the median that includes 80% of the cases or wa ter-level data. The mean range of wetland water levels between the 10th and 90th percentiles is 148.3cm, and the mean range of upland groundwater levels is 167.9 cm (Table 3.5). Thus, the upland wells show greater fluctuations than the wetland wells. The cypress-marsh wetlands have the largest range of wetland water levels and upland groundwater levels between the 10th and 90th percentiles, 198.2 cm and 215.0 cm respectively. While, the cypress wetlands have the smallest range of wetland and upland water levels, 138.6 cm and 151.7 cm respectively.

PAGE 90

78 0 10 20 30 40 50 60 70 80 90 100 -600 -400 -200 0 200 PercentileRelative Water Level (cm) [GE datum]Chart A Wetland EDFs 0 10 20 30 40 50 60 70 80 90 100 -600 -400 -200 0 200 Percentile Relative Water Level (cm) [GE datum]Chart B Upland EDFs 0 10 20 30 40 50 60 70 80 90 100 -600 -400 -200 0 200 PercentileRelative Water Level (cm) [GE datum]Chart C Wetland EDF Trends Mean EDF Std EDF 0 10 20 30 40 50 60 70 80 90 100 -600 -400 -200 0 200 PercentileRelative Water Level (cm) [GE datum]Chart D Upland EDF Trends Mean EDF Std EDF Figure 3.5 Empirical distribution function charts re presenting wetland and upland water levels adjusted to the ground elevation at the well ( GE ).

PAGE 91

79 Table 3.5. Wetland and upland empirical distribution summary statistics adjusted to the ground elevation at the wetland well ( GE ). Relative Water Levels (cm) Wetland Upland Statistic 10th 50th 90th 10th 50th 90th Regional Wetlands Mean -90.8 13.0 57.5 -196.5 -94.3 -28.6 StD 57.6 35.9 35.6 80.7 45.5 24.6 Median -87.0 14.3 49.1 -174.2 -81.4 -23.6 Min -239.0 -103.6 5.5 -514.2 -211.8 -118.9 Max 54.9 92.4 192.9 -73.2 -20.7 13.4 Range 293.8 196.0 187.5 441.0 191.1 132.3 Cypress Wetlands Mean -88.0 11.7 50.6 -176.9 -79.7 -25.2 StD 52.4 32.2 28.1 81.3 37.9 22.2 Median -86.7 14.3 46.5 -157.4 -71.2 -22.3 Min -239.0 -103.6 5.5 -514.2 -181.7 -118.9 Max 54.9 81.4 149.7 -73.2 -20.7 10.1 Range 293.8 185.0 144.2 441.0 160.9 128.9 Marsh Wetlands Mean -78.8 21.1 73.5 -241.8 -135.3 -43.5 StD 63.4 36.8 39.9 62.9 48.6 32.1 Median -66.4 18.3 59.4 -247.8 -135.3 -41.5 Min -176.8 -28.7 37.5 -340.2 -211.8 -99.7 Max 6.1 89.3 154.2 -153.0 -76.2 -5.8 Range 182.9 118.0 116.7 187.1 135.6 93.9 Cypress-Marsh Wetlands Mean -105.3 36.3 92.9 -230.2 -98.7 -15.2 StD 94.6 51.0 60.7 103.2 51.2 25.7 Median -86.9 60.4 81.4 -197.8 -93.6 -9.8 Min -235.0 -19.2 36.9 -356.3 -170.7 -57.0 Max 14.0 92.4 192.9 -94.2 -37.5 13.4 Range 249.0 111.6 156.1 262.1 133.2 70.4 Hardwood Wetlands Mean -93.5 10.4 59.2 -219.0 -100.0 -36.7 StD 41.2 23.4 27.4 34.3 24.0 20.6 Median -108.2 18.9 50.6 -236.2 -94.8 -33.8 Min -125.3 -16.2 37.2 -241.4 -126.2 -58.5 Max -46.9 28.3 89.9 -179.5 -78.9 -17.7 Range 78.3 44.5 52.7 61.9 47.2 40.8 Wet Prairie Wetlands Mean -134.1 -30.9 31.2 -217.4 -133.1 -39.7 StD 58.8 39.1 18.3 70.3 52.3 15.0 Median -140.8 -37.2 38.4 -192.3 -113.7 -42.7 Min -189.3 -66.4 10.4 -296.9 -192.3 -53.0 Max -72.2 11.0 44.8 -163.1 -93.3 -23.5 Range 117.0 77.4 34.4 133.8 99.1 29.6

PAGE 92

80 3.5.1.2. Wetland Dry-Bed Datum ( DB ) The em pirical distribution function plots show n in Figure 3.6 depict the relative water levels, based on the associated wetland dry-bed elevation ( DB ), for all 56 wetland wells (Chart A) and upland wells (Chart B). The charts provide a direct comparison of the wetland and upland water levels, and illustrate the variability in the water levels at each percentile. As before, the ch arts show no obvious distinct pa tterns that could be used to differentiate the various wetland categories. The overlaid water-level distri butions presented in Charts A and B are a bit hard to decipher; therefore, mean-relative-water-level plots representing the wetland water levels (Chart C) and upland groundwater levels (Chart D) were developed from the 56 individual wetland and upland wa ter-level distributions. The charts show the mean water level one standard deviation at each percentile, and show the variab ility of the water levels. The relative water levels at the 10th, 50th, and 90th percentiles corre sponding to the distributions presented in Figure 3.6 are summ arized to provide additional insights into the wetland and upland hydrologic differences and interactions (Table 3.6). On average, the water levels in the wetlands were 85.2 cm below the dry bed elevation at the 10th percentile, and 18.7 cm and 63.1 cm a bove the dry bed elevation at the 50th and 90th percentiles respectively (Table 3.6 Regiona l Wetlands). The mean upland water levels were 102.7 cm and 0.5 cm below the wetland dry bed elevation at the 10th and 50th percentiles respectively, and 65.1 cm above the dry bed elevation at the 90th percentile

PAGE 93

81 (Table 3.6 Regional Wetlands). Further, in general, the water levels in the uplands were deeper than the associated wetland water levels at the 10th and 50th percentiles; while the water levels in the wetlands were deeper than the associated upland water levels at the 90th percentile. The variability, expressed as the standard devi ation (StD), of the we tland water levels and upland groundwater levels was on average largest at the 10th percentile, 58.2 cm and 80.2 cm respectively, and smallest at the 90th percentile, 37.1 cm and 45.8 cm respectively (Figure 3.6 Charts C and D). Also, Charts A and B on Figure 3.6 show the wetland and upland relative water levels associated with the various wetlands crisscross at the three target percentiles, which is substantiated in Table 3.6. For instance, relative water levels at the 50th percentile for the cypress wetlands ranged from 103.3 cm below the wetland dry bed elevation (min) to 82.0 cm above th e wetland dry bed elevation (max), and for the marsh wetlands the 50th percentile water levels range d from 19.2 cm below the well wetland dry bed elevation (min) to 100.0 cm abov e the wetland dry bed elevation (max). In general, the interdecile ra nge of the regional wetland wate r levels is 148.3 cm, and the upland groundwater level range is 167.9 cm (Table 3.6). The cypress-marsh wetlands exhibit the largest range between the 10th and 90th percentiles for both the wetland and upland wells (198.2 cm and 215.0 cm respectively), while the cypre ss wetlands have the smallest range between the 10th and 90th percentiles (138.6 cm and 151.7 cm respectively).

PAGE 94

82 0 10 20 30 40 50 60 70 80 90 100 -500 -400 -300 -200 -100 0 100 200 300 PercentileRelative Water Level (cm) [DB datum]Chart A Wetland EDFs 0 10 20 30 40 50 60 70 80 90 100 -500 -400 -300 -200 -100 0 100 200 300 PercentileRelative Water Level (cm) [DB datum]Chart B Upland EDFs 0 10 20 30 40 50 60 70 80 90 100 -500 -400 -300 -200 -100 0 100 200 300 PercentileRelative Water Level (cm) [DB datum]Chart C Wetland EDF Trends Mean EDF Std EDF 0 10 20 30 40 50 60 70 80 90 100 -500 -400 -300 -200 -100 0 100 200 300 PercentileRelative Water Level (cm) [DB datum]Chart D Upland EDF Trends Mean EDF Std EDF Figure 3.6 Empirical distribution function charts repr esenting the wetland and upland water le vels adjusted to the wetland dry-bed elevation (DB ).

PAGE 95

83 Table 3.6. Wetland and upland empirical dist ribution summary statistics adjusted to the wetland dry-bed elevation ( DB ). Relative Water Levels (cm) Wetland Upland Statistic 10th 50th 90th 10th 50th 90th Regional Wetlands Mean -85.2 18.7 63.1 -102.7 -0.5 65.1 StD 58.2 39.6 37.1 80.2 52.1 45.8 Median -87.0 20.0 55.9 -96.0 -1.8 55.2 Min -238.7 -103.3 13.4 -466.6 -128.9 -48.8 Max 60.4 110.9 207.9 64.6 146.9 236.8 Range 299.0 214.3 194.5 531.3 275.8 285.6 Cypress Wetlands Mean -86.2 13.5 52.5 -100.4 -3.2 51.3 StD 53.8 34.8 22.8 87.5 46.8 30.5 Median -87.0 18.3 49.8 -84.7 0.8 50.3 Min -238.7 -103.3 14.3 -466.6 -128.9 -48.8 Max 55.5 82.0 118.6 64.6 102.1 134.7 Range 294.1 185.3 104.2 531.3 231.0 183.5 Marsh Wetlands Mean -67.8 32.1 84.5 -111.0 -4.5 87.3 StD 52.6 39.0 47.1 74.2 60.4 59.1 Median -68.0 36.0 61.6 -139.9 -14.6 55.5 Min -135.9 -19.2 34.7 -174.7 -53.0 40.8 Max 7.6 100.0 164.9 64.3 146.9 217.3 Range 143.6 119.2 130.1 239.0 199.9 176.5 Cypress-Marsh Wetlands Mean -83.3 58.3 114.9 -85.7 45.8 129.3 StD 105.0 53.4 57.8 91.3 71.4 64.9 Median -71.9 60.4 98.1 -63.7 65.8 124.4 Min -232.3 -14.0 57.9 -200.3 -40.8 66.1 Max 60.4 110.9 207.9 40.5 123.1 236.8 Range 292.6 125.0 150.0 240.8 164.0 170.7 Hardwood Wetlands Mean -82.8 21.0 69.9 -95.8 23.3 86.6 StD 36.4 12.5 24.9 13.7 12.3 10.1 Median -102.1 25.0 73.8 -96.0 18.6 81.7 Min -105.5 7.0 43.3 -109.4 14.0 79.9 Max -40.8 31.1 92.7 -82.0 37.2 98.1 Range 64.6 24.1 49.4 27.4 23.2 18.3 Wet Prairie Wetlands Mean -131.5 -28.2 33.8 -140.8 -56.5 36.9 StD 57.6 38.1 18.2 30.4 13.4 31.7 Median -137.8 -33.5 39.6 -139.3 -60.7 28.3 Min -185.6 -63.4 13.4 -171.9 -67.4 10.4 Max -71.0 12.2 48.5 -111.3 -41.5 71.9 Range 114.6 75.6 35.1 60.7 25.9 61.6

PAGE 96

84 3.5.1.3. Dry Bed ( DB ) and Nor mal Pool (NP ) Relative Frequency Identification The empirical distribution functions shown in Figure 3.6 were used to identify the relative frequency (percentile) co rresponding to the dry-bed elevation F ( DB ) and normal pool elevation F ( NP ) associated with the study wetlands Overall, the wetland dry-bed elevations correspond to a relative frequency of 0.39, ranging from 0.02 to 0.86 (Table 3.7 Regional Wetlands). This indicates 39 % of the recorded water elevations were equal to or less than the wetland dry-bed elev ation. The wet prairie wetlands exhibited the highest dry bed relative frequency (0.70) while the cypress-marsh wetlands had the lowest dry bed relative frequency (0.35). Additionally, on average, the wetland normalpool elevations corresponded to a relative frequency of 0.96, indicating 96% of the recorded wetland water elevati ons are equal to or less than the normal-pool elevation. This value is consistent for all wetland cat egories, ranging from 0.95 (Hardwood) to 0.99 (Cypress-Marsh). Table 3.7. Dry bed (DB ) and normal pool ( NP ) probability. Regional Wetlands Cypress Marsh Statistic F (DB ) F (NP ) ho (cm) F (DB ) F (NP ) ho (cm) F (DB ) F (NP ) ho (cm) Mean 0.39 0.96 74.4 0.38 0.96 61.8 0.38 0.96 94.6 StD 0.20 0.04 43.2 0.18 0.04 25.5 0.20 0.04 59.8 Median 0.39 0.98 59.6 0.37 0.97 55.3 0.31 0.97 72.8 Min 0.02 0.76 36.0 0.02 0.76 36.0 0.13 0.86 46.9 Max 0.86 0.99 250.2 0.86 0.99 148.7 0.65 0.99 235.0 Range 0.84 0.22 214.3 0.84 0.22 112.8 0.53 0.13 188.1 Cypress-Marsh Hardwood Wet Prairie F (DB ) F (NP ) ho (cm) F (DB ) F (NP ) ho (cm) F (DB ) F (NP ) ho (cm) Mean 0.35 0.99 131.4 0.36 0.95 87.4 0.70 0.97 56.1 StD 0.26 0.00 74.4 0.12 0.05 20.5 0.19 0.02 25.7 Median 0.35 0.99 134.1 0.41 0.99 93.0 0.75 0.98 45.7 Min 0.05 0.98 62.2 0.23 0.89 64.6 0.49 0.95 37.2 Max 0.64 0.99 250.2 0.45 0.99 104.5 0.85 0.99 85.3 Range 0.59 0.00 188.1 0.22 0.09 39.9 0.36 0.04 48.2

PAGE 97

85 Last, the maximum wetland pool depth ( ho) estimate is shown on Table 3.7. The pool depth approximation was included to s how depth of the wetland categories that correspond to the normal pool and dry bed elev ation differences. Overall, the regional wetlands are 74.4 cm deep. The wet prairie wetlands are the shallowest (56.1 cm), and the cypress-marsh wetlands are the deepest (131.4 cm). 3.5.2. Wetland Category Water-Level Data Com parisons Kolmogorov-Smirnov Tests 3.5.2.1. Wetland Category Comparisons The two-sample Kolm ogorov-Smirnov tests were performed on the wetland category empirical distribution functions illustrated on Figure 3.7. Summary statistics of the abridged relative water-level data used to populate the respective we tland categories are listed in Table 3.8. All of the wetland category comparisons failed the respective KStest, h = 1 (Table 3.9). The table results are pr esented in a matrix format in which each wetland category (first column) is compared to every other wetland category (top row). The maximum vertical deviation between th e respective category distribution curves ( D stat) vary between 0.10 (Cypress vs. Hardwood) and 0.36 (Hardwood vs. Wet Prairie) for the wetland water levels. These results indicate there are statistical differences, albeit unobvious, between the representati ve probability distributions.

PAGE 98

86 -500 -400 -300 -200 -100 0 100 200 300 0 10 20 30 40 50 60 70 80 90 100 Wetland Relative Water Levels (cm)Cumulative probability (percentile) Cypress Marsh Cyp-Marsh Hardwood Wet Prairie Figure 3.7 Wetland category empirical distri bution functions, relative water levels based on the dry bed datum ( DB ). Table 3.8. Wetland category mont hly water level description. Relative Water Level Summary Statistics (cm) Cypress Marsh CypressMarsh Hardwood Wet Prairie Wetlands 36 9 5 3 3 Records 2,525 663 353 216 224 Mean -5.4 11.1 22.4 10.0 -46.3 StD 68.5 69.5 98.1 57.3 69.1 Median 15.2 23.3 46.0 22.7 -39.9 Min -491.9 -209.4 -388.6 -138.4 -205.7 Max 208.5 156.4 132.3 104.9 103.3

PAGE 99

87 Table 3.9. Wetland category water-level distribution comparisons, KolmogorovSmirnov test results. Category Cypress Marsh Cypress-Marsh Hardwood h p D stat h p D stat hp D stat h p D stat Cypress 0 Marsh 1 0.00 14% 0 Cypres-Marsh 1 0.00 36% 1 0.00 24% 0 Hardwood 1 0.03 10% 1 0.04 11% 1 0.00 27% 0 Wet Prairie 1 0.00 32% 1 0.00 35% 1 0.00 47% 1 0.00 36% 3.5.2.2. Regional Wetland Groups KS-tests we re performed on the aggregate empirical distribution functions illustrated on Figure 3.8. Summary statistics of the rela tive water level data used to populate the respective wetland groups are listed in Table 3.10. Each of the wetland group comparisons failed the respective KS-test, h = 1 (Table 3.11). As before, the table results are presented in a matrix format in which each wetland group (first column) is compared to every other wetland group (top row). Th e maximum vertical de viation between the respective group distribution curves ( D stat) vary between 0.15 (Green Swamp vs. Morris Bridge) and 0.26 (Green Swamp vs. Starke y) for the wetland water levels.

PAGE 100

88 -400 -300 -200 -100 0 100 200 0 10 20 30 40 50 60 70 80 90 100 Wetland Relative Water Levels (cm)Cumulative probability (Percentile) Green Swamp Morris Bridge Starkey UHFDA Figure 3.8 Regional wetland group empirical distribution functi ons, relative water levels based on the dry bed datum ( DB ). Table 3.10. Regional wetland group monthly data description. Relative Water Level Summary Statistics (cm) Green Swamp Morris Bridge Starkey UHFDA Wetlands 9 9 17 9 Records 604 692 1,199 669 Mean 0.8 -16.9 16.4 -31.0 StD 48.0 84.5 69.0 90.7 Median 17.2 8.5 22.3 -17.4 Min -169.5 -388.6 -205.7 -491.9 Max 73.7 156.6 248.1 190.8

PAGE 101

89 Table 3.11. Regional wetland group wate r-level distribution comparisons, Kolmogorov-Smirnov test results. Regional Group Green Swamp Morris Bridge Starkey h p D stat h p D stat h p D stat Green Swamp 0 Morris Bridge 1 0.000.14 0 Starkey 1 0.000.26 1 0.000.14 UHFDA 1 0.000.25 1 0.000.14 1 0.00 0.24 3.6. Discussion 3.6.1. Wetland Hydrologic Evaluation (Frequency Analysis) The em pirical distributions representing th e wetland and upland relative water levels, based on the ground elevation ( GE ) at the well and the bed elevation of the wetland ( DB ), did not provide a clear means of distinguish ing the different wetla nd types (Figures 3.5 and 3.6). The distributions representing the respective wetland water levels (Figures 3.5 and 3.6 Chart A) and upland groundwater levels (Figures 3.5 and 3.6 Chart B) cross at almost every percentile. Based on the di stribution overlap and th e reported standard deviations in Tables 3.5 and 3.6, it is appa rent there was high variability in both the surface and sub-surface water levels between th e 56 study wetlands at each of the target percentiles (10th, 50th, and 90th). Overall, the highest variability was observed at deep groundwater levels (10th percentile), and the lowest va riability was observed at high water levels within the wetland exte nts and in the surrounding uplands (90th percentile). This was generally the case for each wetland category as well.

PAGE 102

90 The increased variability at th e deep groundwater levels (10th percentile) can, in part, be attributed to the depth of the water table. In general, the range of recorded water levels across these wetlands increases as the depth to the water table increases. The point is illustrated in Figure 3.9, which presents a linear regression between the depth to the water table and the range of recorded upland water levels associated with each wetland. The water-table depth was approximated using th e deep groundwater leve ls corresponding to the 10th percentile. The deep upland groundwater levels are su mmarized in Table 3.5. The upland groundwater level ranges for each we tland were calculated from the summary statistics presented in Table 3.4. All upl and water levels were normalized by the ground elevation at the upland well. -600 -500 -400 -300 -200 -100 0 0 100 200 300 400 500 600 Water-table Depth (cm) (Based on 10th percentile)Range of Recorded Water Levels (cm)Upland Groundwater Levels y = 0.96*x + 63 Figure 3.9 Comparison of water-table depth and upland water level range.

PAGE 103

91 Local water-table depths associated with the wetlands in northern Tampa Bay region can be influenced by several factor s or combinations of factors. One is the location of the wetland in the region. For instance, wetlands located in the flat coastal areas on the western side of the region (Figure 3.1) will have local water tables that remain shallow much of the year. Other, wetlands located away from the coast in higher slope areas with more conductive soil types enable higher watertable variability in the groundwater levels throughout the year or during dry years. Yet ot her wetlands are located in areas such as the Green Swamp where the water-table remains near the land surface due to the hydrogeologic characteristics of th e area. In this area, there is an Upper Floridan aquifer groundwater mound that hinders the downward m ovement of water, as well as high clay content in the soils providing surficial aquifer confinement which stabilize the water table (Spechler and Kroening 2007). Further, conf ining layers can be present throughout the region preventing water-table levels from droppi ng. In places wher e the confining layer is thin or has been breached groundwater leve ls can drop considerably Last, the depth of the water-table can be influenced by anthropogenic stresses such as groundwater pumping, which could lower local groundwat er levels, and surface water augmentation which could raise and/or stab ilize local groundwater levels. Conversely, water levels at the 90th percentile within the wetland extents and in the uplands generally had the lowest variability (Tables 3.5 and 3.6). The reduced waterlevel variability within the wetlands is due to the bowl shape near the wetland extents (Haag et al. 2005; Nilsson et al. 2008). The upper bowl near the maximum pool depths tends to fans out, with gradual or low-grad ient topography. Hen ce, for a given water-

PAGE 104

92 level change the incremental volume increase is at its maximum. Therefore, large volume changes are required to see signifi cant changes in the surface water level. Finally, at the maximum pool depth many wetla nds exceed a discharge invert limiting the water levels. As a result, the surface-water levels of the various wetlands are more consistent at the 90th percentile than at the 50th percentile (median water levels) and the 10th percentile (deep groundwater levels). C onsequently the variabil ity in the recorded water levels between the wetlands is lower. The lower variability in the shallow upland groundwater levels (90th percentile) can be attributed to the local topography. As the depth to the water-table decreases, the groundwater becomes an expression of the lo cal topography of the land surface, which is a physical boundary. Also, water levels above ground are associated with rapid surface water overland flow. As a result, high gr oundwater levels in the upland wells become consistent from one wetland to another; hence, the variability of the upland groundwater levels decreases at shallo w water-table depths. 3.6.2. Frequency of Water Levels at Dry Bed ( DB ) and Norm al Pool ( NP ) Two important datums associated with wetlands, the dry bed elevation ( DB ) and the normal pool elevation (NP ) were identified on the respective wetland water-level empirical distribution functions. Based on th ese elevations, there was at least some standing water in all study wetlands on av erage 61% of the time over the study duration (Table 3.7 Regional Wetlands). A compar ison of the cypress and marsh categories, which have the largest number of representa tive wetlands, revealed some standing water

PAGE 105

93 was present in each category 62% of the time, indicating no difference in this variable for two very different vegetative types. The normal pool elevation is identified by stain markers, mosses and unique plant species that are typically found at the wetland extent s. In order for these indicators and plant species to exist, they can onl y be inundated by water a small percentage of time. This analysis verified that the normal pool eleva tions supplied by SWFWMD were, in general, exceeded only 4% of the time for the 56 study wetlands (Table 3.7). This result was approximately the same for all of the wetland categories (Table 3.7). Overall, this finding was consistent with the Districts goal of using vegetation that is intolerant of flooding as an indication of th e normal pool elevations. 3.6.2.1. Analytical Model Application Several analytical m odels used to predic t wetland storage require the maximum pool depth as an input parameter (Hayashi a nd van der Kamp 2000; Nilsson et al. 2008; O'Connor 1989). The maximum pool depth can be obtained via survey or estimate. Surveys are the most accurate measure of the depth. However, depending on the number of wetlands within a study ar ea surveys may be impractical and cost prohibitive. Estimating the wetland pool depth is much le ss costly and may be practical for large study areas, albeit estimation may introduce error into a hydrologic modeling analysis.

PAGE 106

94 The difference in elevation between the wetland normal pool and the dry bed is a good approximation of the wetland maximum pool de pth. The maximum pool depth estimates ( ho) for each wetland category are listed in Tabl e 3.7. These pool depths can be used in analytical models to represent the various wetland categories. Additionally, summary statistics are provided to en able water resource engineer s and hydrologists to account for errors that may be introduced from the norma l variability in the pool depth estimate. Further, the dry bed and normal pool probability results can be used as a calibration tool for the models. For instance, based on the probability data, on average the water levels within the cypress wetlands should be belo w the normal pool eleva tion 96% of the time, and the wetlands should generally be dry 38% of the time. Therefore, some standing water should be present within these indices 58% of the time for the cypress wetlands in order for a model performance to be statistically similar. 3.6.3. Combined Wetland Water-Level Data Comparisons The visual com parison of the individual wetland frequency distributions did not show hydraulic differences between the various wetl and categories. However, combining the water-level data of each wetland within a category revealed that all five wetland categories have statistically unique water-level characteristics (Table 3.9). Based on this finding, the water-level variability associated with individual wetla nd categories could be represented by distinct probability density functions. These functions could be incorporated into hydrologic models to repres ent wetland water-level fluctuations or used to test the model validity.

PAGE 107

95 Similarly, comparisons of four regional wetland groups indicated the water-level variability related to each group are statistica lly different (Table 3.11). This implies the water-level behavior of th ese wetland groups are strongl y influenced by the local hydrogeology, climatology and anthropogenic stresses. 3.7. Application: Impact ed W etland Identification A simple technique of identifying wetlands that may be influenced by anthropogenic activities or natural stresses is presented here For instance, the water levels within and around a wetland may be lowered due to groundwa ter pumping, or the water levels could be augmented from the inadvertent or inte ntional addition of water to the wetland and surrounding upland. Individual wetland and/or upland water-level distributions are compared to the trend distributions (Charts C and D) in Figure 3.6 by visual and numerical inspection. The inspection is us ed to identify two hydr ologic conditions: 1) raised water levels and 2) lowered water le vels. Raised water levels are evident by frequency distributions that lie above the tr end standard deviation, and lowered water levels are evident by frequency distributions th at lie below the trend standard deviation. Also, frequency distributions indicative of outside influences or stresses may deviate from the trend distribution in the form of a ve rtical line or horizontal line. A vertical distribution curve would indicate high water-level fluctuations or variability, while a flat or horizontal curve would indica te low or minimal water-level fluctuations. For example, wetlands that are augmented or in the Gr een Swamp (natural hydrogeologic conditions) may exhibiting minimal wate r-level fluctuations.

PAGE 108

96 Wetlands that exhibit stressed water-level beha vior have all or pa rt of the respective frequency distribution (0th percentile to the 100th percentile) lie outside the standard deviation range (outliers), or deviate from the general dist ribution trend. Depending on the number outliers and the magnitude of the de parture from the standard deviation limit, or trend in the case of vertical and horizontal curves, an i nvestigation can be conducted to determine the hydrologic st ate of the wetland. Two examples of this technique are demonstr ated in Figure 3.10. The Figure shows the respective wetland water-level distribution curves for cypress-marsh wetland 331 (Chart A) and cypress wetland 1322 (Cha rt B) overlaid on the wetla nd water-level distribution trend curves, Figure 3.6 Chart C. The di stribution curve for we tland 331 is above the trend standard deviation, suggesting the wetland may be under the influence of elevated groundwater levels or surface-water runoff, or augmented. The distribution curve for wetland 1322 is below the trend standard deviation, suggesting the water levels associated with the wetland have been lowe red possibly by reduced surface water flows to the wetland.

PAGE 109

97 0 d10 20 30 40 d50 60 70 80 d90 100 -250 -200 -150 -100 -50 0 50 100 150 200 250 PercentileRelative Water Level (cm) [dry bed datum]Chart A Cypress-Marsh Wetland 331 Mean EDF Std EDF ww 2157 0 10 20 30 40 50 60 70 80 90 100 -250 -200 -150 -100 -50 0 50 100 150 200 250 PercentileRelative Water Level (cm) [dry bed datum]Chart B Cypress Wetland 1322 Mean EDF Std EDF ww 1974 Figure 3.10 Individual wetland outlie r distribution functions.

PAGE 110

98 Investigations into the hydrologic state of both wetlands were conducted since both of these wetlands have distribution curves that lie outside the st andard deviation range of the trend at all percentiles. Wetland 331 is locate d in the immediate vicinity of a waste water treatment facility containing two sludge la goons and a spray field (Appendix D). The waste water treatment discharge practices appear to have artificially raised the local water table and/or surface flows to the wetland, hence increasing the frequency of elevated wetland water levels. Conversely, historic observations of we tland 1322 indicate the wetland experiences greatly depressed water levels (Appendix D). The reason of the depressed water level is unknown at this time. Further hydrologic inve stigations need to be conducted to determine the cause. Additionally, Appendix D contains the investigation summaries perf ormed by the District for al l 56 study wetlands listed in Table 3.1. This analysis was conducted on all 56 wetlands presented in Table 3.1. The respective wetland water-level empirical distributions we re compared to the trend distributions (Figure 3.6 Chart C). The empirical distributions were evaluated at the 10th, 50th and 90th percentiles to identify water-level outlier s and potential hydrologic stresses. Twenty wetlands, including the two discu ssed previously, were identifi ed as having low or high outliers (beyond the stan dard deviation range) (Table 3. 12). The wetland locations are shown on Figure 3.11. Each of the respectiv e wetland water-level distributions has an outlier at one or more of the target percentiles, noted by a ( ) or ( ) in the representative column in Table 3.12. The blue ( ) markers indicate the respective wetland water level was to the high side of th e trend standard deviat ion range, and the red

PAGE 111

99 ( ) markers indicate the wetland water level was to the low side of the deviation range. Thirteen wetlands exhibited distribution outliers at the 10th percentile (7 low, 6 high), 12 at the 50th percentile (5 low, 7 high), and 10 at the 90th percentile (3 low, 7 high) (Table 3.13). To put this in perspective, 23% (13 of 56) of the wetlands studied have water-level outliers at the 10th percentile, 21% at the 50th percentile, and 18% at the 90th percentile. Table 3.12. Wetland percentiles exceedi ng one standard devi ation (outliers). Wetland Well Outliers per Target Percentile Type Location UID ID 10th 50th 90th 21 1943 M S 81 1971 M 154 1948 CM CB 183 1954 M MB 261 2076 C 276 1984 C SP 331 2157 CM S 379 1951 CM MB 489 1927 C S 501 1941 CM S 505 1950 C MB 605 1966 M 1317 1934 WP S 1320 1960 M CB 1322 1974 C UHFDA 1323 1975 C UHFDA 1324 1976 C UHFDA 1325 1977 WP UHFDA 3713 2064 C 4352 2253 C S = High Frequency Distribution outliers = Low Frequency Distribution outliers

PAGE 112

100 # # # # # # # # # # ## # # # # # # # #CROSS BAR RANCH STARKEY SOUTH PASCO ELDRIDGE WILDE MORRIS BRIDGEPASCO HILLSBOROUGH POLK PINELLAS SUMTER HERNANDO 81 21 605 501 489 331 261 183 154 1320 4352 505 379 276 3713 1325 1324 1323 1322 1317 WellFields GreenSwamp LAKE UPLAND NORTHERN GULF COASTAL LOWLANDS WESTERN VALLEY Impacted Wetlands#High EDF#Low EDF 081 6 4 Kilometers4 Figure 3.11 Wetlands with empirical distribution outliers ( 1 StD from mean).

PAGE 113

101 The wetlands that exhibit stre ssed water-level behavior are located throughout the region, from coastal areas to the Green Swamp (F igure 3.11). Reasons for the respective elevated or depressed water leve ls for given wetlands are difficult to determine from their region location, because wetlands with high water levels (blue ( ) markers) as well as wetlands with low water levels (red ( ) markers) are located in the vicinity of one another. Further complicating matters, seve ral wetlands within and around the Starkey well field have elevated water le vels. Contrary to this, it w ould be expected that wetlands located on well fields or possibly in the vici nity of well fields would have lower water levels due to pumping stresses. This sugge sts that the local hydr olgeologic properties, such as soil composition and the presence or in tegrity of the intermediate confining unit, associated with the various wetlands could be the dominate factor affecting the water levels of these wetlands. Therefore, detailed hydrogeologic surveys need to be performed in order to determine the exact reason these wetlands exhibit the water level behavior. This analysis demonstrates that empirical (f requency) distribution f unctions can be used to identify potentially impacted wetlands wh ere traditional tempor al water-level plots may not. Also, based on this analysis, a pproximately 80% of the wetland water-level data are within the empirical trend standard deviation range (Figure 3.6 Chart C). This could be another indication th at the 56 west-central Florida wetlands exhibit similar hydraulic behavior.

PAGE 114

102 3.8. Conclusions The purpose of this chapter was to characterize wetland water-levels based on the probability of inundation and the frequency distri bution of the depth to the water table. Empirical distribution functions were devel oped from historic wetland water elevation records and upland groundwater elevation records associated with 56 different isolated wetlands in west-central Florida. The empirical distribution functions provide a means to analyze the water-level data using frequenc ies and probabilities of occurrence of water levels over time. Further, the distributions were used to compare the hydraulic characteristics of five wetland categories and four regional wetland groups, and to identify wetlands that are potentially under adverse hydrologic stresses. In general, standing water was present in these wetlands 61% of the time over the seven year study. Also, the water levels in th e wetlands exceeded the normal pool vegetative markers only 4% of the time. These levels re present critical indica tors for the hydrologic condition or state of the wetland, and may serve as useful para meters to calibrate or test hydrologic models. Also, an estimate of th e maximum pool depth can be obtained using the dry bed and normal pool elevations. This depth can be incorpor ated in the storage model presented in Chapter Two to devel op wetland stage-storage characteristics. Surprisingly, individual wetla nd categories could not be identified simply by viewing the representative empirical distribution functions associated with each wetland type. The variability in water levels between th e regional wetlands and wetland types was significant. Consequently, individual wetland categories could not be identified via

PAGE 115

103 simple inspection of the respective water-le vel distributions. Additionally, there was higher variability in the gr oundwater levels beneath the wetlands than in the surfacewater levels within the wetlands. The high variability in the groundwater levels is most likely a reflection of varying water-table dept h across the west-central Florida region. The depth of the water table can be a ffected by pumping stresses, surface water augmentation or local hydrogeology. The reduced variability at high water levels within the wetlands is attributed to the natural shape of the wetlands. The pooled water fluctuations are small due to large changes in volume near the wetland extents. Frequency distributions can be used as a comparison tool to identify similarities and differences between representati ve data sets and to identify a typical hydrologic behavior. Statistical tests performed on frequency dist ributions representing the combined water levels of the various wetlands within a wetland category showed significant differences in the water-level behavior for the specific wetland categories. Further, wetlands were identified that might be adversely influenced by anthropogenic activities or natural stresses. This was accomplished by simply comparing the respective wetland empirical distribution functions to the ge neral trend distribution curve de veloped in this chapter to determine the number and magnitude of water-l evel measurements that fall outside the trend standard deviation range. Hence, pr obability distributions can be used in hydrologic modeling to test the water-level behavior, trends, or stresses to individual wetlands and wetland categories.

PAGE 116

104 3.9. Acknowledgments I would like to graciously acknowledge the contributions of Michael C. Hancock, Senior Professional Engineer, R esource Projects Department, Southwest Florida Water Management District, who provide d the wetland database used in this dissertation. He provided valuable advice and insights throughout the development of this dissertation.

PAGE 117

105 CHAPTER 4 THE EXTENT AND PREVAL ENCE OF GROUNDWATER RECHARGE/DISCHARGE CONDITIONS IN WEST-CENTRAL FLORIDA ISOLATED WETLANDS 4.1. Introduction Closed-bas in wetlands are perhaps the most numerous and prominent freshwater systems in west-central Florida (Dah l 2006; Lee et al. 2009). These wetlands are hydrologically connected to the shallow su rficial aquifer (Haag et al. 2005), and in many instances hydrologically connected to the deeper Florid an aquifer where the confining layer is breached or thin due to subaerial erosion or su bterranean karst collapse (Lee et al. 2009). Groundwater levels in and ar ound these wetlands in both th e surficial and Floridan aquifers are typically within a few meters of the land surface for much of the year. Therefore, these wetlands provide a direct interface between surface water and groundwater during much of the year. In west-central Florida, pr ecipitation is approximately 140 centimeters per year, with approximately 60% of the precipitation fa lling during the four-month period of JuneSeptember (Southeast Regional Climate Center 2010). The mean evapotranspiration (ET) for the region is approximately 100 centimet ers per year (Bidlake et al. 1996), and is generally considered higher in wetlands th an uplands (Hill and Neary 2007). Because of

PAGE 118

106 this, it is often assumed that wetlands serve as drains that lower the local water-table thereby creating local water-table depressions (Whigham and Jordan 2003). However, empirical data generally supporting this assumption are largely lacking. There are sound hydrogeologic reasons that may indicate wetlands are groundwater recharge features on average or at least some of the time. Evapotranspiration in these wetlands is temporally variable by season, with wetland evapotranspiration strongly declining in the dry season due to a reduced availability of water and a lowering of the water-table. Also, winter time senescence of the leaves of the typical dominant or codominant tree, Bald Cypress ( Taxodium distichum ), results in reduced evapotranspiration. Furthermore, the specific yiel d (Sy), defined as the volume of water a water-table aquifer releases from or takes into storage per uni t aquifer area per unit change in water-table elevation (Freeze and Cherry 1979), is much higher in the surface-wate r systems of these wetlands, when water is present, than in the groundwater systems of the surrounding uplands. In surface-water systems, such as these wetlands, specific yield typically is assumed to be 1.0 (Hill and Neary 2 007; Mitsch and Gosselink 2000), while in groundwater systems, such as the uplands adjace nt to these wetlands, specific yield is on the order of 10-1 (Johnson 1967). Therefore, surface-water drawdown in the wetlands will be lower than groundwater drawdown in the uplands. Other possible explanations are: 1) the leakage through the wetlands may be slower than the leakage in the intermediate confining unit of the surrounding uplands, and 2) the wetland can be located at a topographic low point in the watershed, a bove the water table, th at can collect runoff from the surrounding upland areas and rechar ge the local groundwater system.

PAGE 119

107 The objective of this chapter is to characterize the groundwater recharge potential or trends of 56 various isolated wetlands in west-central Florida by comparing wetland water levels to surrounding upla nd water levels. It is hoped that this empirical data analysis will provide new insight into the groundwate r recharge or discharge characteristics of these wetlands that has larg ely been lacking to date. Long-term, paired wetland-upland monitoring well data as well as peak dry season (e.g., March-May) and wet season (e.g., July-September) data were used to determine the water-level relationships and recharge characteris tics between these wetlands and surrounding uplands. 4.2. Methods 4.2.1. Head Differences between Paired Wetland and Upland Water Levels The water e levation data for the wetland wells and upland wells presented in Chapter 3.2.3 were used to evaluate the groundwater recharge between the 56 various isolated wetlands and surrounding uplands. To ensure representative analysis, only paired wetland and upland water eleva tion records measured on the same date over the seven year study were used in this analysis. The groundwater recharge conditions were evaluated by calculating the difference in hydraulic head between the wetland water elevations and upland groundwater elevations for the matched data records. This net hydraulic head was used to determine the poten tial flow of the surfic ial aquifer, either into the wetland (groundwate r sink) or out of the wetla nd (groundwater source).

PAGE 120

108 Additionally, paired wetland a nd upland water elevation data, measured on the same date, from the peak dry season (March May) and the peak wet season (July September) were used to understand the seasonal rech arge characteristics of these wetlands. Typically the lowest measured water-table elevations occur during or near the end of the dry season, and the highest wate r-table elevations are measured during or near the end of the wet season. The two seasons represent the extreme water-table elevations, which help shed light on the surficial aquifer rech arge conditions of these wetlands under these hydrologic conditions. Empirical distribution functions (EDFs) defined in Chapter 3.2.4 were used to provide additional insights into the groundw ater recharge characteristic s of these wetlands at low, median and high water levels. Paired wetla nd and upland water elevations, measured on the same date, were used to develop re presentative wetland and upland frequency distributions. The net hydraulic head asso ciated with each distribution frequency or percentile was calculated by s ubtracting the upland water el evation from the wetland water elevation at the particular relative frequency, i.e. the 10th percentile. The hydraulic head differences were then evaluate d at the three target percentiles: 10th representing low water elevations, 50th representing median water elevations, and 90th representing high water elevations.

PAGE 121

109 4.2.2. Seasonal Group Water Level Analyses This analy sis focused solely on the wetland a nd upland water elevati on data recorded in the peak dry season (March May) and th e peak wet season (July September). Monthly data, based on a single measurement or the mean of all water elevations recorded during the respective month, were used in this analysis. Monthly values were used to eliminate any bias brought about by co mbining data sets comprised of daily water elevations and monthly elevati ons. Further, the use of mont hly data did not influence the results because data sets comprised of monthly data are not statisti cally different from data sets comprised of daily data (Appendix C). Seasonal analyses were conducted by groupi ng the dry season and wet season wetland water elevations and the upla nd groundwater elevations. Relative water levels (RWL) were developed to provide a means to co mbine the respective wetland and upland well data into four seasonal groups: 1) dry season wetland water levels (WWDS), 2) dry season upland groundwater levels (UWDS), 3) wet season wetland water levels (WWWS), and 4) wet season upland groundwater levels (UWWS). The relative water levels were developed by normalizing the recorded water elevation data with respect to th e associated wetland dry bed (DB) elevation (Table 3.1). The normalized data is presented as centimeters above or below the wetland dry-bed datum.

PAGE 122

110 Further, empirical distribution functions we re developed from the combined wetland and upland water levels associated with the four seasonal groups to provide additional insights into the groundwa ter recharge/discharge at low, median and high water levels. The frequency distributions were evaluated at the 10th, 50th, and 90th target percentiles. A battery of Wilcoxon rank sum tests were performed on the paired wetland well and upland well seasonal water level data to test whether sample combinations are drawn from the same population (Dallal 2007). The Wilcoxon test is the nonparametric alternative to the student t test. This test was implemented due to the non-normal distribution of the wetland and upland water level data (Appendix B). The Wilcoxon rank sum test is used to evaluate two distri butions to determine if the data sets are independent samples from identical continuous distributions with equal medians (null hypothesis), against the alternat ive that they do not have equal medians. The Wilcoxon test returns the result of h = 1 which indicates a rejection of the null hypothesis at the 5% significance level, and h = 0 which indicates a failure to reject the null hyp othesis at the 5% significance level. Si mply, a test result of h = 0 indicates the distributions are statistically similar and can be consider ed representative of one another. The Wilcoxon rank sum test was used to comp are four combinations of the seasonal relative water-level data. The first test evaluated wetland water levels recorded during the dry season and wet season re spectively. The test was used to determine if the recorded wetland water levels in the dry seas on were statistically different than those recorded during the wet season. Likewise the second test evaluated the seasonal upland

PAGE 123

111 groundwater levels to determine if the gr oundwater levels recorded during dry season were statistically different than those reco rded during the wet seas on. The third test focused solely on the wet season water levels. The test was setup to determine if the wetland water levels were significantly differe nt from the adjacent water-table elevations, recorded at the upland wells, during the wet seas on. Last, the fourth test focused solely on the dry season water levels. The test was setup to determine if the wetland water levels were significantly differe nt from the adjacent water-ta ble elevations during the dry season. 4.3. Results 4.3.1. Head Difference between Wetland and Upland W ater Levels (Surficial Aquifer) 4.3.1.1. Standard Statistical Analyses The hydraulic head com parison showed wetland water elevations were generally 9.2 cm higher than the paired upland surficial groundwater elevations over the seven year study (Table 4.1 All). The head differences ra nge from -177.7 cm to 227.7 cm. Also, based on this data set, 36 wetlands had positive he ad differences, and 20 wetlands had negative head differences. Table 4.1. Wetland-upland head difference. Statistic Separation Head Difference (cm) Distance All Dry season Wet season Data Count (m) 5178 1237 1255 Mean 56 9.2 12.8 0.3 StD 25 37.1 39.3 34.7 Median 57 3.7 5.5 -1.5 Min 8 -177.7 -139.6 -177.7 Max 149 227.7 215.2 189.6 Range 142 405.4 354.8 367.3

PAGE 124

112 Additionally, on average the wetland water el evations were 12.8 cm higher than the surrounding upland surficial groundwater elevat ions during the dry season (March May) (Table 4.1 Dry Season). The differen ce in the hydraulic heads ranged from -39.2 cm to 215.2 cm. During the wet season (July September), the wetland water elevations were nearly the same as the surrounding upla nd surficial groundwater elevations (Table 4.1 Wet Season). On average the wetland water levels were the only 0.3 cm higher than the surficial levels ranging from -177.7 cm to 189.6 cm 4.3.1.2. Frequency Analyses Em pirical distribution functions were deve loped to further investigate the wetland recharge characteristics at low, median and high water elevations. In general the water levels within the wetland extents were highe r than the upland groundwater levels at the 10th and 50th percentiles, 8.8 cm and 16.3 cm resp ectively (Table 4.2 All Records). However, the groundwater levels in the upl and were 2.2 cm higher than the wetland water levels at the 90th percentile. Furthermore, the an alysis revealed that 18 of the wetlands had positive head differences at all three of the target percentiles, and nine of the wetlands had negative head differences at all of the target percentiles. Table 4.2. Wetland surficial aquifer head differen ce at particular frequency indices. Statistic All Records (cm) Dry Season (cm) Wet Season (cm) 10th 50th 90th 10th 50th 90th 10th 50th 90th Mean 8.8 16.3 -2.210.418.5 5.91.5 2.7-8.5 StD 43.3 31.1 18.152.636.7 26.047.0 26.618.9 Median 1.5 9.6 -1.7-0.913.0 3.8-5.5 2.0-7.9 Min -103.0 -65.8 -59.1-102.1-57.9 -56.1-139.3 -77.4-46.0 Max 166.4 132.9 94.2210.0157.0 130.8164.6 95.791.1 Range 269.4 198.7 153.3312.1214.9 186.8303.9 173.1137.2

PAGE 125

113 The empirical distribution functions repres enting the wet season a nd dry season wetland water elevation and upland groundwater elevation head diffe rences showed slightly different trends (Table 4.2 Dry Season). On average, the dry season water levels were higher in the wetlands than th e upland groundwater levels at each target percentile as indicated by the positive head differences: 10th (10.4 cm), 50th (18.5 cm) and 90th (5.9 cm). Additionally, the analysis revealed that 19 of the wetlands had positive head differences at all of the target percentiles while 10 of the wetlands had negative head differences at the target percentiles. Unlike the dry season distributions, the wet season distributions showed the wetland water elevations were slightly higher than the upland groundwater elevations at the 10th and 50th percentiles, 1.5 cm and 2.7 cm respectiv ely (Table 4.2 Wet Season). However, the upland groundwater levels were on aver age 8.5 cm higher than the wetland water levels at the 90th percentile. In addition, eight of the wetlands had positive head differences at all of the target percentiles whereas 21 of the wetla nds had negative head differences at all target percentiles. 4.3.2. Seasonal Group Water-level Conditions 4.3.2.1. Standard Statistical Analyses The four seasonal wetland and upland relativ e w ater level groups: 1) dry season wetland water levels (WWDS), 2) dry season upland groundwater levels (UWDS), 3) wet season wetland water levels (WWWS), and 4) wet season upland groundwater levels (UWWS) are represented by the box-and-whisker plots shown on Figure 4.1. The wetland water levels

PAGE 126

114 were generally higher than the upland surficial groundwater levels in each season (Figure 4.1 and Table 4.3). The average water leve ls within the wetland extents were 21.1 cm below the wetland bottom, and the average surf icial groundwater levels in the adjacent uplands were 37.8 cm below the wetland bo ttom during the dry season. Conversely, during the wet season the mean water levels within the wetlands were 13.0 cm above the wetland bottom, and the mean surficial groundw ater levels in the upland were 14.4 cm above the wetland bottom. Also, both the wetl and and upland surficia l water levels were regularly below the wetland dry bed elevation in the dry season, and above the dry bed elevation in the wet season. WWds WWws UWds USws -500 -400 -300 -200 -100 0 100 200 300 Seasonal Water Level GroupsRelative Water Level (cm) [dry bed datum] = Median water level = Inter quartile range = Outliers Figure 4.1 Seasonal wetland and upland surficial aquifer water levels.

PAGE 127

115 Table 4.3. Seasonal wetland and upland surficial water levels. Statistic Dry Season Water Levels (cm) Wet Season Water Levels (cm) Wetland Upland Wetland Upland Data Count 961 1075 946 965 Mean -21.1 -37.8 13.0 14.4 StD 76.6 89.4 72.2 82.3 Median -3.7 -32.0 32.9 30.5 Min -482.8 -493.2 -482.2 -484.0 Max 143.6 211.8 208.5 272.2 Range 626.4 705.0 690.7 756.2 4.3.2.2. Frequency Analyses Em pirical distribution functions were deve loped from the seasonal groupings of the wetland water levels and upland groundwater levels associated with the 56 wetlands. In general, the water levels at the upland su rficial well were lower than the corresponding water levels at the wetland well during the dry season (Table 4.4 Dry Season). This observation was true for all three of the distri bution target percentiles. For instance, at the 50th percentile the mean water levels in side the wetland were 6.3 cm below the wetland dry bed, and the mean surficial groundwater levels in th e nearby upland were 33.5 cm below the wetland dry bed. The wet se ason results were different. In general, the water levels at the upland surficial well were lower than the subsequent wetland well water levels at the 10th percentile (Table 4.4 Wet Seas on). Conversely, on average the wetland water levels were lower than the upland groundwater levels at the 50th and 90th percentiles. For instance, the mean groundwater levels in the adjacent uplands were 10.2 cm higher than the corresponding we tland water levels at the 90th percentile.

PAGE 128

116 Table 4.4. Seasonal wetland and upland surficial water levels at particular frequency indices. Statistic Dry Season Wet Season Wetland Water Levels (cm) Wetland Water Levels (cm) 10th 50th 90th 10th 50th 90th Mean -107.8 -6.3 49.5 -70.6 32.2 67.9 StD 62.4 47.2 32.7 65.8 43.3 28.7 Median -103.2 -2.6 40.8 -84.1 31.4 60.8 Min -268.5 -107.9 -12.5 -238.7 -136.9 29.6 Max 59.7 102.7 139.0 89.3 124.1 154.8 Range 328.3 210.6 151.5 328.0 260.9 125.3 Upland Groundwater Levels (cm) Upland Groundwater Levels (cm) 10th 50th 90th 10th 50th 90th Mean -125.8 -33.5 47.2 -81.5 34.7 78.1 StD 92.6 61.9 47.7 74.7 49.3 47.5 Median -104.4 -36.9 39.0 -77.9 32.2 65.7 Min -490.1 -175.9 -71.6 -310.6 -139.0 -44.5 Max 37.8 101.5 204.5 124.4 187.8 257.6 Range 527.9 277.4 276.1 434.9 326.7 302.1 4.3.2.3. Wilcoxon Rank Sum Tests The W ilcoxon rank sum test results compari ng the four seasonal water level groups presented in Figure 4.1 identified Test 3, th e comparison of the wet season wetland water levels and the adjacent upland groundwater le vels, as the only scenario that failed to reject (Table 4.5). In th is instance the null hypothesis indicated failure to reject h = 0, which suggests the two samples come from iden tical continuous dist ributions with equal medians. Furthermore, a strong result is evident based on the high p-value, 0.9.

PAGE 129

117 Table 4.5. Wilcoxon rank sum test results. Trial Condition h p 1 WWDS vs. WWWS 1 0.0 2 UWDS vs. UWWS 1 0.0 3 WWWS vs. UWWS 0 0.9 4 WWDS vs. UWDS 1 0.0 WWDS = Dry season wetland water levels WWWS = Wet season wetland water levels UWDS = Dry season upland groundwater levels UWWS = Wet season upland groundwater levels 4.4. Discussion 4.4.1. Head Differences between Paired Wetland and Upland Water Levels 4.4.1.1. Complete Data Set The water levels within the wetlands, whether as standing wa ter above the wetland bottom or as a groundwater levels below it, were generally higher than the adjacent upland water-table. Under natu ral conditions, water flows from high head conditions to low head conditions. Therefore, this head difference allowed water to flow out of the wetland into the surrounding upl and recharging the local surf icial groundwater system. A negative head difference would have indi cated water was flowing from the local groundwater system into the wetlands, hen ce lowering the local water-table. These wetlands were most often recharge features replenishing the local surficial groundwater system over the seven year study. Moreover, frequency analyses added additional insights into the recharge/discharge characteristics of these wetlands by evaluati ng the hydraulic head differences at low (10th percentile), median (50th percentile) and high (90th percentile) wetland and upland water

PAGE 130

118 elevations. The frequency analysis reveal ed the local groundwater system was usually recharged by the study wetlands at the low and median water elevations (Table 4.2 All Records). Conversely, at high water elevations (90th percentile) water was generally flowing from the upland either as surface runoff or local groundwater flow during wet conditions replenishing wetland water levels Additionally, the frequency analysis revealed that the study wetlands were groundw ater recharge zones 59% of the time over the seven years. 4.4.1.2. Seasonal Data Sets The seasonal data sets w ere used to characte rize the recharge/discharge characteristics of the study wetlands during the dry season (M arch May) and the wet season (July September). The head difference between the wetland water elevations and upland surficial groundwater eleva tions indicated these wetlands were generally groundwater recharge zones in the both the dry season and in the wet season. Albeit the net head difference in the wet season was very small, 0.3 cm (Table 4.1). These results were based on the mean water levels in each seas on, which may provide limited insight into the recharge characterist ics of these wetlands. For this reason, frequency analyses were used to further characterize the seasonal recharge/discharge conditions associated with the study wetla nds. In general, the study wetlands were groundwater recharge zones at all three targ et percentiles, representing low, median and high water levels, in the dr y season (Table 4.2 Dry Season). Further, the wetlands were groundwater recharge zones, on average, 61% of the time during the

PAGE 131

119 dry season. This result could be an indi cation that groundwater mounds form beneath these wetlands during the dry season. The mounds may be caused by slower leakage through the soils beneath the we tlands than in th e intermediate confining unit of the surrounding uplands. Also, these wetlands may be local recharge point s in the landscape because they are located in, or in effect ar e, low points in the lo cal topography that are above the water table. The frequency analysis of the wet season hydraulic head differences indicated the study wetlands were generally surf icial groundwater recharge features at the low (10th percentile) and median (50th percentile) water elevati ons, though the average head differences were very small, 1.5 cm and 1. 2 cm respectively (Table 4.2 Wet Season). Conversely, the head difference at the high water levels (90th percentile) indicated the movement of water during these wet periods was into the wetland and out of the local groundwater system. This could be an indica tion that the local water table becomes an expression of the local topogra phy during the wet season. As discussed previously, the local topography generally increa ses in elevation outside of the wetland extents; therefore as the water table approaches the land su rface the head difference between the wetland and upland water levels would decrease or reverse creating a possible scenario in which groundwater can flow into the respective we tland. Overall, the study wetlands were groundwater recharge features 47% of the time during the wet season.

PAGE 132

120 4.4.1.3. Consistent Recharge Feature Spatial Locations Each statis tical analysis revealed that a certain number of wetla nds were consistent groundwater recharge features or discharge features over the seve n year study period. The particular number and wetlands varied depending on the analysis conducted (Section 4.3.2). The number of wetlands that were consistent recharge or discharge features varied due in part to the t ype of analysis conducted. The frequency analyses are more robust than basis statistical analyses, as a re sult, provide a more detailed look into the recharge/discharge characterist ics of the study wetlands. Furthe r, the available data does not explain why certain features were consistent recharge or discharge features. There were mixed wetland types for both conditions and there was no geographical significance that could provide a reasonable explanation for these observations. The distinction between these wetlands could be in the way th ey were formed, i.e. karst collapse and/or small topographical depression, or simply due to the hydraulic va riations in these complex natural systems. Also, the location of the upland well relative to the wetland could affect the results. 4.4.1.4. Recharge Wetland Versus Flow-through Wetland The recharg e characteristics of a wetland might be misinterpreted since the head difference between the wetland water elevat ions and the adjacen t upland groundwater elevations was determined using a single paired wetland well and a single upland well (Rosenberry and Winter 1997). The head di fferences might indicate the wetland was a recharge feature when in fact the wetland wa s a flow through feature (Lee et al. 2009). Figure 3.2 shows the concept of a groundwater recharge we tland (A) and a groundwater

PAGE 133

121 flow through wetland (B). The problem can occur when an upland well is located on the predominant outflow side of a wetland, for instance along a regional water-table gradient toward a river. In this instance the head difference between the wetland water elevations and the upland groundwater elevations could indicate the wetl and is a recharge feature where in actuality it is a flow through feature. Further, th e paired well system might not provide sufficient evidence to determine di scharge conditions either. A study using a single upland well may indicate the local gr oundwater was flowing into a wetland, where in actuality the wetland was in a flow thr ough condition. Ideally, two upland monitoring wells would need to be installed, on opposite sides of the wetland along the regional water-table gradient, in order to determin e with certainty the recharge/discharge condition of the wetland.

PAGE 134

122 Figure 4.2 Conceptualized interactions of wetlands with (A) groundwater recharge and (B) groundwater flow through (m odified from Lee et al., 2009). 4.4.2. Seasonal Group Water Levels 4.4.2.1. Recharge Characteristics Wetland and upland water levels were groupe d into com prehensive dry season and wet season data sets to substantiate the head analysis results, and to gain a more complete understanding of the wetland recharge behavior at extreme water-table elevations. The initial analyses showed the wetland water levels were generally higher than the associated upland surficial groundwater levels during the dry season (Table 4.3 Dry Season), which coincides with the head results in Table 4.1. Even though the mean water levels within the wetlands and in the surrounding upl ands were below the dry bed elevation of the wetlands, the results s uggest these wetlands were gr oundwater recharge features during the dry season. During the wet season, typically the upland water levels were

PAGE 135

123 slightly higher (1.4 cm) than the associated wetland water levels implying the wetlands were groundwater sinks (Tab le 4.3 Wet Season). Furthermore, the frequency analysis re vealed the study wetlands were overall groundwater recharge features at all three target percentiles during the dry season, as evident by the wetland water levels being hi gher than the upland surficial groundwater levels, e.g. -6.3 cm versus -33.5 cm at the 50th percentile (Table 4.4 Dry Season). This is consistent with the head analyses presen ted in Table 4.1. During the wet season the study wetlands were in general groundw ater recharge features at the 10th percentile, and groundwater discharge features at the 90th percentile which corresponds to the head analysis (Table 4.4 Wet Season). However, at the 50th percentile (median water levels) the wetlands were groundwater depressions instead of recharge zones. 4.4.2.2. Wilcoxon Rank Sum Tests The com parison of the wet season wetland water levels and upland surficial groundwater levels, using the Wilcoxon rank sum test, indi cates the respective wetland water levels and upland surficial groundwater levels are statistically similar (Table 4.5 Trial 3). The result suggests the surface-wate r levels in the wetlands are as sociated with the depth of the surrounding water table, and implies th at these wetlands b ecome surface-water expressions of the local groundwater system as the water table approaches the land surface. Hence, as the water table rises dur ing the wet season, the water levels in the wetland and upland become similar as shown by similar mean and median water levels in Tables 4.3 and 4.4.

PAGE 136

124 The opposite test scenario comparing the dry season wetland wate r levels and upland groundwater levels failed the rank sum test (Table 4.5 Trial 4). The result indicates the wetland water levels and upland groundwater levels recorded in the dry season are statistically dissimilar. This test result indicates the wetland wa ter levels and upland groundwater levels uncouple during the dry s eason as the local water-table drops. 4.5. Conclusions The objective of this chapter was to char acterize the groundwater recharge potential between iso lated wetlands in west-centr al Florida and surrounding uplands, and to provide an empirical data analysis addressing the assumption that wetlands are local water-table depressions. Standard statistic al analyses showed these wetlands were generally groundwater recharge zones over th e seven year study. Additionally, seasonal analyses, utilizing water eleva tion data from the peak dry season (March May) and the peak wet season (July September), indicate d these wetlands were overall groundwater recharge zones during both seasons. Further, frequency analyses employing empiri cal distribution functi ons shed additional light into the recharge char acteristics of the study wetla nds. On the whole, these wetlands were groundwater rechar ge features at least 59% of the time, over the seven year study. The seasonal analyses indicat ed the wetlands were groundwater recharge zones 61% of the time during the dry season and 47% of the time during the wet season.

PAGE 137

125 Last, Wilcoxon rank sum tests showed the we t season wetland water levels and upland groundwater levels were statistically similar, which suggests that the wetland water levels are governed by the depth of the water-table. Also, this may indicate that the land surface of the wetland perimeters is the controlling/limiting elevation in the wetland and surrounding water table, and that runoff ma y be occurring under these conditions. Further, this indicates these wetlands ar e surface-water expres sions of the local groundwater system during the wet season. Conversely, the rank sum tests showed dry season wetland water levels and the asso ciated upland groundwater levels were statistically independent of each other, indicating the wetl and water levels and upland groundwater levels disassociate during the dr y season as the local water-table drops. Overall, the study showed these wetlands tend to be groundwater rech arge zones in the dry season and surface-water expressions of lo cal groundwater levels in the wet season. Also, the study revealed that these wetland s were largely groundwater recharge zones over the seven year period. Hydrologic mode lers should be aware of these findings to ensure regional models accurately represent water-table fluctuations.

PAGE 138

126 CHAPTER 5 PROBABILITY DENSITY FUNCTION REPRESENTATIONS OF WESTCENTRAL FLORIDA ISOLATED WETLAND WATER LEVELS 5.1. Introduction Probability distr ibution functions or probability models can be developed that represent wetland water-level fluctuations. These mode ls can be used to approximate the water levels of similar wetland types without havi ng to install and mon itor wetland wells and staff gauges. This could save time and money in data collection, and/or provide hydrologic modelers with a reasonable means to approximate ungaged wetland waterlevel fluctuations. Also, the use of probabi lity models to represent the water-level fluctuations of wetlands will enable e ngineers and hydrologists to make objective comparisons between individual wetlands, wetland types, assign probability to a particular event, a nd test hypotheses.

PAGE 139

127 A probability density function xfX describes the relative likelihood that a continuous random variable X takes on different values, and a cumulative distribution function 1xF X defines the probability P that random variable X is less than or equal to a real number 1x (Hogg and Ledolter 1987; Maidment 1993): 11 1 x X XdxxfxFxXP (5.1) These functions are known as continuous distributions because specific function parameters can be defined to describe the functio n or curve over an explicit range of data. The parameters can be arbitrarily defined base d on experience and/or calculated by fitting a probability distribution to a set of data. The parameters enable quantiles and expectations, e.g. wetland water elevations, to be calculated with the fitted probability mode l (Maidment 1993). Further, fitting a probability distribution to a se t of hydrologic data enables a large amount of probabilistic information to be efficiently summarized in the function and its associated parameters (Chow et al. 1988; Maidment 1993). This also provides a smooth and compact representation of the data. The objective of this chapter is to develop best-fit probability de nsity functions that accurately represent the water levels associat ed with the west-central Florida isolated wetlands presented in Chapter Three. Exp licitly, probability models were developed

PAGE 140

128 from the aggregate recorded water levels for each wetland category and all 56 wetlands in the northern Tampa Bay region. These models can be used to repres ent water levels of various wetland types in the absence of recorded hydrologic data. This is useful in hydrologic studies with large numbers of wetla nds and limited water-level data. Further, smallest extreme value models representing th e hydrologic characteri stics of five wetland categories, and four wetland groups were compared to identify any distinguishable differences or similarities. Last, an applic ation was presented demonstrating the use of the probability models to assign/predict wetland water levels based on antecedent moisture conditions, or projected moisture c onditions that could be used in extended period hydrologic model simulations. 5.2. Water-Level Data Wetland water-elevation data presented in Chap ter 3.2.3 were used to develop em pirical distribution functions represen ting the five wetland categories as well as a west-central Florida regional group comprised of all 56 study wetlands. Monthl y data, based on a single measurement or the mean of all wate r elevations recorded during the respective month (Chapter 3.4.1.3), were normalized with respect to the wetland dry-bed datum and combined into the respective wetland categor y and regional group (Table 5.1). Relative water level statistics, presented as centimeter above or below the dry bed datum, as well as the number of monthly wate r-level records are listed for each wetland category and the regional group in Table 5.1. The cypress wetland category is comprised of 36 wetlands and contains the most water-level record s, 2,525. The hardwood wetland category is comprised of three wetlands and contains the least water-level records, 216.

PAGE 141

129 Table 5.1. Wetland category monthly data description. Relative Water Level Summary Statistics (cm) Region Cypress Marsh CypressMarsh Hardwood Wet Prairie Wetlands 56 36 9 5 3 3 Records 3,980 2,525 663 353 216 224 Mean -0.8 -5.4 11.1 22.4 10.0 -46.3 StD 74.3 68.5 69.5 98.1 57.3 69.1 Median 16.8 15.2 23.3 46.0 22.7 -39.9 Min -491.9 -491.9 -209.4 -388.6 -138.4 -205.7 Max 248.1 208.5 156.4 132.3 104.9 103.3 5.2.1. Best-Fit Probability Distribution Identification The Anderson-Darling test (Stephens 1974) was used to identify the best-fit cum ulative distribution function type for the five wetla nd categories and for all the wetlands in the study (regional group). The Ande rson-Darling test provides a means to evaluate different cumulative distribution functions to determin e the best-fit for the respective data. Generally, the smaller the test statistic the be tter the distribution re presents the data. The test was applied to each of the aggregat e wetland category data sets as well as the regional group data set listed in Table 5. 1. The Smallest Extreme Value (SEV) distribution was identified by the AndersonDarling test, based on the smallest test statistic, as the best-fit distribution for all of the we tland categories except the marsh wetlands (Table 5.2). The Logistic distributi on was identified as the best-fit distribution for the marsh wetlands. However, a comparison of the two di stributions showed negligible gains in using the logistic distribution over the small extr eme value distribution when predicting the marsh water levels. Therefore, the smallest extreme value distribution will be used to represent the marsh water levels as well.

PAGE 142

130 Table 5.2. Comparison of alternative probability distributions, A nderson-Darling test. Anderson-Darling Test Statistic Distribution Type Region CypressMarsh CypressMarsh Hardwood Wet Prairie Smallest Extreme Value 39.5 22.6 7.4 1.2 0.7 0.8 Logistic 54.9 62.5 3.3 4.0 2.4 2.1 Normal 75.1 80.6 5.3 6.3 3.3 2.1 Gamma (3-Parameter) 79.2 83.8 5.9 6.7 3.7 2.3 Laplace 80.7 83.4 4.4 4.6 3.9 3.0 Largest Extreme Value 23.0 19.5 9.5 6.1 Exponential Gamma Loglogistic Lognormal Pareto Uniform Weibull Weak Distribution Fit Note: Bold numbers represent best-fit distributions. 5.2.2. Wetland Category Em pirical Distribu tions Empirical distribution functions were deve loped from the respective wetland category and regional wetland group data sets (summarized in Tabl e 5.1) using the procedure outlined in Chapter 3.4.1.1. Eleven target percentiles {0.05, 0.10:0.10:0.90, 0.95} were selected to represent the empirical distributions in this chapter. The 11 target percentiles were chosen to provide a detailed representation of the dist ributions developed from the water-level data sets. The associated relative water levels for the five wetland categories and the regional wetland group are summarized in Table 5.3. Additionally, the mean interdecile range ( IntDEDF) is listed for each wetland category. The cypress-marsh wetlands have the largest interdecile range (201.8cm) and the cypress wetlands have the smallest interdecile range (144.3 cm).

PAGE 143

131 Table 5.3. Wetland category empirical distri bution function statisti cs per percentile. Wetland Category Relative Water Levels per Percentile (cm) Region Cypress Wetlands Marsh Wetlands Percentile Mean StD Min Max Mean StD Min Max Mean StD Min Max 95th 69.4 38.9 25.4 224.6 56.7 22.9 25.4 120.4 95.0 49.9 37.5 184.4 90th 62.8 37.3 14.3 209.7 52.1 23.1 14.3 119.5 84.1 46.5 36.0 164.9 80th 50.7 35.5 -18.3 170.1 42.6 24.5 -18.3 108.5 67.5 44.0 24.4 147.2 70th 41.3 36.2 -38.4 158.5 34.6 25.9 -38.4 104.9 56.0 41.9 10.8 135.0 60th 29.4 36.9 -59.4 125.4 24.3 29.5 -59.4 97.5 41.4 38.8 -7.0 110.3 50th 16.5 40.0 -107.9 112.8 13.2 35.4 -107.9 82.0 26.9 38.0 -28.7 100.0 40th -0.7 44.6 -159.7 103.3 -2.2 41.9 -159.7 77.7 10.2 41.7 -41.1 89.3 30th -22.1 51.0 -192.6 101.3 -23.1 48.1 -192.6 74.4 -12.1 37.7 -68.0 29.7 20th -50.9 55.2 -204.8 86.1 -53.0 51.9 -204.8 67.7 -39.4 46.7 -101.5 18.7 10th -91.6 58.2 -238.7 59.7 -92.2 52.6 -238.7 55.5 -77.2 57.0 -165.2 -9.1 5th -117.3 61.3 -268.5 42.7 -119.2 58.5 -267.6 42.7 -97.8 63.9 -197.2 -16.4 IntDEDF 154.4 144.3 161.3 Cypress-Marsh Wetlands Hardwood Wetlands Wet Prairie Wetlands Percentile Mean StD Min Max Mean StD Min Max Mean StD Min Max 95th 121.8 62.3 65.5 224.6 76.9 29.2 44.8 102.0 49.7 19.8 36.0 72.4 90th 114.6 58.8 62.5 209.7 71.7 27.6 42.1 96.8 32.4 15.1 15.8 45.4 80th 97.7 50.9 40.2 170.1 58.6 20.5 39.6 80.4 10.9 21.0 -12.3 28.5 70th 85.9 54.0 27.3 158.5 50.4 19.6 37.7 73.0 -6.2 26.6 -31.4 21.6 60th 69.2 53.8 2.4 125.4 37.8 16.5 25.8 56.5 -20.1 34.2 -52.3 15.8 50th 52.8 58.3 -26.7 112.8 17.7 11.2 4.9 25.6 -36.0 34.6 -60.8 3.5 40th 19.8 69.7 -79.6 103.3 4.7 15.1 -7.3 21.6 -55.2 31.5 -75.4 -18.9 30th -3.8 89.6 -140.2 101.3 -10.0 24.0 -24.7 17.7 -82.0 41.8 -116.7 -35.5 20th -33.3 95.1 -171.3 86.1 -33.8 21.6 47.4 -8.8 -106.9 52.7 -149.4 -48.0 10th -87.2 106.9 -234.8 59.7 -85.6 35.1 109.4 -45.3 -140.3 53.4 -193.3 -86.5 5th -128.1 97.7 -268.5 -4.6 -97.4 35.9 -118.7 -56.0 -155.2 44.3 -198.5 -110.0 IntDEDF 201.8 157.4 172.6

PAGE 144

132 5.3. Methods 5.3.1. Smallest Extreme Value Distribution The sm allest extreme value distribution, identi fied as the best-fit distribution to represent the wetland water levels by the Anderson-Darlin g test (Section 5.2.2), is a form of the extreme value type I distribution, aka the Gumbel distribution (NIST/SEMATECH eHandbook of Statistical Methods 2010). The distribution is based on the minimum extreme value. The smallest extreme value density function f(x) and cumulative distribution function F(x) are defined as: x x xf exp exp 1 ),| ( (5.2) x xF expexp1),| ( (5.3) where [L] is the location parameter and [L] is the scale parameter (> 0 and x is a real number, in this instance x [L] is the wetland surface and sub-surface water level observation (Weisstein 2010a). The location pa rameter is an indication of the central location of the given distributions, and the scal e parameter is an indication of the spread of the water level data around the location parameter. The smallest extreme value distribution density function is typically identified by a long ta il and is skewed to the left (Figure 5.1). The distribution is not bounded, i.e. defined on the entire real axis ; x, therefore, can be used to represen t the wetland relative water levels.

PAGE 145

133 -5 -4 -3 -2 -1 0 1 2 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Data (x)f(x) densitySmallest Extreme Value Distribution Parameters: Location = 0 Scale = 1 Figure 5.1 Smallest extreme valu e distribution, general case. The final variant of the smallest extreme value distribution of interest is the inverse cumulative distribution function: p pF 1lnln),|(1 (5.4) where F -1( p) is a unique relative water level x [L] (quantile) that is coupled with cumulative probability p or F ( x ) from Eq. (5.3) (NIST/SEMATECH e-Handbook of Statistical Methods 2010). The inverse cumulative di stribution function was used to predict relative water levels at the target percentiles for comparison to the associated mean recorded relative water levels presented in Table 5.3.

PAGE 146

134 5.3.2. Smallest Extreme Value Parameter Identification Representative sm allest extreme value distribution location () and scale () parameters were calculated for each of the five wetland cat egories and the regional group (Table 5.1) using the maximum likelihood method (Chow et al. 1988; Maidment 1993; Weisstein 2010b). The method returns the best-fit estim ates of the smallest extreme value distribution parameters for each data set. The smallest extreme value distribution parameters are predicated on the relative water levels within a respective wetland category. As a result, a specific distributi on function, hence forth referred to as the SEV model, was developed for each of the wetland categories and regional group. Additionally, Kolmogorov-Smirnov tests (Ma ssey 1951; Weisstein 2009b), outlined in Chapter 3.4.2, were performed between the SEV model distributions and the corresponding empirical distribution functions (T able 5.1) for each category to verify the SEV model robustness.

PAGE 147

135 5.3.3. SEV Model Evaluation 5.3.3.1. Wetland Category Analyses The SEV m odels were evaluated by comparing the predicted relative water levels to the mean relative water levels for each wetland cat egory (Table 5.3) to determine how well the smallest extreme value distribution repr esented the category as a whole. The SEV models were evaluated using a root-mean-square d-error (RMSE) analysis and an absolute water level error analysis. The root-mean-squared-error analysis is: %100 )()( 11 2 EDF k p pEDF pSEV EDF CATIntD xx k IntD RMSE RMSE (5.5) p = {0.05,0.10:0.10:0.90,0.95} where CATRMSE is the normalized RMSE, EDFIntD [L] is the interdecile range of the mean relative water level distribution fo r a specific wetland category (Table 5.3), p is a target percentile, k is the total number of percentiles used to evaluate the distributions, SEVx [L] is the SEV model predicted relative water level (quantile) at percentile p, and EDFx [L] is the recorded mean relative water level at percentile p. The absolute water level error (CATAE ) for the category comparison is: %100* )()( 11 k p EDF pEDF pSEV CATIntD xx ABS k AE ; (5.6) p = {0.05,0.10:0.10:0.90,0.95}

PAGE 148

136 where ABS is the absolute value. Further a relative water level error (PRE) analysis was conducted to determine if the SEV models under or over predicted the respective mean recorded relative water levels at each of the target percentiles. The relati ve water level error was calculated as: %100* )()( p EDF pEDF pSEV PR xx RE ; (5.7) p = {0.05,0.10:0.10:0.90,0.95} where EDFR [L] is the range of relative water levels at each target percentile (Table 5.3). 5.3.3.2. Individual Wetland Comparisons To this poin t the SEV model comparison analyses have focused on the mean relative water levels of each wetland category. Th e final analysis technique compares the SEV model distributions to the individual wetland distributions in each category in order to gain understanding into the pr edictive capabilities of the SEV model on an individual wetland basis. The analysis was conducte d using a root-mean-squared-error analysis defined by: n w p wEDF SEV Pxx n RMSE1 2)()( 1; (5.8) p = {0.05,0.10:0.10:0.90,0.95}

PAGE 149

137 where PRMSE [L] is the root-mean-squared-error at target percentile p, w is the individual wetland identifier, n is the total number of wetlands in the category, SEVx [L] is the SEV model predicted relative water level (quantile) at percentile p, and EDFx [L] is the recorded relative water level for wetland w at target percentile p. 5.4. Results and Discussion 5.4.1. Smallest Extreme Value Parameter Identification Specific smallest ex treme value distribution function parameters were developed for each wetland category and the west-central Florid a regional wetland grouping (Region). The maximum likelihood estimates for the wetland category location parameter () and scale parameter () [Eq. (5.2)] are presented in Table 5.4. The location parameters range from -13.4 cm for the wet prairie wetlands to 73.9 cm for the cypress-marsh wetlands. The scale parameters range from 46.0 cm for the hardwood wetlands to 83.9 cm for the cypress-marsh wetlands. In addition, the va riation in the shape and scale parameters, represented by the standard devia tion (StD), is listed in Table 5.4. The standard deviation was calculated from individual wetland di stribution parameters (i.e. 36 location parameters for the cypress category) for the respective wetland category and the regional group. The standard deviation for the location parameters range from 11.0 cm (Hardwood wetlands) to 60.0 cm (Cypress-Mars h wetlands), and the standard deviation for the scale parameters range from 17.3 cm (Cypress wetlands) to 26.5 cm (CypressMarsh wetlands).

PAGE 150

138 Table 5.4. SEV distribution function parame ters and distribution fit test results. Location (cm) Scale (cm) Kolmogorov-Smirnov test Category StD StD h p Dstat Region 33.3 35.6 65.2 19.3 0 0.58 0.09 Cypress 24.5 27.1 51.2 17.3 0 0.37 0.09 Marsh 48.8 38.1 67.5 20.9 0 0.34 0.10 Cypress-Marsh 73.9 60.0 83.9 26.5 0 0.94 0.06 Hardwood 36.6 11.0 46.0 17.4 0 0.97 0.06 Wet Prairie -13.4 22.5 58.4 20.2 0 1.00 0.05 The best-fit SEV models for the regional wetland group as well as the five wetlands categories are shown in Figures 5.2, 5.3 and 5.4. The best-fit smallest extreme value density functions [Eq. (5.2)], defined by the specific location and scale parameters, for the respective categories are overl aid on the water-level histograms (Charts A, C, E, G, I and K). Note the skewed nature of the empirical histograms and the best-fit density functions. This lends credence to the use of the smallest extreme value probability density function to represent the wetla nd water-level data (Section 5.3.1). Another observation is the noticeable under prediction of the water levels at the distribution mode, e.g. Figure 5.2 Chart A. This devia tion may be misleading due to the histogram bin sizes. Bin sizes can be set several different ways, each affecting the amount of data represented in the bin, whic h will distribute the da ta in the histogram accordingly. Also, by definition both the hi stogram bin densities and the area under the density function curve have to add up to one. Therefore, the density function is a more consistent representation of the water-level data.

PAGE 151

139 In addition, the best-fit cumulative distributi on functions [Eq. (5.3)] are overlaid on the respective empirical di stribution functions for the region al wetlands (Chart B) and the five wetland categories (Charts D, F, H, J and L) on Figures 5.2, 5.3 and 5.4. By inspection it is apparent the best-fit SEV models match the regional group and cypress category empirical distributions, the two categ ories with the largest populations (Table 5.1). Similar results were observed for the other wetland categories, with the largest apparent deviation observed for the ma rsh wetlands. Further, all of the SEV model distributions passed the Kolmogorov-Smirnov tests, h = 0 (Table 5.4). This verifies the SEV model shape and scale parameters effectively reproduce the water levels for each wetland category.

PAGE 152

140 -500 -400 -300 -200 -100 0 100 200 0 0.002 0.004 0.006 0.008 0.01 Relative Water Level (cm) Chart A Regional Wetlands Density, f(x) Histogram SEV (33.3cm, 65.2cm) -500 -400 -300 -200 -100 0 100 200 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Relative Water Level (cm) Chart B Regional Wetlands Cumulative probability, F(x) EDF SEV (33.3cm,65.2cm) -500 -400 -300 -200 -100 0 100 200 0 0.002 0.004 0.006 0.008 0.01 Relative Water Level (cm) Chart C Cypress Wetlands Density, f(x) Histogram SEV (24.5cm, 51.2cm) -500 -400 -300 -200 -100 0 100 200 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Relative Water Level (cm) Chart D Cypress Wetlands Cumulative probability, F(x) EDF SEV (24.5cm, 51.2cm) Figure 5.2 SEV Model best-fit distributions, regional group and cypress wetlands.

PAGE 153

141 -200 -150 -100 -50 0 50 100 150 200 0 2 4 6 8 x 10-3 Relative Water Level (cm) Chart E Marsh Wetlands Density, f(x) Histogram SEV (48.8cm, 67.5cm) -200 -150 -100 -50 0 50 100 150 200 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Relative Water Level (cm) Chart F Marsh Wetlands Cumulative probability, F(X) EDF SEV (48.8cm, 67.5cm) -300 -200 -100 0 100 200 0 1 2 3 4 5 x 10-3 Relative Water Level (cm) Chart G Cypress-Marsh Wetlands Density, f(x) Histogram SEV (73.9cm, 83.9cm) -400 -300 -200 -100 0 100 200 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Relative Water Level (cm) Chart H Cypress-Marsh Wetlands Cumulative probability, F(x) EDF SEV (73.9cm, 83.9cm) Figure 5.3 SEV Model best-fit distributions, mars h and cypress-marsh wetlands.

PAGE 154

142 -100 -50 0 50 100 0 0.002 0.004 0.006 0.008 0.01 0.012 Relative Water Level (cm) Chart I Hardwood Wetlands Density, f(x) Histogram SEV (36.6cm, 46.0cm) -100 -50 0 50 100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Relative Water Level (cm) Chart J Hardwood Wetlands Cumulative probability, F(x) EDF SEV (36.6cm, 46.0cm) -200 -150 -100 -50 0 50 100 0 1 2 3 4 5 6 x 10-3 Relative Water Level (cm) Chart K Wet Prairie Wetlands Density, f(x) Histogram SEV (-13.4cm, 58.4cm) -200 -150 -100 -50 0 50 100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Relative Water Level (cm) Chart L Wet Prairie Wetlands Cumulative probability, F(x) EDF SEV (-13.4cm, 58.4cm) Figure 5.4 SEV Model best-fit distributions, ha rdwood and wet prairie wetlands.

PAGE 155

143 5.4.2. Smallest Extreme Value ( SEV ) Distribution Models 5.4.2.1. SEV Model W ater Level Predictions Water levels were calculated for each wetland category using the smallest extreme value inverse cumulative distributi on function [Eq. (5.4)] and th e respective location and shape parameters listed in Table 5.4. The water levels were calculated at the 11 target percentiles. The predicted wate r levels (SEV) for each wetland category are listed next to the recorded or actual relative water levels (M ean) at 11 target percentiles (Table 5.5). For example, the predicted water levels fo r the cypress wetlands range from 127.5 cm below the wetland dry bed elevation at the 5th percentile to 80.7 cm above the wetland dry bed elevation at the 95th percentile. The corresponding recorded water levels are: 119.2 cm (5th) and 56.7 cm (95th) respectively.

PAGE 156

144 Table 5.5. SEV Model predicted water leve ls and category evaluation. Region (cm) Cypress (cm) Marsh (cm) Percentile Mean SEV REP Mean SEV REP Mean SEV REP 95th 69.4 104.9 17.8% 56.7 80.7 25.3% 95.0 122.9 19.0% 90th 62.8 87.7 12.7% 52.1 67.2 14.4% 84.1 105.2 16.3% 80th 50.7 64.4 7.3% 42.6 48.9 5.0% 67.5 81.0 11.0% 70th 41.3 45.4 2.1% 34.6 34.0 -0.4% 56.0 61.4 4.4% 60th 29.4 27.6 -1.0% 24.3 20.1 -2.7% 41.4 42.9 1.3% 50th 16.5 9.4 -3.2% 13.2 5.8 -3.9% 26.9 24.1 -2.2% 40th -0.7 -10.5 -3.7% -2.2 -9.8 -3.2% 10.2 3.5 -5.1% 30th -22.1 -33.9 -4.0% -23.1 -28.2 -1.9% -12.1 -20.8 -8.9% 20th -50.9 -64.5 -4.7% -53.0 -52.2 0.3% -39.4 -52.5 -10.8% 10th -91.6 -113.4 -7.3% -92.2 -90.6 0.5% -77.2 -103.1 -16.6% 5th -117.3 -160.3 -13.8% -119.2 -127.5 -2.7% -97.8 -151.7 -29.8% RMSECAT 13.6% 6.9% 13.6% AECAT 7.1% 5.5% 11.4% Cypress-Marsh (cm) Hardwood (cm) Wet Prairie (cm) Percentile Mean SEV REP Mean SEV REP Mean SEV REP 95th 121.8 165.9 27.8% 76.9 87.0 17.7% 49.7 50.7 2.8% 90th 114.6 143.9 19.9% 71.7 74.9 5.8% 32.4 35.3 10.0% 80th 97.7 113.8 12.4% 58.6 58.4 -0.3% 10.9 14.4 8.6% 70th 85.9 89.4 2.7% 50.4 45.1 -15.1% -6.2 -2.6 6.9% 60th 69.2 66.5 -2.2% 37.8 32.5 -16.9% -20.1 -18.5 2.4% 50th 52.8 43.1 -6.9% 17.7 19.7 9.9% -36.0 -34.8 1.8% 40th 19.8 17.5 -1.3% 4.7 5.7 3.5% -55.2 -52.6 4.5% 30th -3.8 -12.7 -3.7% -10.0 -10.8 -2.0% -82.0 -73.6 10.3% 20th -33.3 -52.0 -7.3% -33.8 -32.4 3.7% -106.9 -101.0 5.8% 10th -87.2 -115.0 -9.4% -85.6 -66.8 29.3% -140.3 -144.8 -4.3% 5th -128.1 -175.4 -17.9% -97.4 -99.9 -4.0% -155.2 -186.9 -35.8% RMSECAT 12.2% 4.4% 6.0% AECAT 10.1% 9.8% 8.5%

PAGE 157

145 The predicted water levels ( SEV model) and the correspond ing mean recorded water levels for each wetland category are presented on inverse quantile plots on Figure 5.3. Figure 5.3 is comprised of six charts each representing a wetland category. The SEV models are portrayed as a solid line because it is a continuous distribution, and the mean recorded water levels, empirical data, are re presented as single points at each target probability level. Additionally, each chart shows one positive and one negative standard deviation about the mean recorded water le vels, as well as the minimum and maximum water levels corresponding to the target probabi lity. The last object on each chart is the SEV model error bars. The bars represent th e model errors associated with individual wetland water level predictions at each target percentile.

PAGE 158

146 Regional Wetlands (56 ea. 3,980 records) -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 00.10.20.30.40.50.60.70.80.91 PercentilesRelative Water Levels (cm ) SEV Model EDF Min Max StD+ StDCypress Wetlands (36 ea. 2,525 records) -300 -250 -200 -150 -100 -50 0 50 100 150 00.10.20.30.40.50.60.70.80.91 PercentilesRelative Water Levels (cm ) SEV Model EDF Min Max StD+ StDFigure 5.5 SEV Model predicted water levels and recorded water levels for the regional and cypress wetlands.

PAGE 159

147 Marsh Wetlands (9 ea. 663 records) -250 -200 -150 -100 -50 0 50 100 150 200 00.10.20.30.40.50.60.70.80.91 PercentilesRelative Water Levels (cm ) SEV Model EDF Min Max StD+ StDCypress-Marsh Wetlands (5 ea. 353 records) -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 00.10.20.30.40.50.60.70.80.91 PercentilesRelative Water Levels (cm ) SEV Model EDF Min Max StD+ StDFigure 5.6 SEV Model predicted water levels and recorded water levels for the marsh and cypress-marsh wetlands.

PAGE 160

148 Hardwood Wetlands (3 ea. 216 records) -150 -100 -50 0 50 100 150 00.10.20.30.40.50.60.70.80.91 PercentilesRelative Water Levels (cm ) SEV Model EDF Min Max StD+ StDWet Prairie Wetlands (3 ea. 224 records) -250 -200 -150 -100 -50 0 50 100 00.10.20.30.40.50.60.70.80.91 PercentilesRelative Water Levels (cm ) SEV Model EDF Min Max StD+ StDFigure 5.7 SEV Model predicted water levels and recorded water levels for the hardwood and wet prairie wetlands.

PAGE 161

149 5.4.2.2. SEV Model Evaluation 5.4.2.2.1. Wetland Category Evaluation The sm allest extreme value distribution func tions were first evaluated by comparing the predicted water levels [Eq. (5.4)] to the mean recorded water levels for each wetland category. The root-mean-squared-errors for th e category analyses [Eq. (5.6)] range from 4.4% for the hardwood wetlands to 13.6% for the regional and mars h wetlands (Table 5.5 CATRMSE ). Further, the absolute water level e rror [Eq. (5.7)] ranges from 5.5% for the cypress wetlands to 11.4% for the marsh wetlands (Table 5.5 CATAE ). Additionally, water level error analyses [Eq. (5.7)] compared the predicted water levels to the mean recorded water levels at ea ch target percentile (Table 5.6 PRE). Overall, for all the wetland categories, the analyses show the SEV model under predicted the mean water levels at the 5th thru the 60th percentiles (i.e. produced lo wer water levels) and over predicted the mean water levels at the 70th thru 95th percentiles (i.e. predicted higher water levels). 5.4.2.2.2. Individual Wetland Comparisons The final an alyses investigated the capability of the respective SEV model to predict the recorded water levels for each individual we tland within a category. The root-meansquared-error ( RMSEP) [Eq. (5.8)] for the regional wetland group range from 36.1 cm at the 70th percentile to 74.4 cm at the 5th percentile (Table 5.6). Further, the RMSE associated with each percentile are illust rated on the respective wetland category chart as error bars about the respective SEV model continuous distribution (Figures 5.5 thru 5.7).

PAGE 162

150 The length of each error bar is twice the RMSEP value listed in Table 5.6 centered about the SEV model predicted water level. Table 5.6. SEV model prediction RMSE per percentile ( RMSEP). RMSE P (cm) Percentile Region Cypress Marsh CypressMarsh Hardwood Wet Prairie 95th 52.4 33.0 54.7 71.1 25.9 16.2 90th 44.5 27.3 48.6 60.2 22.8 12.7 80th 37.8 25.0 43.6 48.2 16.8 17.5 70th 36.1 25.6 39.9 48.4 16.9 22.0 60th 36.6 29.4 36.6 48.2 14.4 27.9 50th 40.3 35.7 36.0 53.1 9.4 28.2 40th 45.3 42.0 39.9 62.4 12.4 25.8 30th 51.9 47.7 36.6 80.6 19.6 35.2 20th 56.4 51.2 45.9 87.1 17.7 43.4 10th 61.6 51.9 59.7 99.6 34.3 43.8 5th 74.4 58.3 80.8 99.4 29.4 48.1 5.4.2.3. Discussion SE V Model Performance The SEV models reproduced the mean water levels for each wetland category and the regional group adequately when considering the overall error and visual fit. The predicted water levels for each category are within the recorded empirical distribution deviation range at each of the target percentiles (Figures 5.5 thru 5.7). However, upon closer inspection, the SEV models do not predict the water levels at the distribution tails as good. This is confirmed by the relative error (REP) and root-mean-squared-error ( RMSEP) calculated at each target percentile (Tables 5.5 and 5.6 respectively). In general, the SEV models under predicted the low water levels (5th percentile) and over predicted the high water levels (95th percentile). The reduced predictive capabilities of the SEV models at the tails might be due to the limited amount of data recorded at the

PAGE 163

151 extreme water elevations. Also, the best-fit distributions are fitted to the bulk of the water-level data, which is centered on the location parameter values for the representative SEV model away from the tails. Furthermore, the largest prediction errors were generally observed at the 5th and 10th percentiles. This could be attributed to the high variability in the deep wetland water levels discussed in Chapter Three. The SEV models can be applied to specific we tland types where representative water level data is not available. Fu rther, the distributions can be used as a calibration tool to indicate whether a hydrologic model is portraying the respective wetland or wetland category water levels adequately. For ex ample, an extended period simulation model should produce cypress wetland water levels th at form a probability distribution curve similar to the one on Figure 5.5 Cypress Wetlands. 5.4.3. SEV Models Probability Plots Statistical com parisons of the five wetla nd categories and four regional groups of wetlands located in different ar eas throughout the region were pr esented in Chapter 3.5.2. A battery of Kolmogorov-Smirnov tests discu ssed in Chapter 3.4.2 indicated observed water levels for the same period were unique. It was conjectured that SEV models could be developed to verify and fu rther explore their uniqueness. SEV models representing the various wetland categories and wetland groups were plotted on smallest extreme value probability scale (Figures 5.8 and 5.9). Probabi lity plots are used to portray distributions in a linear manner (bmxy ) so that one can easily interpolate, extrapolate, or compare the data in a simpler manner (C how et al. 1988). This is accomplished by

PAGE 164

152 rearranging the inverse smallest extreme valu e distribution function [Eq. (5.4)] into a linearized form: xp 1 1ln ln (5.9) where x is the relative water level (quantile) [L] and p 1lnln is the corresponding scaled cumulative probability, F ( x ) in Eq. (5.3). -200 -150 -100 -50 0 50 100 150 200 0.05 0.1 0.25 0.5 0.75 0.9 0.95 Relative Water Levels (cm)SEV ProbabilityWetland Categories Cypress Marsh CypMarsh Hardwood Wet Prairie Figure 5.8 SEV Model probability plots for all wetland categories.

PAGE 165

153 -250 -200 -150 -100 -50 0 50 100 150 0.05 0.1 0.25 0.5 0.75 0.9 0.95 Relative Water Levels (cm)SEV ProbabilityWetland Groups GS MB S UHFDA Figure 5.9 SEV Model probability plots for the wetland groups. Based on the observation of the linearized proba bility plots, each distribution exhibits similar means, extremes and ranges in water-le vel behavior, however, has a unique slope. The slopes of the fitted distributions are defined by 1 [L-1]. The slopes of the wetland categories ranged from 0.0119 cm-1 (cypress-marsh) to 0.0217 cm-1 (hardwood), and the slopes of the wetland groups ranged from range from 0.014 cm-1 (UHFDA) to 0.026 cm-1 (Green Swamp). The use of scal ed probabilities to display the SEV model distributions clearly shows there are hydrologic differences between the individua l wetland categories and the respective wetland groups. Further, the variability in water levels asso ciated with each wetland category or regional group is related to the slope of the respective probability curv e (Figures 5.8 and 5.9). For instance, the steeper the SEV model curve (higher slope) the lower the variability in water

PAGE 166

154 levels. Conversely, high variability in water levels corresponds to a lower slope or a flat curve. As an example, the hardwood wetlands have the highest slope (steepest curve) of the five wetland categories, suggesting low water-level variability. This is confirmed on the SEV model comparison chart (Figur e 5.7 hardwood wetlands). As seen on the chart and from the corresponding data in Table 5.3, the hardwood wetland category generally has the tightest standard deviation, and minimum and maximum water-level range at each of the percentiles. Also, the SEV model errors are generally small for this category (Table 5.6). This could be indicative of a more consistent water-table depth, which makes ecological sense since hardwood species generally have a lower to lerance to extreme water-level variability (Mitsch and Ewel 1979). Furthermore, the reduced variability in the water levels could be due to the limited water-level data available for the hardwood category. 5.4.4. Theoretical Application The sm allest extreme value probability models can be used in conjunc tion with current or projected meteorological data to determine representative water levels for wetlands. For example, extended period hydrologic simulati ons need to account for changing climate conditions, precipitation patterns in particular The changing precipitation patterns will affect the antecedent moisture conditions w ithin a region or hydrologic study area. The moisture conditions can be above or below l ong-term averages, and at varying departures from the norm, i.e. 20% below or 40% above normal conditions.

PAGE 167

155 The normal moisture conditions and departures can be transferred to the smallest extreme value inverse cumulative distribution functions [Eq. (5.4)] to predict a representative water level for a particular wetland, wetla nd category or group of wetlands within a hydrologic study area. For instance, moisture conditions projected to be 50% above normal, will equate to the 75th percentile for the SEV Model (50% above the median percentile). Incorporating this percentile into the specific inverse cu mulative distribution functions will yield representative wetland water levels. Further, the predicted water levels can be adjusted using the SEV model error bars or any of the other boundary parameters on the respective chart on Figures 5.5, 5.6 and 5.7. The adjustments can be made based on modeling experience and knowledge of the hydrogeology. 5.5. Conclusions The sm allest extreme value ( SEV ) probability distribution was identified as the best-fit model to represent the wate r levels of five wetland cat egories and a region group comprised of 56 wetlands located in the northern Tampa Bay region. SEV models were developed to represent water levels of vari ous wetland types in the absence of recorded hydrologic data or where an analytical represen tation is desired. The probability models were shown to adequately represent the rela tive water levels associated with the wetland categories as well the regional wetland group. The predicted water levels were usually close to the mean recorded water levels for a given category, generally falling within one standard deviation of the mean recorded wa ter levels. On average, the discrepancy between the predicted water levels and the recorded water levels was less than 10%, shown by a root-mean-squarederror analysis. Further observations showed the SEV

PAGE 168

156 models under predicted the wetland water levels at the low probability levels and over predicted the water levels at the high probability levels. In addition, smallest extreme value dens ity functions representing the five wetland categories as well as four groups of wetlands spread across the region were linearized and compared on special probability axes. The slopes of the linear distributions indicated differences between the individual wetland cat egories, and between the wetland groups. This result suggests there are distinct hydrologic differences between the various wetland categories and wetland groups in west-central Fl orida. Further, this indicates wetlands subjected to similar hydrologic st resses behave similarly. Overall, the smallest extreme value probability models can be used to represent water levels of various wetland types in the abse nce of recorded hydrologic data. With the information in this chapter water resour ce engineers and hydrologists can develop representative water-level characteristics of wetlands with quantifiable errors. This is useful in developing accurate hydrologic mode ls to be used in hydrologic studies with large numbers of wetlands and limited water level data.

PAGE 169

157 CHAPTER 6 SUMMARY AND CONCLUSIONS Hydrologic data including water-leve l observations, stage/storage, and surface/groundwater interactions do not generall y exist for the vast majority of wetlands within west-central Florida let alone other less studied regions throughout the United States and the world. Therefore, this disse rtation presents several improved analytical and empirical methods designed to provide a better means to quantify the above ground storage of wetlands, and characterize the wate r level fluctuations as well as the surface and groundwater interactions associated with wetlands. Firs t, an analytical method was developed to describe the stor age characteristics of wetlands and lakes in the absence of detailed hydrologic and bathymetric data. S econd, an empirical probabilistic approach was developed to characterize the water levels associated with isolated wetlands, and to provide insights into surface and groundwater interactions within and adjacent to the wetlands. Third, analytical probability mode ls were developed to represent the water levels of wetlands in the absence of detail ed hydrologic data. Th e end product is to improve the accuracy of hydrologic model predic ted water levels and fluctuations within wetlands, and associated surface and groundw ater exchange between the wetland and local surficial aquifer system.

PAGE 170

158 In Chapter Two wetland and lake stage-storag e relationships were defined using a powerfunction model that is based on a single fitting parameter and two physically-based parameters: the reference wetland or lake plan ar area (e.g. a GIS data coverage), and the corresponding maximum pool depth. General models were developed based on detailed bathymetry of wetlands and lakes located in west-central Florida, North Dakota and Canada, representing different geologic settings. These models were then used to predict the storage behavior of multiple wetland and lake combinations. To determine the strength of the general mode l in the absence of detailed survey data, the model was applied to an independent validation data set comprised of 21 lakes in west-central Florida. This work demonstrated that a single wetland shape parameter can be used to represent the storage of a single or multiple wetlands and/or lakes with acceptable and quantifiable error in field, theoretical and modeling st udies. Additionally, the power-function model shape parameter(s) could be used as a calibration tool in hydrologic models, as opposed to individually adjusting rating relationship terms thereby easing calibration difficulty and reducing over parameterization. Chapter Three focused on the hydrologic charact erization of 56 various isolated wetlands in west-central Florida. Empirical distribu tion functions or freque ncy distributions were developed from historic al paired wetland and upland wa ter elevation records collected over seven years. The empirical distributi ons provided a means to analyze the waterlevel data using frequencies a nd probabilities of occurrence of water levels over time.

PAGE 171

159 Further, the distributions were used to compare the water-level fluctuations of five wetland categories (cypress, marsh, cypressmarsh, hardwood and wet prairie), and to identify potentially impacted wetlands. In general, at least some standing water was present in these wetlands 62% of the time over the seven year study. Also, the water levels in the wetlands exceeded the normal pool vegetative markers only 4% of the time. These crucial parameters can be used as a calibration tool to ensure hydrologic models accurately represent water levels in wetlands, as well as means to determine indicative behavior for normal or impaired wetland hydroperiods. Variability in water levels between the wetla nds in the west-centr al Florida region was significant. Consequently, individual wetland categories could not be identified via simple inspection of the respective water-le vel distributions. Additionally, there was higher variability in the gr oundwater levels beneath the wetlands than in the surfacewater levels within the wetlands. The high variability in the groundwater levels is most likely a reflection of varying water-table dept h across the west-central Florida region. The depth of the water table can be a ffected by pumping stresses, surface water augmentation or local hydrogeology. Water leve l variability within the wetlands near the wetland extents (maximum pool de pth) was lower due in part to the natural shape of the wetlands. The incremental rise or fall of th e wetland surface water levels will be small in comparison to the increase or decrease of the wetland pool volume near the wetland extents effectively stabilizing the surf ace water levels within the wetlands.

PAGE 172

160 Frequency distributions can be used as a comparison tool to identify similarities and differences between representative data sets and to identify atypical hydrologic behavior in wetlands. Statistical tests performed on frequency distributions representing the combined water levels within a wetland categ ory showed significant differences in the water-level behavior for the specific wetland categories. Further, wetlands that may be adversely influenced by anthropogenic activitie s or natural stresses were identified using a simple technique comparing a respective wetland empirical distribution to a general trend distribution curve developed in this work. These are examples that show probability distributions can be used in hydr ologic modeling to test the water-level behavior, trends, or stresses to indivi dual wetlands and wetland categories. Chapter Four provided insight into the groundw ater recharge/discharge characteristics between 56 isolated wetlands and surrounding uplands. The analysis was performed to test the assumption that wetlands are local wa ter-table depressions. The results indicated these wetlands were groundwater recharge zones 59% of the time over the seven year study. This was based on the head difference between the paired wetland and upland well water elevations. Additional seasonal analyses, utilizing water elevation da ta from the peak dry season (March May) and the peak wet season (Jul y September), indica ted these wetlands were groundwater recharge zones 61% of th e time during the dry season and 47% of the time during the wet season. Further, statistical tests comparing the seasonal wetland water levels and upland groundwater levels indicated these wetlands are surface-water

PAGE 173

161 expressions of the local groundwater system during the wet season, and indicated the wetland and upland water levels disassociate during the dry season as the local water table drops. Hydrologic modele rs should be aware of these findings to ensure regional models accurately represent water-table fluctuations. The aim of Chapter Five was to identify specifi c probability models th at can be used to represent the surface and subsur face water-level behavior of various wetland types. The application of these probability models woul d be to predict wetland range of water-level fluctuations especially during different seas onal conditions, in the absence of recorded hydrologic data. Furthermore, the models we re used to discern hydrologic differences between the various wetland categories and four groups of wetlands located in different hydrogeologic settings. The smallest extreme value probability distribut ion was identified as the best-fit model to represent the water levels associated with th e five wetland categories as well as a regional group comprised of all 56 wetlands. Specifi c distributions, predicated on respective location and scale parameters, were used to pr edict the water levels associated with each wetland category. Overall, th e discrepancy between the pred icted water levels and the recorded water levels was less than 10%. Additional observations showed the smallest extreme value models under predicted the wetl and water levels at the low probability levels, and over predicted the relative water levels at the high probability levels.

PAGE 174

162 In addition, smallest extreme value probability models representing the five wetland categories as well as four groups of wetlands spread across the region were linearized on probability scale. Based on the probability models, the various wetland categories exhibited similar means, extremes and ranges in water-level behavi or, but unique slopes in frequency distributions. The slopes of the linear distributions indicated distinct hydrologic differences between the individual we tland categories, as we ll as between the wetland groups. The result suggests there ar e different hydrologic pr operties associated with the various wetland categories in west -central Florida, and indicate wetlands subjected to similar hydrol ogic stresses or conditions behave similarly. Water resource engineers and hydrologists ca n utilize the represen tative wetland and/or wetland category characteristics presented in this work to develop comprehensive and more accurate hydrologic models. The accuracy of hydrologic models is predicated on sound and complete input parameters such as the surface and sub-surface water storage associated with wetlands, which are often estimated due to the lack of detailed bathymetry and water-level data. However, now these parameters can be estimated with quantifiable error using the methods and an alytical techniques presented in this dissertation. The storage model in conjunction with the fr equency analysis and probability models will improve the accuracy of wetland representati on in hydrologic models. The methods and techniques can be utilized to define wetland water-level and storage characteristics derived from various anthropogenic and clim atic stresses. Further, they will aid

PAGE 175

163 engineers and hydrologists in predicting surface water runoff, river stage and discharge, and groundwater fluctuations. Overall, wetland water-level fluctuations were characterized and coupled with improved analytical methods geared toward modeling the storage and water-level behavior in wetlands. The techniques and methodologies presented in this dissertation are not solely for the purposes of understanding the hydrology, and especially the surface/groundwater interactions of west-central Florida wetlands. These techniques can be applied to other areas to help understand the hydr ology of these wetlands in va rious geologic and climatic settings.

PAGE 176

164 REFERENCES Altman, D. G., and Bland, J. M. (1994). "Sta tistics Notes: Quartiles, quintiles, centiles, and other quantiles." BMJ 309(6960), 996. Bicknell, B., Imhoff, J. C., Kittle, J. L., Jr., Jobes, T. H., and Donigian, A. D., Jr. (2001). "Hydrological simulation programF ORTRAN (HSPF): Users manual for Version 12." U.S. Environmental Protection Agency, Athens, GA. Bidlake, W. R., Woodham, W. M., and Lopez, M. A. (1996). Evapotranspiration from areas of native vegetation in West-Central Florida U.S. Geological Survey water-supply paper ; 2430. Bradley, C. (2002). "Simulation of the a nnual water table dyna mics of a floodplain wetland, Narborough Bog, UK." Journal of Hydrology 261(1-4), 150. Bras, R. L., and Rodrg uez-Iturbe, I. (1993). Random functions and hydrology Dover Publications, New York. Brooks, R. T., and Hayashi, M. (2002). "Depth-Area-Volume and Hydroperiod Relationships of Ephemeral (Vernal) Fo rest Pools in Southern New England Wetlands 22(2), 247-255. Bullock, A., and Acreman, M. (2003). "The ro le of wetlands in the hydrological cycle." Hydrology and Earth System Sciences 7(3), 358-389. Carr, D. W., and Rochow, T. F. (2004). "C omparison of six biologic indicators of hydrology in isolated Taxodium ascendens domes." Brooksville, Southwest Florida Water Management District, Tec hnical Memorandum, April 19, 2004, 4. Chow, T. V., Maidment, D. R., and Mays, L. W. (1988). Applied Hydrology McGrawHill, New York. Dahl, T. E. (2005). "Florida's Wetlands: An Update on Status and Trends 1985 to 1996." U.S. Department of the Interior; Fish a nd Wildlife Service, Washington, D.C., 80 pp. Dahl, T. E. (2006). "Status and trends of wetlands in the conterminous United States 1998 to 2004." U.S. Department of the In terior; Fish and Wildlife Service, Washington, D.C., 112 pp.

PAGE 177

165 Dallal, G. E. (2007). "No nparametric Statistics." http://www.jerrydallal.com/LHSP/npar.htm (Jun. 2008). Environmental Systems Research Institute In c. (ESRI). (2007). ArcMap 9.2, Redlands, Calif. Florida Center for Community Design and Research. (2007). "Hillsborough County Watershed Atlas ." http://www.hillsborough.watera tlas.usf.edu/ (Jun. 06, 2007). Freeze, R. A., and Cherry, J. A. (1979). Groundwater, Prentice-Hall, Englewood Cliffs, N.J. Haag, K. H., Lee, T. M., and Herndon, D. C. (2005). "Bathymetry and Vegetation in Isolated Marsh and Cypress Wetlands in the Northern Tampa Bay Area, 20002004." U.S. Geological Survey Scientific Investigations Report 2005-5109, 49. Hammersmark, C. T., Solomon, Z. D., Mark, C. R., and Jeffrey, F. M. (2009). "Simulated Effects of Stream Restoration on the Di stribution of Wet-Meadow Vegetation." Restoration Ecology Hayashi, M., and van der Kamp, G. (2000). "S imple equations to represent the volumearea-depth relations of shallow wetla nds in small topographic depressions." Journal of Hydrology 237(1-2), 74-85. Hayashi, M., van der Kamp, G., and Rudolph, D. L. (1998). "Water and solute transfer between a prairie wetland and adjacent uplands, 1. Water balance." Journal of Hydrology 207(1-2), 42. Hill, A. J., and Neary, V. S. (2007). "Es timating Evapotranspiration and Seepage for a Sinkhole Wetland From Diurnal Surface-Water Cycles." JAWRA Journal of the American Water Resources Association 43(6), 1373-1382. Hogg, R. V., and Ledolter, J. (1987). Engineering statistics Macmillan, New York. Johnson, A. I. (1967). Specific yield : compilation of spec ific yields for various materials U.S. G.P.O., Washington, D.C. Johnson, W. C., Boettcher, S. E., Poiani, K. A., and Guntenspergen, G. (2004). "Influence of Weather Extremes on the Water Levels of Glaciated Prairie Wetlands." Wetlands 24(2), 385-398. Lee, T. M., Haag, K. H., Metz, P. A., and Sacks, L. A. (2009). "Comparative Hydrology, Water Quality, and Ecology of Selected Natural and Augmented Freshwater Wetlands in West-Central Florida." U.S. Geological Survey Professional Paper 1758, 152.

PAGE 178

166 Lee, T. M., and Swancar, A. (1997). "Inf luence of evaporation, ground water, and uncertainty in the hydrologic budget of Lake Lucerne, a seepage lake in Polk County, Florida." U.S. Geological Survey water-supply paper ; 2439, 61. Maidment, D. R. (1993). Handbook of hydrology McGraw-Hill, New York. Massey, F. J., Jr. (1951). "The Kolmogorov-Smirnov Test for Goodness of Fit." Journal of the American Statistical Association 46(253), 68-78. Mitsch, W. J., and Ewel, K. C. (1979). "Comparative Biomass and Growth of Cypress in Florida Wetlands." American Midland Naturalist 101(2), 417. Mitsch, W. J., and Gosselink, J. G. (2000). Wetlands, John Wiley, New York. National Oceanic and Atmospheric Admini stration. (2009). "Annual Climatological Summary." http://cdo.ncdc.noaa.gov/ancsum/ACS (Dec. 2009). Nestler, J. M., and Long, K. S. (1997). "D evelopment of hydrologi cal indices to aid cumulative impact analysis of riverine wetlands." Regulated Rivers: Research & Management 13(4), 317-334. Nilsson, K. A., Ross, M. A., and Trout, K. E. (2008). "Analytic method to derive wetland stage-storage relationships using GIS areas." Journal of Hydrologic Engineering 13(4), 278-282. Nilsson, K. A., Trout, K. E., and Ross, M. A. (In Press). "A General Model to Represent Multiple Wetland and Lake Stage-Storage Behavior." Journal of Hydrologic Engineering-ASCE. NIST/SEMATECH e-Handbook of Statistical Methods (2010). "Extreme Value Type I Distribution." http://www.itl.nist.gov/div898/handbook/eda/section3/eda366g.htm (Feb. 2010). O'Connor, D. J. (1989). "Seasonal and Long-Te rm Variations of Dissolved Solids in Lakes and Reservoirs." Journal of Environmen tal Engineering-ASCE 115(6), 1213-1234. Rains, M. C., Mount, J. F., and Larsen, E. W. (2004). "Simulated Changes in Shallow Groundwater and Vegetation Distributions Under Different Reservoir Operations Scenarios." Ecological Applications 14(1), 192-207. Rosenberry, D. O., and Winter, T. C. (1997). "D ynamics of water-table fluctuations in an upland between two prairie-pot hole wetlands in North Dakota." Journal of Hydrology 191(1-4), 266-289. Shjeflo, J. B. (1968). "Evapotranspiration a nd the water budget of prairie potholes in North Dakota." U.S. Geological Survey Professional Paper, 585-B, 47.

PAGE 179

167 Singh, V. P., and Woolhiser, D. A. ( 2002). "Mathematical Modeling of Watershed Hydrology." Journal of Hydrologic Engineering 7(4), 270. Southeast Regional Climate Center. (2010). "H istorical Climate Summaries for Florida." Data for Plant City, Florida 1931-2007, http://www.sercc.com/climateinfo/hist orical/historical_fl.html (Mar. 2010). Southwest Florida Water Management Dist rict. (2007). "Data & Maps, GIS Data." http://www.swfwmd.state.fl.us/data/ (Jun. 2007). Spechler, R. M., and Kroening, S. E. (2007) "Hydrology of Polk County, Florida." U.S. Geological Survey Scientific Inve stigations Report 2006-5320, 114. StatSoft Inc. (2010). "Electronic Statisti cs Textbook." Tulsa, OK: StatSoft. WEB: http://www.statsoft.com/textbook/stathome.html (Dec. 2009). Stephens, M. A. (1974). "EDF Statistics for Goodness of Fit and Some Comparisons." Journal of the American Statistical Association 69(347), 730-737. U.S. Environmental Protection Agency. ( 2009). "Laws and Regulations." Clean Water Act, http://www.epa.gov/regulations /laws/cwa.html (Nov. 24, 2009). U.S. Fish and Wildlife Service. (2007a). "National Wetlands Inventory website. U.S. Department of the Interior, Fish a nd Wildlife Service, Washington, D.C." http://www.fws.gov/wetlands/index.html (Jun. 2007). U.S. Fish and Wildlife Service. (2007b). "Wetlands Geodatabase." Wetlands Digital Data, http://wetlandsfws.er.usgs.gov/NWI/download.html (Jul. 2007). Weisstein, E. W. (2009a). "D istribution Function." From MathWorld -A Wolfram Web Resource. http://mathworld.wolfram.com /DistributionFunction.html (Dec. 2009). Weisstein, E. W. (2009b). "K olmogorov-Smirnov Test." From MathWorld --A Wolfram Web Resource. http://mathworld.wol fram.com/Kolmogorov-SmirnovTest.html (Jun. 2008). Weisstein, E. W. (2009c). "Plotting Position." From Math World -A Wolfram Web Resource. http://mathworld.wolfram.com/PlottingPosition.html (Dec. 2009). Weisstein, E. W. (2010a). "Gumbel Distribution." From MathWorld --A Wolfram Web Resource. http://mathworld.wolfram.com /GumbelDistribution.html (Feb. 2010). Weisstein, E. W. (2010b). "Maximum Likelihood." From MathWorld --A Wolfram Web Resource. http://mathworld.wolfram.c om/MaximumLikelihood.html (Feb. 2010). Whigham, D. F., and Jordan, T. E. (2003). "Isolated Wetlands and Water Quality." Wetlands 23(3), 541-549.

PAGE 180

168 Winter, T. C. (1999). "Relation of streams, lakes, and wetlands to groundwater flow systems." Hydrogeology Journal 7(1), 28. Wise, W. R., Annable, M. D., Walser, J. A. E., Switt, R. S., and Shaw, D. T. (2000). "A wetland-aquifer interaction test." Journal of Hydrology 227(1-4), 257.

PAGE 181

169APPENDICES

PAGE 182

170 Appendix A: Staff Gauge and Wetland Well Data Correlations Spearm an rank correlation coefficient tests (Dallal 2007) were performed comparing the wetland well data set to the a ssociated staff gauge data sets for each wetland in this study to determine if the respective data sets are statis tically similar. The data sets used for this analysis were developed from paired water level measurements that were matched based on the data collection date. Fifty-four paired wells were used in this analysis because two of the staff gauges did not have recorded data. The wetland well and staff gauge time series correlations were performed to valid ate the use of the wetland well data set to characterize the wetland pooled wa ter levels in stead of using the staff gauge data sets. The Spearman rank correlati on tests the hypothesis that there is no correlation ( p = 1) against the alternative there is a nonzero correlation at the 5% confidence interval. The Spearman rank correlation was used for this anal ysis due to non-parametric nature of the well and staff gauge data sets. The average Spearmans rho (correlation coe fficient) for the 54 wetland well and staff gauge comparisons was 0.90, and the corresponding p-value for each wetland pair was 0.0. Based on the results of the Spearman ra nk correlation tests the wetland well data sets were strongly related to the corresponding staff gauge data se ts. The analysis showed the general trend behavior between the staff gauge and wetland well was statistically similar, thus allowing the use of the wetland well to characterize the pooled wetland hydrologic behavior. Also, based on the strong correlation, the staff gauge data was used to fill in the associated wetland well data gaps.

PAGE 183

171 Appendix A: (Continued) Reference: Dallal, G. E. (2007). "Nonparametric Statistics." http://www.jerrydallal.com/LHSP/npar.htm (Jun. 2008).

PAGE 184

172 Appendix B: Wetland Well and Upland Well Data Normality Check A Lilliefors test (Conover 1980) was perform ed on the wetland and upland well data sets to determine if the data were normally distri buted. The Lilliefors test is a two-sided goodness-of-fit test suitable when a fully-specified null distribution is unknown and the respective parameters must be estimated. The well water level distribution type was evaluated to ensure the proper statistical models and methods were used in the analyses, i.e parametric or non-parametric (Dallal 2007 ; StatSoft Inc. 2010). The null hypothesis being that the sample comes from a distribution in the normal family, against the alternative that it does not come from a norma l distribution. The test returns the logical value h = 1 if it rejects the nu ll hypothesis at the 5% significance level, and h = 0 if it cannot. Five of the wetland well data sets passed the Lilliefors normality test, null hypothesis h = 0, and 51 wetland well data sets failed the test h = 1. Two of the well data sets that passed the test had very small p-values (0.17 and 0.07 respec tively) suggesting the test results were very weak, and three of the wetland well data sets had moderate p-values of 0.50 indicating the test results were good. Additionally, the L illiefors test rejected the null hypothesis for 37 of the upland well data sets Ten of the well da ta sets returned a logical value h = 0 with very small p-values, average of 0.12. Nine of the upland well data sets passed the null hypothesis h = 0 with moderate p-values, average of 0.47. The Lilliefors normality test showed that th e majority of the wetland and upland well data sets did not pass the normality tests. Furthermore, sample probability density functions

PAGE 185

173 Appendix B: (Continued) for the paired wetland and upland wells associ ated with wetland 20 clearly showed that the data distributions were skewed to the low end of the water levels (Figure B.1). This could be due to the relatively short time pe riod the data set comprises, or due to the topography of the wetlands. For instance, wetlands typically fan out at the upper pool depths covering more area (Brooks and Hayashi 2002; Hagg et al. 2005; Nilsson et al. 2008). Thus little increases in water depth can equate to large increases in pooled surface area. This would tend to shift the mean of the mass distributions to the higher water levels. Since the majority of the data sets failed the Lilliefors normality test, the statistical analyses performe d in this study were designed to handle non-parametric or distribution-free data. 30 32 34 36 38 40 42 44 0 0.05 0.1 0.15 0.2 0.25 Well Head (NGVD)ProbabilityWetland 20 Well Data Records ww1959 fit uw1792 fit Figure B.1 Typical wetland and upland well probability density functions.

PAGE 186

174 Appendix B: (Continued) References: Brooks, R. T., and Hayashi, M. (2002). "Depth-Area-Volume and Hydroperiod Relationships of Ephemeral (Vernal) Fo rest Pools in Southern New England Wetlands 22(2), 247-255. Conover, W. J. (1980). Practical Nonparametric Statistics Wiley, NJ. Dallal, G. E. (2007). "Nonparametric Statistics." http://www.jerrydallal.com/LHSP/npar.htm (Jun. 2008). Haag, K. H., Lee, T. M., and Herndon, D. C. (2005). "Bathymetry and Vegetation in Isolated Marsh and Cypress Wetlands in the Northern Tampa Bay Area, 20002004." U.S. Geological Survey Scientific Investigations Report 2005-5109, 49. Nilsson, K. A., Ross, M. A., and Trout, K. E. (2008). "Analytic method to derive wetland stage-storage relationships using GIS areas." Journal of Hydrologic Engineering 13(4), 278-282. StatSoft Inc. (2010). "Electronic Statisti cs Textbook." Tulsa, OK: StatSoft. WEB: http://www.statsoft.com/textbook/stathome.html (Dec. 2009).

PAGE 187

175 Appendix C: Monthly Vers us Daily Data Comparison Two-sam ple Kolmogorov-Smirnov tests (KS-te st) were performed to determine if wetland well monthly water level records are statistically different from the corresponding daily water level records. Th e KS-test is a form of minimum distance estimation used to compare the empirical dist ribution functions of two samples, and tries to determine if two datasets differ signifi cantly (Massey 1951; StatSoft Inc. 2010). The KS-test makes no assumption about the data dist ribution, i.e. it is a non-parametric and distribution free test. The test was used due to the non-para metric nature of the water elevation data sets (Appendix B). Th e Kolmogorov-Smirnov statistic quantifies a distance between the empirical distributi on functions of two samples. The null hypothesis is that the two data sets are from the same continuous distribution; else they are from different continuous di stributions. The result of h = 1 is returned if the test rejects the null hypothesis at the 5% significance level, otherwise the result of h = 0 is returned indicating a failure to reject the null hypothesis. The KS-test comparing wetland well daily dist ributions of values and representative monthly distributions of values was used to determine if monthly well records are statistically similar to daily we ll records. The null hypothesis fo r this test is that the daily data set and the monthly data set were draw n from the same con tinuous distribution. Two sets of KS-tests were performed compar ing the following data sets: 1) daily well values versus the well record value recorded on the 15th of each month and 2) daily well values versus monthly average values. The tests were limited to 19 wetland wells due to limited daily recorded well levels.

PAGE 188

176 Appendix C: (Continued) The Kolmogorov-Smirnov test results comparing: 1) daily well values versus monthly average values and 2) daily well values vers us the well record value on recorded on the 15th of each month are shown in Table C.1. Based on the two-sample tests, the null hypothesis could not be rejected for either test scenario. Therefore, both the water level distributions comprised of monthly average va lues and the single day value were drawn from the same distribution as the daily wetland well samples for all of the 19 wetland wells. The test results for each of the 38 individual tests had h = 0 and p = 1.0. The analysis indicates that monthly well record time series are statistically indistinguishable from the daily well record time series. This finding is supported by (Foster et al. 2008) in which the authors comp ared frequency distri butions based on daily and monthly data. These results indicate th at hydrologic studies c ould be designed based on monthly data records instead of daily data records, which could eliminate the use of costly continuous data recorders or the need to manually record well water levels on a daily basis. References: Foster, L. D., Shah, N., Ross, M., Ladde, G. S., and Wang, P. (2008). "Using frequency analysis to determine wetland hydroperiod." Neural, Parallel Sci. Comput. 16(1), 17-34. Massey, F. J., Jr. (1951). "The Kolmogorov-Smirnov Test for Goodness of Fit." Journal of the American Statistical Association 46(253), 68-78. StatSoft Inc. (2010). "Electronic Statisti cs Textbook." Tulsa, OK: StatSoft. WEB: http://www.statsoft.com/textbook/stathome.html (Dec. 2009).

PAGE 189

177 Appendix C: (Continued) Table C.1. Wetland well daily versus monthly data distribution comparisons. 15th of the Month Monthly Average Wtld WW Data Count Kolmogorov-Smirnov test Wilcoxon Rank-Sum Kolmogorov-Smirnov test Wilcoxon Rank-Sum ID ID Daily Monthly h p D stat p valid h p h p D stat p valid h p 20 1959 1819 61 0.000 0.9940.054 59 0.0000.891 0.0000.9990.04759 0.000 0.907 70 1932 1440 47 0.000 0.9650.072 46 0.0000.907 0.0000.9970.05847 0.000 0.964 84 1989 1399 46 0.000 0.9220.081 45 0.0000.801 0.0000.8710.08745 0.000 0.625 170 1987 1329 43 0.000 0.9640.076 42 0.0000.782 0.0000.9790.06944 0.000 0.749 183 1954 1741 57 0.000 1.0000.046 55 0.0000.959 0.0000.9440.06858 0.000 0.947 196 1992 1362 45 0.000 0.9910.065 44 0.0000.837 0.0000.9660.07245 0.000 0.961 215 1929 1797 58 0.000 0.9950.055 56 0.0000.856 0.0000.9790.06159 0.000 0.872 295 1990 1302 43 0.000 0.8790.089 42 0.0000.670 0.0000.9900.06643 0.000 0.882 388 1988 1402 46 0.000 0.9850.067 45 0.0000.836 0.0000.9800.06845 0.000 0.803 541 1991 1175 38 0.000 0.9470.084 37 0.0000.755 0.0000.7880.10339 0.000 0.759 605 1966 1178 38 0.000 0.9600.082 37 0.0000.734 0.0000.9110.08740 0.000 0.833 1319 1961 2144 69 0.000 0.9800.057 67 0.0000.876 0.0000.9940.04971 0.000 0.844 1320 1960 2131 69 0.000 0.9980.047 67 0.0000.771 0.0000.9880.05371 0.000 0.991 1325 1977 1950 64 0.000 0.9300.068 62 0.0000.815 0.0000.6580.09064 0.000 0.831 1326 1978 1670 55 0.000 0.9990.049 53 0.0000.996 0.0001.0000.04755 0.000 0.997 1329 1981 1813 59 0.000 0.8750.077 57 0.0000.795 0.0000.6940.09060 0.000 0.761 1337 1995 1175 38 0.000 1.0000.057 37 0.0000.783 0.0000.9810.07339 0.000 0.792 3713 2064 1956 65 0.000 0.9960.051 63 0.0000.910 0.0000.9550.06463 0.000 0.731 3715 2060 1959 65 0.000 0.9910.054 63 0.0000.904 0.0000.8530.07563 0.000 0.696

PAGE 190

178 Appendix D: SWFWMD White Papers on Wetland Histories Uid 20 Morris Bridge E ast Cypress Marsh The Morris Bridge East Cypress Marsh was one of several wetlands selected by the District for monitoring in th e 1970s. Hydrologic monitoring be gan in 1977. On or about this time a stilling well recorder was instal led. In 2000 a shallow upland well was added and in 2001 a 6-inch shallow wetland well. Hydrologic information is part of the District's Hydrologic Data Base (HDB). Along with other Morris Bridge wetlands sele cted for monitoring in the 1970s, plots and transects were installed at approximately the time of staff gage inst allation. Biological monitoring was conducted at least yearly from the 1970s through the 1990s. From 20002006 East Cypress Marsh was monitored using the Wetland Assessment Procedure (WAP). The 1983 Review report indicate s that plant specie s characteristic of more upland areas invaded wetland monitoring plots in the early 1980s. These included broomsedge bluestem (Andropogon virginicus), dogfennel (Eupatorium ca pillifolium) and slender flattop goldenrod (Solidago micr ocephala). Sustained ground-water production during this time was thought to be a contributor to reduced wetland hydr operiod and invasion by upland species. Improved wetland conditions in recent years might be attributed to a sizable reduction in overall we llfield pumping along with mo re normal rainfall. Wetland conditions in the East Cypress Marsh are chan ged somewhat from those seen in the 1970s although the cypress canopy remains in good condition. References: Lopez, M. 1980. Hydrobiological monitori ng of Morris Bridge Wellfield, Hillsborough County, Florida: 1978-1979 update. SWFWMD Environm ental Section Technical Report 1980-1. 68 pp. Lopez, M. 1983. Hydrobiological Mon itoring of Morris Bridge Well Field, Hillsborough County, Florida. A Review: 1977-1982. Environmental Section Technical Report 1983-5. 96 pp. Mumme, R.L. 1978. Hydrobiological m onitoring of Morris Bridge Wellfield, Hillsborough County, Florida. SWFWMD Environmental Section Technical Report 1978-3. 42 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 191

179 Appendix D: (Continued) Uid 21 STWF "B" (Grass Prairie) STWF "B" marsh is located in the western part of the Starkey Wellfield just east of the large pasture south of the entrance road to the Park. The large marshy wetland is usually called Grass Prairie. Marsh "B" may historic ally have been a large grassy lake since there is considerable elevational decline from the palmetto fringe to the marsh center. In 1975 a staff gauge, transect and two meter-square vegetation plots were installed in the marsh. Transect and meter-square monitori ng of vegetation was conducted from 1975 to 2001. After 2001, vegetational monitoring in formation comes from the WAP (Wetland Assessment Procedure). A surficial upland monitoring well was added by SWFWMD in 1999 and a surficial wetland monitoring well next to the staff in 2001. Hydrological information from SWFWMD's installations is part of the WMDB. Another SWFWMD staff and wetland shallow well (STWF "G") is located on the eas tern side of the Grass Prairie marsh. Water fluctuations at "B" and "G" are very similar and suggest that water is part of one system rather two separate pools. Tampa Ba y Water monitors Grass Prairie with a site called S-24 near the northern end of the wetland system. Surface waters at the Marsh "B" site have n early always been low relative to control marshes during 30 years of SWFWMD mon itoring. It has been assumed that groundwater production from four wells in the we stern part of the wellfield starting prior to 1975 has been a major influence in the rela tively low waters levels. In some years water levels have only risen slig htly at the staff gauge and have been well below the level of saw palmettos at th e edge of the marsh. In 1975 pickerelweed (Pontederia cordata) was quite abundant in the staff gauge area but over the years pickerelweed has either disa ppeared or been much less abundant. Fennel (Eupatorium spp.) and bluestem (Andropogon spp. ) have often occupied the area near the staff. Spadeleaf (Centella as iatica), bluestem and fennel have often been abundant in the marsh fringe area over the years. Vegetati onal trends are depicted in graphs that accompany the history. The accompanying file of photos taken at Grass Prairie "B" shows th e vegetational trends noted. The photos also add evidence to obser vations that the stand of red maples (Acer rubrum) in the central area of the Grass Prairie marsh has expanded and has mostly blocked the view across the marsh. The area of red maples is part of a floating mat occupying an extensive area of the marsh. Due to the invasion of fennel, bluestem and other shallow-water plants, the health of Grass Prairie "B" has been poor in below ra infall years although recovery occurs in above-normal rainfall years. In addition to th e effects of rainfall, wetland conditions in the marsh are likely affected by the level of water production from wells in the western Starkey area.

PAGE 192

180 Appendix D: (Continued) STWF Marsh "B" was last visited for vegetati onal observations in July, 2006. The most noteworthy new occurrence was a large cons picuous soil slump f eature about 50 feet north of the staff. The soil slump area was es timated to have dimens ions of 30 x 200 feet and to be 1-2 feet deep. The slumped ar ea was photographed and the photograph added to the photo-file that accompanies the history. The most likely cause of the soil slump feature is a history of depressed su rface water levels in the marsh. References: Southwest Florida Water Management District 1976. Biological assessment of the Jay B. Starkey Wilderness Park. SWFWMD E nvironmental Section Technical Report 19764. 135 pp. Rochow, T.F. 1982. Biological assessment of the Jay B. Starkey Wilderness Park --1982 update. SWFWMD Environmental Sec tion Technical Report 1982-9. 58 pp. Rochow, T.F. 1983. 1983 Photographic survey of the Jay B. Starkey Wilderness Park. SWFWMD Environmental Section Technical Memorandum 4-27-83. Rochow, T.F. 1984. 1984 Photographic survey of the Jay B. Starkey Wilderness Park. SWFWMD Environmental Section Technical Memorandum 4-27-84. Rochow, T.F. 1985. Biological assessment of the Jay B. Starkey Wilderness Park --1985 update. SWFWMD Environmental Sec tion Technical Report 1985-4. 105 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 193

181 Appendix D: (Continued) Uid 51 EWWF #5 History of the EWWF #5 cypress dome located in the western part of the Eldridge-Wilde Wellfield extends back to the early 1980s. The EWWF #5 dome was visited many times during the period from 1982 to 1994 partly to collect environmental information for SWFWMD's consumptive use evidentiaries (CUP 202673). During this time the dome was called the TR-PMD #5 dome. A consider able number of leaning and fallen cypress trees were noted as well as visible fire burn scars on trees. Mo re light than normal appeared to penetrate through the cypress canopy leading to considerable grasses and sedges in the understory. The dome appeared drier than no rmal when observed on site visitations. Observations through the 1990s showed at times dense fennel (Eupatorium spp.) along with considerable amoun ts of blackberries (Rubus spp.). In 1989 a staff gauge was established in EWWF #5 and in 2001 wetland and upland surficial wells were added. The hydrologic information is part of the WMDB. Observations of Dome #5 continued and incr eased as part of the Northern Tampa Bay Water Resources Assessment Project (1996). Mo re than forty site visitations occurred between 1989 and 1999. The dome continued to have an impacted appearance during this time. Only on rare occasions was standing water noted. Abundant fennel (Eupatorium spp.) was commonly noted as we ll as some blackberry (Rubus spp.). From 2000 through 2006, the dome was assessed using the WAP (Wetland Assessment Procedure). Over the entire period of obser vation the health of the dome has generally been poor although a considerable number of cypress trees are still standing. General surveillance of aerial photography over the wellfield starti ng before wellfield pumpage leads to the conclusion that impacts to the EWWF #5 dome started not too long following the initiation of pump ing at the wellfield in 1956.

PAGE 194

182 Appendix D: (Continued) References: Rochow, T.F. 1988. Eldridge-Wilde Well Field (CUP 202673) environmental evaluation. Southwest Florida Water Management District memorandum. August 24, 1988. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp. Rochow, T.F. 1998. Investigation of hi storic aerial photography in and around the Eldridge-Wilde Wellfield. SWFW MD Memorandum. August 31, 1998. Rochow, T.F. and P. Rhinesmith. 1991. Comp arative analysis of biological conditions in five cypress dome wetlands at the Starkey and Eldridge-Wilde well fields in southwest Florida. SWFWMD Environmental Section Technical Report 1991-1. 67 pp. Southwest Florida Water Management District 1982a. Evidentiary evaluation, CUP No. 202673, Eldridge-Wilde Wellfield, Renewal. February 24, 1982. Southwest Florida Water Management District. 1982b. Historic impact on wetlands within the Eldridge-Wilde We llfield, Work Order Number 238. April 13, 1982. Memorandum by Rock G. Taber. Southwest Florida Water Management District 1989. Staff Report. Consumptive Use Permit Application No. 202673.02, Eldridge-Wilde Wellfield, May 25, 1989. Southwest Florida Water Management Dist rict. 1996. Northern Tampa Bay Water Resources Assessment Project. Volume one. Surface-Water/Ground-Water Interrelationships. Resource Evaluation Section, SWFWMD

PAGE 195

183 Appendix D: (Continued) Uid 70 STWF "FF" The STWF "FF" dome lies in S. 12, T. 26, R. 17 south of the Starkey Wellfield and the Anclote River and west of the Suncoast Park way. The wetland is surrounded mostly by flatwoods. The dome lies approximately on e mile from the easternmost Starkey production well. A staff was installed in the dome in 1988 as part of the Northern Tampa Bay Water Resources Assessment Project (SWFWMD, 1996). During the 1990s a stilling well with recorder was added. A shallow upland well was installed in 2000 a nd a six-inch wetland well next to the staff in 2001. Hydrologic da ta is part of the Water Management Data Base. Quantitative meter-square monitoring has not been conducted in the "FF" dome but numerous observations with writ ten notes as well as photographs have been taken. The photographs are in the photograp hic file which accompanies th e history. Approximately forty site visitations with not es were recorded from 1989-1999. On site visitations the dome has often b een well hydrated although notes reveal water levels were sometimes lower-than-expected when compared to control cypress dome wetlands. At times scattered fennel (Eupatori um spp.) has been observed. The ecologic conditions in the dome are regarded as good although it is possible that some surface water depression has occurred at times. Since 2000, environmental conditions have been monitored with the WAP. Reference: Northern Tampa Bay Water Resources Assessm ent Project. 1996. Volume One. Surface-Water/Ground-Water Interrelationships. Resource Evaluation Section, Southwest Florida Water Manage ment District. March 1996

PAGE 196

184 Appendix D: (Continued) Uid 81 UHFDA North Marsh The North Marsh is located in SWFWMD's Upper Hillsborough Flood Detention Area close to the north fence line (S. 8, T. 26, R. 22). The North Marsh is composed of a large central marsh surrounded by a narrow fringe of cypress trees. The marsh was first observed in the late 1970s and the earliest photographs in the photo-file accompanying the history date from this time. A staff was installed in 1982. A shallow upl and well was installed in 2000 and a shallow wetland well in 2001. The current staff in the marsh lies in a deep gator-like depression. In the years since 1982 a manufactured housing pa rk was built just north of the marsh on private land. Runoff from the park and/or package treatment plant at times augments marsh waters. The marsh is very large however, so water augmenting the marsh is likely not great. The District has not monitored the North Mars h with quantitative monitoring installations but has relied on the photographic record and observations for knowledge of ecologic conditions. Starting in 2000 the North Marsh was monitored using the Wetland Assessment Procedure (WAP). At times the marsh-like area of North Mars h has had considerable amounts of duckpotato (Sagittaria lancifolia), pick erelweed (Pontederia cordata) spatterdock (N uphar lutea), maidencane (Panicum hemitomon) and sm artweed (Polygonum hydropiperoides). At other times dog fennel (Eupatorium capillifoliu m) has been abundant. Observations over a long time period have shown that peak wate r levels seldom ente r the cypress fringe therefore allowing wax myrtle a nd slash pines to invade along the ground. In addition to generally depressed water levels, dry cycles ap pear to be more severe in the North Marsh than in control wetlands. Whether this apparent abnormality in the hydrologic behavior of the marsh is valid need further analysis. The UHFDA North Marsh has not been men tioned in any District publications and therefore no references are given. Hydrologi c measurements are part of the District's Hydrologic Data Base.

PAGE 197

185 Appendix D: (Continued) Uid 84 Green Swamp Dome #3 Green Swamp Dome #3 along with five ot her Green Swamp domes was selected for monitoring in 1979 (S. 20, T. 24, R. 24). A staff gauge was in stalled and hydrologic monitoring began in May 1979. A stilling well was installed not too long after staff reading began. The depth of the stilling well is not available a nd a somewhat sketchy historical record indicates that the stilling well may have been deepened one or more times over the lengthy period of observation. A two-inch upland surficial well was installed in 1999 and a six-inch wetland surficial well in 200 1 Instrumentation on the stilling well was moved to the wetland surficial well after it was drilled. Three meter-square vegetational plots (A-C) were installed in vegeta tional zones from the center of dome (A) to the edge (C). The gr aphical plots are in a file accompanying the history. The plots were initi ally sampled for percent plant species coverage in May, 1981 and November, 1981 --at a later date samp ling was changed to once per year in MayJune. Yearly quantitative vegetational sampling continued until June, 2002. A report on 1979-1982 monitoring was completed in 1984. Many photographs were taken over the years of the exterior and inte rior of the dome. Some of the photos are shown in the photo-file which accompanies the history. Hydrologic conditions over the years in the dome have been close to what is normally reported in the literature for isolated cypre ss domes. Dome #3 at the present time is in good condition. Reference: Rochow, T.F. and M. Lopez. 1984. Hydrobiological monitoring of cypress domes in the Green Swamp area of Lake and Sumter c ounties, Florida 1979-1982. Environmental Section Technical Report 1984-1. Southwest Fl orida Water Management District. 79 pp.

PAGE 198

186 Appendix D: (Continued) Uid 89 J.B. Starkey #2 The J.B. Starkey #2 cypress dome (S. 23, T. 26, R. 17) was selected for monitoring in 1989 and a staff gauge placed in the dome at this time. A 2-inch shallow wetland well and a 2-inch shallow upland well were added in 2001. The water level record is part of the District's Hydrologic Data Base (HDB). J.B. Starkey dome #2 has predominantly been a hydrologic monitori ng site with ecologic monitoring of secondary importance unt il WAP monitoring was begun in 2000. However, nearly forty site visitations with observations of conditions were made and notes taken during the 1990s. Hydrology of the dome and vegetation appeared close to control domes in the District 's monitoring network. From 2000 to 2006 the dome was monitored w ith the WAP. WAP monitoring scores appear to indicate that the dome continues to be in good condition. Photographs of conditions in the dome are included in the pho to-file that accompanies this history.

PAGE 199

187 Appendix D: (Continued) Uid 112 STWF "BB" The Starkey "BB" cypress dome (S. 2, T. 26, R. 17) was selected for monitoring in 1985 and a staff gauge installed at the time. A shallow upland well was drilled in 2000 and a shallow wetland well next to the staff in 2001. No quantitative monitoring has been c onducted in the STWF "BB" dome but a descriptive transect was inst alled in the dome in 1985 with detailed descriptions at intervals along the transect Notes indicate the dome was in good health with considerable amounts of chain fern (Woodwardia virginica), lesser pipewort (Eriocaulon compressum), and giant pipewort (Eriocaulon decangulare) in the unde rstory. Fetterbush (Lyonia lucida) was noted as common th rough the dome. The cypress canopy was healthy. Many observations of STWF "BB" were made in the period from 1985 to 2005. Healthy canopy, shrub, and understory conditions were noted during this time. Observations showed the dome to be reason ably well hydrated wit hout the appearance of depressed surface water levels shown by other do mes in the central part of the wellfield. Starting in 2000 the "BB" dome was assesse d with the Wetland Assessment Procedure (WAP). WAP scores indicate the cypress stand is in good condition.

PAGE 200

188 Appendix D: (Continued) Uid 136 STWF "EE" The Starkey "EE" cypress dome (S. 1, T. 26, R. 17) in the eastern area of the wellfield was selected for monitoring in 1988 and a staff gauge installed at this time. A shallow upland well was drilled in 2000 and a shallo w wetland well next to the staff in 2001. No plot monitoring has been conducted in the STWF "EE" dome but a descriptive transect was installed in the dome in 1988 with detailed descriptions made at intervals along the transect. Notes indicate the dom e was in good health with a good sandweed (Hypericum fasciculatum) fringe and c onsiderable amounts of lesser pipewort (Eriocaulon compressum) in the outer cypres s fringe area. Fetterbush (Lyonia lucida) was noted to be common through the dome The cypress canopy was healthy. Many observations of STWF "EE" were made in the period from 1988 to 2005. Healthy canopy, shrub, and understory conditions were noted during this time. The dome is about 4000 feet from the easternmost Starkey water production well. At th is distance signs of water table drawdown are not very apparent and hydrology of the dome is close to that expected under natural conditions. Starting in 2000 the "EE" dome was assessed with the Wetland Assessment Procedure (WAP). WAP scores indicate that vegeta tional conditions in th e dome are quite good.

PAGE 201

189 Appendix D: (Continued) Uid 143 STWF "T" The Starkey "T" dome (S. 2, T. 26, R. 17) lo cated just west of the Cross Cypress slough in the central part of the wellfield was select ed for monitoring in 1983. A staff gauge was placed in the dome in 1983 along with a shallow upland well in 2000 and a shallow wetland well in 2001. The hydrologic record is part of the District's Hydrologic Data Base (HDB). Vegetational conditions in the dome were m onitored from 1983 to 2003 with inner and outer meter-square plots. The inner plot is close to the staff and there is little rooted vegetation in the plot due to deep water duri ng the summer rainy season. Floating eastern purple bladderwort (Utricularia purpurea) has typically b een abundant during wet years in the monitoring plot. During the first ten years of monitoring, le sser pipeworts (Eriocaulon compressum) were common in the outer meter-square plot. In 1993 the plant covered 70% of the plot. Since 1993 lesser pipeworts have been considerably less abundant in the outer meter-square area. Following 2000, lesser pipeworts essent ially disappeared from the monitoring area although a few could be found nearby. Horned rush (Rhynchospora corniculata) appears to have replaced lesser pipeworts in the area. Dome "T" was last observed in June, 2006. When Dome "T" was initially observed in th e early 1980s the dome was called "Polygala Head" due to the abundance of yellow milk worts (Polygala cymosa). From recent observations the plant appears to have a much-reduced presence in the dome possible for the same reasons as lesser pipeworts. Since 2000 the Wetland Assessment (WAP) has b een used to monitor vegetation in the dome. Observations have shown that in recent years wax myrtle has moved a considerable distance into the edge of the dome on the ground. This could be due to somewhat depressed surface-water levels in recent years. When the dome was last observed in 2006 wax myrtle was noted to be stressed/dead perhaps indicating improved water levels. The overall health of the dome is good with a full cypress canopy and no leaning or fallen trees. A photo-file accompanies the history sh owing some of the conditions in the dome.

PAGE 202

190 Appendix D: (Continued) References: Rochow, T.F. 1984. 1984 Photographic survey of the Jay B. Starkey Wilderness Park. SWFWMD Environmental Section Technical Memorandum 4-27-84. Rochow, T.F. 1985. Biological assessment of the Jay B. Starkey Wilderness Park --1985 update. SWFWMD Environmental Sec tion Technical Report 1985-4. 105 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 203

191 Appendix D: (Continued) Uid 154 Pasco Trails Wetland monitoring at the Pasco Trails Cypress Marsh (S. 11, T. 25, R. 18) was initiated in 1984 with the installation of a staff gauge. The wetland is about one mile south of the Cross Bar Wellfield and three miles northwest of the Cypress Creek Wellfield. The wetland has an extensive marsh in the area of the staff and dense cypress at the wetland edge. Entrance to the site on private property is by license agreement. In 2001 a shallow wetland well was installed next to the staff and a shallow upland well in the uplands at the edge of the wetland. Unlike a number of the District's wellfield monitoring sites, no sampling plots were installed at the Pasco Trails Cypress Marsh. Monitoring was accomplished in the 1980s and 1990s by yearly examination of a descri ptive transect, photography, and written observations. During this time records show more than forty visits to the wetland for staff reading and observation of environmenta l conditions. Photographs are included in the photo-file which accompanies the history. During the 1980s a dense growth of picker elweed (Pontederia co rdata) and bulltongue arrowhead (Sagittaria lancifolia) existed in the marsh area near the staff. At more moderate depths in the marsh several sp ecies were common including spikerush (Eleocharis spp.), horned beaksedge (Rhynchospora cornic ulata), white water lily (Nymphaea odorata), and floating hearts (Nym phoides aquatica). Lesser pipeworts (Eriocaulon compressum) were common in the cypress fringe area. During the 1990s water levels in the cypress marsh were observed to be much lower than in control wetlands some distance from wellfield pumping. At this time maidencane (Panicum hemitomon) and fennel (Eupatorium spp.) were frequently noted in the central marsh area. In response to depressed wa ter levels, blue maidencane (Amphicarpum muhlenbergianum) in the 1990s became abundant in the cypress fringe. In recent years water levels in the Pasco Trai ls Cypress Marsh appear to have improved. Wetland vegetation in the central marsh area has improved but the appearance of the wetland has changed from what was originally seen in the 1980s. A small area of soil slumping and fissuring exists not far from the staff. From 2000-2006 the Cypress Marsh has been monitored with the Wetland Assessment Procedure (WAP).

PAGE 204

192 Appendix D: (Continued) Uid 165 Morris Bridge X-6 Dome The Morris Bridge X-6 Dome was added to monitoring program at the Morris Bridge Wellfield in 1985 with the inst allation of a staff gage. The monitoring program began in the 1970s but the need was seen in the 1980s to expand geographic coverage of the program and hence several other wetlands were added in the 1980s. A shallow upland well was installed for hydrologic monitoring in 1999 and a shallow wetland well next to the staff in 2001. Hydrologic information is pa rt of the District's Hydrologic Data Base (HDB). Unlike wetlands selected for monitoring in the 1970s, the X-6 dome was not monitored quantitatively using plots and transects. It has also not been mentioned in several comprehensive technical reports on Morris Bridge monitoring since the last such monitoring report on the wellfie ld was in 1983. Biological information on the dome has relied on photography, observations, and re peat site visitations from 1985 to 2005. Photographs of wetland conditions are include d in the photo-file which accompanies the history. Since 2000 the Wetland Assessment Procedure (WAP) has been used at the wetland to assess environmental conditions. Relatively high pumpage rates at Morris Bri dge in the early 1980s may have caused a drying of the X-6 dome in the mid to late 1980s (Rochow, 1998). In recent years hydrologic and biologic conditions have been quite good on site visitations. References: Lopez, M. 1983. Hydrobiological Mon itoring of Morris Bridge Well Field, Hillsborough County, Florida. A Review: 1977-1982. Environmental Section Technical Report 1983-5. 96 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 205

193 Appendix D: (Continued) Uid 170 Green Swamp Dome #1 Green Swamp Dome #1 along with five ot her Green Swamp domes was selected for monitoring in 1979 (S. 30, T. 24, R. 24). A staff gauge was in stalled and hydrologic monitoring began in May 1979. A stilling well was installed not too long after staff reading began. The depth of the stilling well is not available a nd a somewhat sketchy historical record indicates that the stilling well may have been deepened one or more times over the lengthy period of observation. A two-inch upland surficial well was installed in 1999 and a six-inch wetland surficial well in 200 1. Instrumentation on the stilling well was moved to the wetland surficial well after it was drilled. Four meter-square vegetational plots (A-D) were installed in vegeta tional zones from the center of dome (A) to the edge (D). The graphical plots are in a file accompanying the history. The plots were initi ally sampled for percent plant species coverage in May, 1981 and October, 1981 --at a later date sampling wa s changed to once per year in May-June. Yearly quantitative vegetational sampling continued until June, 2002. A report on 19791982 monitoring was completed in 1984. Many photographs were taken over the years of the exterior and interior of the dome. Some of the photos are shown in the photo-file which accompanies the history. Dome #1 was severely burned in November 1980 by a fire through the entire dome. Even though a number of cypress were lost from the canopy, by spring1981 surviving cypress were beginning to resprout Over the next several ye ars, the effects of the fire gradually disappeared with the vigorous resprout ing of the cypress and understory shrubs. Dome #1 at the present time appears much as it did prior to th e fire. Hydrologic conditions over the years in the dome have been close to what is normally reported in the literature for isolated cypress domes. There have been considerable fluctuations in understory plant composition over the years in Dome #1. Much of the fluctuation is attributed to the opening of the dome to light af ter the fire and then the gradual shading of the understory as shrubs and the de nse cypress overstory was restored. Reference: Rochow, T.F. and M. Lopez. 1984. Hydrobiological monitoring of cypress domes in the Green Swamp area of Lake and Sumter c ounties, Florida 1979-1982. Environmental Section Technical Report 1984-1. Southwest Fl orida Water Management District. 79 pp.

PAGE 206

194 Appendix D: (Continued) Uid 183 Morris Bridge Trout Creek Marsh Morris Bridge Trout Creek Marsh was one of se veral wetlands selected by the District for monitoring in the 1970s. Hydrologic monitoring began in 1977. On or about this time a stilling well recorder was installed. In 1999 a shallow upland well was added and in 2001 a 6-inch shallow wetland well. Hydrologic information is part of the District's Hydrologic Data Base (HDB). Along with other Morris Bridge wetlands sele cted for monitoring in the 1970s, plots and transects were installed at approximately the time of staff gage inst allation. Biological monitoring was conducted at least yearly from the 1970s through the 1990s. From 20002006 Trout Creek Marsh was monitored us ing the Wetland Assessment Procedure (WAP). Trout Creek Marsh is somewhat more distant from the center of the wellfield than other monitored Morris Bridge wetlands and was observed to experience more moderate wellfield impacts when 17.2 mgd (yearly av erage) was pumped from the wellfield in 1982. The 1983 review report indicates a reduction of sa ndweed (Hypericum fasciculatum) and branched hedgehyssop (Gra tiola ramosa). Dogfennel (Eupatorium capillifolium) became prominent for a time in the early 1980s. Continued observations of Trout Creek Marsh as wellfield producti on was reduced indicate improvement in hydrologic and biologic conditions in the ma rsh although vegetation differs somewhat from than of the 1970s.

PAGE 207

195 Appendix D: (Continued) References: Hancock, M.C., T. Rochow and J. Hood. 2005. Review of original wetland assessment procedure (WAP March 2000) an d test results of a proposed revision to the WAP, May 2004. Southwest Florida Water Manageme nt District, Brooksville, FL 121 pp. Hancock, M.C., T. Rochow, and J. Hood. 2005. Test results of a proposed revision to the Wetland Assessment Procedure (WAP), Oc tober 2004 and Development of the Final WAP Methodology Adopted in April 2005. Southwest Florida Water Management District, Brooksville, FL 147 pp. Lopez, M. 1980. Hydrobiological monitori ng of Morris Bridge Wellfield, Hillsborough County, Florida: 1978-1979 update. SWFWMD Environm ental Section Technical Report 1980-1. 68 pp. Lopez, M. 1983. Hydrobiological Mon itoring of Morris Bridge Well Field, Hillsborough County, Florida. A Review: 1977-1982. Environmental Section Technical Report 1983-5. 96 pp. Mumme, R.L. 1978. Hydrobiological m onitoring of Morris Bridge Wellfield, Hillsborough County, Florida. SWFWMD Environmental Section Technical Report 1978-3. 42 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 208

196 Appendix D: (Continued) Uid 196 Green Swamp Dome #6 Green Swamp Dome #6 along with six othe r Green Swamp domes was selected for monitoring in 1979 (S. 13, T. 24, R. 23). A staff gauge was in stalled and hydrologic monitoring began in May 1979. A stilling well was installed not too long after staff reading began. The depth of the stilling well is not available a nd a somewhat sketchy historical record indicates that the stilling well may have been deepened one or more times over the lengthy period of observation. A two-inch upland surficial well was installed in 1999 and a six-inch wetland surficial well in 200 1 Instrumentation on the stilling well was moved to the wetland surficial well after it was drilled. Four meter-square vegetational plots (A-D) were installed in vegeta tional zones from the center of dome (A) to the edge (D). The graphical plots are in a file accompanying the history. The plots were initi ally sampled for percent plant species coverage in May, 1981 and October, 1981 --at a later date sampling was changed to once per year in MayOctober. Yearly quantitative vegetational sampling continued until June, 2002. A report on 1979-1982 monitoring was completed in 1984. Many photographs were taken over the years of the exterior and interior of the dome. Some of the photos are shown in the photo-file which accompanies this history. Hydrologic conditions over the years in the dome have been close to what is normally reported in the literature for isolated c ypress domes. Dome #6 has remained in good condition over the entire period of monitoring. References: Berryman & Henigar, Inc. 2005. Vertical di stribution of vegetation species relative to normal pool elevations in ten isolated we tlands in the Northern Tampa Bay area. Prepared for Tampa Bay Wa ter, Clearwater FL Rochow, T.F. and M. Lopez. 1984. Hydrobiological monitoring of cypress domes in the Green Swamp area of Lake and Sumter c ounties, Florida 1979-1982. Environmental Section Technical Report 1984-1. Southwest Fl orida Water Management District. 79 pp.

PAGE 209

197 Appendix D: (Continued) Uid 201 Morris Bridge West Cypress Morris Bridge West Cypress was one of severa l wetlands at the Morris Bridge Wellfield selected by the District for monitoring in the 1970s. A staff gage was installed in 1977. A shallow upland well was installed in 1999 and a shallow wetland well in 2001. The hydrologic record is part of the Distri ct's Hydrologic Data Base (HDB). Along with other Morris Bridge wetlands sele cted for monitoring in the 1970s, plots and transects were installed at approximately the time of staff gage inst allation. Biological monitoring was conducted at least yearly from the 1970s through the 1990s. From 20002006 Clay Gully Cypress was monitored using the Wetland Assessment Procedure (WAP). The 1983 review report notes 17.2 mgd (yearly average) for 1982 was pumped from Morris Bridge. Although several other cypr ess and marsh systems in the wellfield showed adverse hydrologic and biologic impact s at this time, wetland vegetation at West Cypress remained healthy. Unlike some of the other wetland systems, lesser pipewort continued to be abundant in the wet meadow zone near the edge of the dome (Lopez, 1983). Observations up though 2005 have indicated a healthy cypress canopy and understory in the wetland. References: Hancock, M.C., T. Rochow, and J. Hood. 2005. Test results of a proposed revision to the Wetland Assessment Procedure (WAP), Oc tober 2004 and Development of the Final WAP Methodology Adopted in April 2005. Southwest Florida Water Management District, Brooksville, FL 147 pp. Lopez, M. 1980. Hydrobiological monitori ng of Morris Bridge Wellfield, Hillsborough County, Florida: 1978-1979 update. SWFWMD Environm ental Section Technical Report 1980-1. 68 pp. Lopez, M. 1983. Hydrobiological Mon itoring of Morris Bridge Well Field, Hillsborough County, Florida. A Review: 1977-1982. Environmental Section Technical Report 1983-5. 96 pp. Mumme, R.L. 1978. Hydrobiological m onitoring of Morris Bridge Wellfield, Hillsborough County, Florida. SWFWMD Environmental Section Technical Report 1978-3. 42 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 210

198 Appendix D: (Continued) Uid 215 J.B. Starkey #1 The J.B. Starkey #1 cypress dome (S. 24, T. 26, R. 17) was selected for monitoring in 1989. A 6-inch stilling well with a water leve l recorder was installed in the dome at or shortly after this time. A 6-inch shallow wetland well and a 2-inch shallow upland well were added in 2001. The water level record is part of the District's Hydrologic Data Base (HDB). J.B. Starkey dome #1 has predominantly been a hydrologic monitori ng site with ecologic monitoring of secondary importance. Observa tions of conditions were made periodically and notes taken during the 1990s. Genera lly the dome was in good ecological and hydrologic condition during this time. It is worth noting that a reason for selecting the dome as part of the District's monitori ng program was that loblolly bay (Gordonia lasianthus) existed at the e dge of the dome. Therefore the wetland was different from others in the District's we tland monitoring network and pr ovided diversity in wetlands being monitored. From 2000-2006 J.B. Starkey #1 was monitored systematically using WAP methodology. The very central part of the dome near the recorder is open to light --pickerelweed (Pontederia cordata) and bladde rwort (Utricularia spp.) are often seen in this area. The dome understory is hummocky with ferns on the hummocks but litt le noteworthy groundcover vegetation. The transition of the dome to the uplands is narrow with inkberry (Ilex glabra) and fetterbush (Lyonia lucida). Muscad ine (Vitis rotu ndifolia) has been noted in the trees at the very edge of the dome. P hotographs of the wetland are in the photo-file which accompanies this history.

PAGE 211

199 Appendix D: (Continued) Uid 252 STWF "C" History STWF "C" dome located near the Power Li nes in the western portion of the Starkey Wellfield was equipped with a staff gauge, transect, and two meter-square vegetation monitoring plots in 1975. In 1975 there was no wellfield/pipe line road and water production was 2-3 miles away in the far west ern part of the wellf ield. Transect and meter-square monitoring of vegetation was conducted from 1975 to 2001. After 2001, vegetational monitoring information comes from the WAP with occasional monitoring of the meter-square quadrats. Hydrological inform ation from 1975 to present is part of the District's WMDB. Surficial upland and wetland monitoring we lls were installed in 20002001. Further information on hydrologic installations is found in the EXCEL file in the uid 252 STWF "C" History. Vegetational conditions were re latively stable in the dome from 1975 into the mid 1980s. From the mid to late 1980s up to the pres ent time there have been notable understory changes. Juncus repens, or iginally common in the inner-meter square near the staff gauge, disappeared from the area after the initial ten years of monitoring. In its place chain fern (Woodwardia virginica) has become quite common. This is evident from the inner meter-square figure in the History file. In the outer meter-square near the cypr ess fringe, Eriocaulon compressum (lesser pipewort) virtually disappeared by the late 1980s and Blechnum serrulatum (swamp fern) along with some Woodwardia virginica great ly increased in abundance. Historic photographs in the photo-file along with the outer meter-square figure confirm these vegetational changes. The changes occurred within a few years after water production wells were drilled and water production began in the central area of the wellfield in the 1980s. The dominance of ferns and absence of lesser pipewort continued to be evident when the meter square was visited in June, 2006. Photographic information and observations show that Pinus elliottii (slash pine) has increased in abundance in the outer cypress fr inge area. Observations during the 1990s indicate Amphicarpum muhlenbergianum (blu e maidencane) invaded the edge of the dome. STWF "C" was last visited in June, 2006.

PAGE 212

200 Appendix D: (Continued) References: Southwest Florida Water Management District 1976. Biological assessment of the Jay B. Starkey Wilderness Park. SWFWMD E nvironmental Section Technical Report 19764. 135 pp. Rochow, T.F. 1982. Biological assessment of the Jay B. Starkey Wilderness Park --1982 update. SWFWMD Environmental Sec tion Technical Report 1982-9. 58 pp. Rochow, T.F. 1983. 1983 Photographic survey of the Jay B. Starkey Wilderness Park. SWFWMD Environmental Section Technical Memorandum 4-27-83. Rochow, T.F. 1984. 1984 Photographic survey of the Jay B. Starkey Wilderness Park. SWFWMD Environmental Section Technical Memorandum 4-27-84. Rochow, T.F. 1985. Biological assessment of the Jay B. Starkey Wilderness Park --1985 update. SWFWMD Environmental Sec tion Technical Report 1985-4. 105 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 213

201 Appendix D: (Continued) Uid 261 Lansbrook East History: The Lansbrook East dome is locat ed north of Village Center Dr ive and east of East Lake Tarpon Road (S. 27, T. 27, R. 16) in Pinellas County. A staff gauge was installed in the wetland in 1989. Shallow wetland and upland wells were added in 2001. Water level data is part of the Distri ct's WMDB. When the dome was first observed in 1981 the surrounding land was undeveloped. In recent years a shopping cen ter parking lot and roads to residential area have surro unded the dome on all sides. Quantitative monitoring installations do not exist in the Lansbrook East dome but starting in the early 1980s observations and photogr aphs were taken on a number of site visitations as part East Lake Tarpon Wellfield study, the NTB Water Resources Assessment Project (1996) and s ubsequent studies. During s ite visitations in the late 1980s into the 1990s the dome sometimes app eared drier than normal. However, as development occurred in the 1990s, runoff fr om a parking lot, detention pond, and roadways appears to have augmented water levels in the dome. The cypress canopy at Lansbrook East since 1980 has been healthy. A few Chinese tallowtrees (Sapium sebiferum) have been observed in the dome but since the dome is large these invaders are not conspicuous. Due to deep water during the rainy season there is little understory vegetation in the interior of the dome. Starting in 2000, monitoring of the Lansbrook East dome has taken place using the Wetland Assessment Procedure (WAP). Reference: Northern Tampa Bay Water Resources Assessm ent Project. 1996. Volume One. Surface-Water/Ground-Water Interrelationships. Resource Evaluation Section, Southwest Florida Water Manage ment District. March 1996

PAGE 214

202 Appendix D: (Continued) Uid 276 SPSP-6 (NW-50) The SPSP-6 cypress is located in the eastern part of the St. Petersburg-South Pasco Wellfield. Historiically, the dome has likely been connected to the large interior cypress stand at times of high water. Monitoring of SPSP-6 began in the early 1970s. Plot sampling of the dome was undertaken along a transect in the wetland from the early 1970s until the early 1980s. The transect was called TR-6. Five SWFWMD reports documented the initial five y ears of monitoring. Although a temporary staff was placed in the wetland, hydrologic data was not put in a permanent database. The final 1982 report in the Conclusions secti on mentioned impacts to isolated wetlands in the wellfield (Bradbury and Courser, 1982). In general these were descri bed as: blow-downs of trees, late leaf-out of cypre ss, invasion of weedy terrestrial plants into formerly wetland areas, reduction and loss of wetla nd understory plants, increase d susceptibility to fire damage, expansion of drier fringe communities, and reduction of the aquatic communities." It is assumed that some of these impacts were observed at SPSP-6. In 1989 SWFWMD installed a staff gauge in th e wetland as part of the Northern Tampa Bay Water Resources Assessment Project (1996). A wetland surficial well was installed next to the staff in 2001 and an upland surficial well in 2002. Data collected from the staff and wells are in the WMDB. Forty site visitations were made to SPSP-6 as part of the NTB project from 1989-1999. Surface water in SPSP-6 was absent much of th e time and water levels were judged to be much lower-than-expected when compared to those in domes far removed from wellfield pumpage. Fennel (Eupatorium spp.) at time s was abundant. Soil subsidence of 6 inches or more was noted at cypre ss bases. Many leaning and fa llen cypress were observed as well as burn marks on cypress. During the NTB project the dome was called NW-50. During the period from 2000-2006 SPSP-6 was assessed using the WAP (Wetland Assessment Procedure). The dome continues to show a considerable levels of impacts.

PAGE 215

203 Appendix D: (Continued) References: Putnam, S.A. and R.J. Moresi. 1974. Annual report of the St. Petersburg-South Pasco Wellfield study. SWFWMD Environmen tal Section Technical Report 1974-2. Putnam, S.A. and R.J. Moresi. 1975. Sec ond annual report of the St. Petersburg-South Pasco Wellfield study. SWFWMD Environmental Section Technical Report 1975-1. Courser, W.D. and P.A. Hernandez. 1977. Third annual report of the St. PetersburgSouth Pasco Wellfield study. SWFWMD E nvironmental Section Technical Report 19771. Bradbury, K.R. and W.D. Courser. 1977. Fourth annual report of the St. PetersburgSouth Pasco Wellfield study. SWFWMD E nvironmental Section Technical Report 19774. Bradbury, K.R. and W.D. Courser. 1982. Fi fth annual report of the St. Petersburg-South Pasco Wellfield study. SWFWMD Environmental Section Technical Report 1982-6. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp. Southwest Florida Water Management Dist rict. 1996. Northern Tampa Bay Water Resources Assessment Project. Volume one. Surface-Water/Ground-Water Interrelationships. Resource Evaluation Section, SWFWMD

PAGE 216

204 Appendix D: (Continued) Uid 295 Green Swamp Dome #4 Green Swamp Dome #4 along with five ot her Green Swamp domes was selected for monitoring in 1979 (S. 17, T. 24, R. 24). A staff gauge was in stalled and hydrologic monitoring began in May 1979. A stilling well was installed not too long after staff reading began. The depth of the stilling well is not available a nd a somewhat sketchy historical record indicates that the stilling well may have been deepened one or more times over the lengthy period of observation. A two-inch upland surficial well was installed in 1999 and a six-inch wetland surficial well in 200 1 Instrumentation on the stilling well was moved to the wetland surficial well after it was drilled. Three meter-square vegetational plots (A-C) were installed in vegeta tional zones from the center of dome (A) to the edge (C). The gr aphical plots are in a file accompanying the history. The plots were initi ally sampled for percent plant species coverage in May, 1981 and November, 1981 --at a later date samp ling was changed to once per year in MayJune. Yearly quantitative vegetational sampling continued until June, 2002. A report on 1979-1982 monitoring was completed in 1984. Many photographs were taken over the years of the exterior and inte rior of the dome. Some of the photos are shown in the photo-file which accompanies the history. Hydrologic conditions over the years in the dome have been close to what is normally reported in the literature fo r isolated cypress domes. Do me #4, although burned severely in the November 1980 Green Swamp fire, has r ecovered and at the present time is in good condition. Reference: Rochow, T.F. and M. Lopez. 1984. Hydrobiological monitoring of cypress domes in the Green Swamp area of Lake and Sumter c ounties, Florida 1979-1982. Environmental Section Technical Report 1984-1. Southwest Fl orida Water Management District. 79 pp.

PAGE 217

205 Appendix D: (Continued) Uid 301 Pine Ridge Cypress Dome history: The Pine Ridge Cypress Dome is located east of East Lake Tarpon Road (S. 22, T. 27 R. 16) in Pinellas County. A staff gauge was installed in the wetland in 1989 with a stilling well recorder shortly thereafter. The reco rder was vandalized dur ing the 1990s with water level instrumentation being restored at a later date. A shallow upland well was installed in 2000 and a wetland well in 2001. Water level data is part of the District's WMDB. The dome is located on District Lower Brooker Creek land and is surrounded by flatwoods with considerable woody growth. Quantitative monitoring installa tions do not exist in the Pine Ridge dome but photographs and observations have been taken on a number of site visitations as part of the NTB Water Resources Assessment Project (1996) and subsequent studies. During site visitations over the years water levels have appeared close to normal based on observations of control domes. The cypress canopy and understory of the Pine Ridge Dome since first observed in 1988 have appeared healthy. Since 2000, monitoring of the Pine Ridge dome has been conducted using the Wetland Assessment Procedure (WAP). Reference: Northern Tampa Bay Water Resources Assessm ent Project. 1996. Volume One. Surface-Water/Ground-Water Interrelationships. Resource Evaluation Section, Southwest Florida Water Manage ment District. March 1996

PAGE 218

206 Appendix D: (Continued) Uid 304 STWF Marsh "Y" Marsh "Y" (S. 1, T. 26, R. 17), in the eastern portion of the Starkey Wellfield, was selected as part of the District's wetla nd monitoring program in 1982 and a staff gauge placed in the marsh. Marsh "Y" is less th an 0.25 acre but was chosen for monitoring since there are few marshes in the eastern pa rt of the wellfield. The marsh lies a few hundred feet east of the most eas terly of the production wells in the wellfield. STWF "Y" is distinctly sink-like in appearance and likel y formed in a manner similar to other small sinks after a collapse in the underlying kars t geology. Shallow wetland and upland wells were added in 1999. Marsh "Y" was monitored yearly from 1983 to 2002 with descriptive observations along a transect from the saw palmetto edge to the staff as well as inner and outer meter-square plots. Originally present on the inner mete r-square near the staff were creeping rush (Juncus repens), grassy arrowhead (Sagittaria graminea) and maidencane (Panicum hemitomon). Sampled on the outer meter-square during the initial years of monitoring were species such as southern beaksedge (Rhynchospora microcarpa), spikerush (Eleocharis sp.), southern umbrellasedge (Fuirena scirpoidea), sandweed (Hypericum fasciculatum), and broomsedge (Andropogon virg inicus). Photographs in the photo-file and observations along the transect indicate a distinct fringe of healthy sandweed in the early 1980s. The sandweed fringe, which give s the marsh a stratified appearance, is typical of marshes with normal hydrology --Marsh "Y" had 11-12 months of standing water at the staff in 1983 and 1984 (Rochow, 1985). Marsh "Y" began a drying trend in the mid to late 1980s that has lasted through the present time. Water production beginning in 1989 from the easternmost Starkey production well probably contributed to this drying. Dry marsh conditions led to an increase in maidencane, broomsedge, and fe nnel in the marsh. Sandweed moved from the marsh edge to the central marsh area and decreased in abundance. In the 1990s fire from the flatwoods burned through the marsh and further altered vegetational conditions. From 2000-2004, Marsh "Y" was monitored with the Wetland Assessment Procedure (WAP). The Marsh was removed from the list for WAP monitoring in 2005 due to its small size. Based on a long record of monito ring, Marsh "Y" is considered a severely impacted wetland. References: Rochow, T.F. 1985. Biological assessment of the Jay B. Starkey Wilderness Park --1985 update. SWFWMD Environmental Sec tion Technical Report 1985-4. 105 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 219

207 Appendix D: (Continued) Uid 320 Morris Bridge X-4 Dome The Morris Bridge X-4 Dome was added to mo nitoring program at the wellfield in 1985 with the installation of a staff gage. The m onitoring program at Morris Bridge began in the 1970s but the need was seen in the 1980s to expand geographic coverage of the program and hence several other wetlands were added in the 1980s. A shallow upland well was added for hydrologic monitoring in 200 0 and a shallow wetland well next to the staff in 2001. Hydrologic information is part of the District's Hydrologic Data Base (HDB). In recent years, w ith Chapter 40D-8, F.A.C. the X-4 dome has become one of four domes at Morris Bridge that are esp ecially important for assessing water level conditions over time. Unlike wetlands selected for monitoring in the 1970s, the X-4 dome was not monitored quantitatively using plots and transects. It has also not been mentioned in several comprehensive technical reports on Morris Bridge monitoring since the last such monitoring report on the wellfield in 1983. Biological informa tion on the dome has relied on photography, observations, and re peat site visitations from 1985 to 2005. Photographs of wetland conditions are include d in the photo-file which accompanies the history. Since 2000 the Wetland Assessment Procedure (WAP) has been used at the wetland to assess environmental conditions. The photographs in the photo-f ile show quite clearly that the cypress canopy of X-4 was considerably stressed in the mid 1980s when the staff gage was instal led. Relatively high pumpage rates at Morris Bridge in the 1980s are believed to be the cause of this canopy stress (Rochow, 1998). The cypress canopy ha s appeared somewhat better in recent years although evidence of stress is still evident.

PAGE 220

208 Appendix D: (Continued) References: Hancock, M.C., T. Rochow and J. Hood. 2005. Review of original wetland assessment procedure (WAP March 2000) an d test results of a proposed revision to the WAP, May 2004. Southwest Florida Water Manageme nt District, Brooksville, FL 121 pp. Hancock, M.C., T. Rochow, and J. Hood. 2005. Test results of a proposed revision to the Wetland Assessment Procedure (WAP), Oc tober 2004 and Development of the Final WAP Methodology Adopted in April 2005. Southwest Florida Water Management District, Brooksville, FL 147 pp. Lopez, M. 1983. Hydrobiological Mon itoring of Morris Bridge Well Field, Hillsborough County, Florida. A Review: 1977-1982. Environmental Section Technical Report 1983-5. 96 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 221

209 Appendix D: (Continued) Uid 331 J.B. Starkey #4 The J.B. Starkey #4 cypress marsh (S. 27, T. 26, R. 17) was selected for monitoring in 1989 and a staff gauge placed in the wetland at this time. A 2-inch shallow wetland well and a 2-inch shallow upland well were added in 2002. The water level record is part of the District's Hydrologic Data Base (HDB). At the time the wetland was selected the area surrounding was improved pasture. Not long after selection land use changed with the construction of a Pasco County wastewater treatment plant. Treated waste-water from a nearby spray field may at times enter the cypress marsh. The J.B. Starkey #4 cypress marsh has pre dominantly been a hydrol ogic monitoring site with ecologic monitoring of secondary importa nce until WAP monitoring began in 2000. However, observations of conditions were ma de periodically and notes taken during the 1990s. Generally the cypress marsh was in good ecological and hydrologic condition during this time. From 2000 to 2006 the wetland was monitored with the WAP. Photographs of conditions in the cypress-ma rsh are included in the photo-file that accompanies the history. The J.B. Starkey #4 wetland has not specifically been mentioned in any District report. In March, 2007 Pasco County was observed excavating and pumpi ng from a large pit close to the Starkey #4 wetland (see photos in the photo-file). According to communication with workers, the operation was being undertaken to lower the water table so that the nearby spray field would absorb water more efficiently. The operation is believed to have the potential to lower water levels in the cypress marsh wetland.

PAGE 222

210 Appendix D: (Continued) Uid 379 Morris Bridge South Cypress Marsh The Morris Bridge South Cypress Marsh was one of several wetla nds selected by the District for monitoring in the 1970s. A sta ff gage was installed in 1977. A shallow upland well was added in 1999 and a shallow wetland well in 2001. Hydrologic information is part of the District's Hydr ologic Data Base (HDB). The South Cypress Marsh has also been monitored for many year s by Biological Research Associates (BRA) in support of permit application for the wellf ield. BRA's name for this wetland is MBR29. Therefore, hydrologic and biologic info rmation is available from different sources for this dome. Along with other Morris Bridge wetlands sele cted for monitoring in the 1970s, plots and transects were installed at approximately the time of staff gage inst allation. Biological monitoring was conducted at least yearly from the 1970s through the 1990s. From 20002006 Clay Gully Cypress was monitored using the Wetland Assessment Procedure (WAP). The 1983 review report notes 17.2 mgd (yearly average) for 1982 was pumped from Morris Bridge. At this time a reduction in hydroperiod was noted at the South Cypress Marsh leading to invasion by weedy, terrestrial plant species and disappearance of obligate aquatic and semi-aquatic plants. Cert ain of these changes are apparent in the photo-file that accompanies the history --especially noteworthy are photographs showing severe canopy stress. With a cut-back in pumpage in the 1990s water levels and wetland health improved (Rochow, 1998). However, the stressed canopy is still very evident.

PAGE 223

211 Appendix D: (Continued) References: Hancock, M.C., T. Rochow, and J. Hood. 2005. Test results of a proposed revision to the Wetland Assessment Procedure (WAP), Oc tober 2004 and Development of the Final WAP Methodology Adopted in April 2005. Southwest Florida Water Management District, Brooksville, FL 147 pp. Lopez, M. 1980. Hydrobiological monitori ng of Morris Bridge Wellfield, Hillsborough County, Florida: 1978-1979 update. SWFWMD Environm ental Section Technical Report 1980-1. 68 pp. Lopez, M. 1983. Hydrobiological Mon itoring of Morris Bridge Well Field, Hillsborough County, Florida. A Review: 1977-1982. Environmental Section Technical Report 1983-5. 96 pp. Mumme, R.L. 1978. Hydrobiological m onitoring of Morris Bridge Wellfield, Hillsborough County, Florida. SWFWMD Environmental Section Technical Report 1978-3. 42 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 224

212 Appendix D: (Continued) Uid 384 Morris Bridge Well Marsh Morris Bridge Well Marsh was one of severa l wetlands selected by the District for monitoring in the 1970s. A staff gage wa s installed in 1977. Shallow upland and wetland wells were added in 2000. Hydrologic information is part of the District's Hydrologic Data Base (HDB). Well Marsh has also been monitored for many years by Biological Research Associates (BRA) in suppor t of permit application for the wellfield. BRA's name for this wetland is MBR-42. Therefore, hydrologic and biologic information is available from two sources for this wetland. Along with other Morris Bridge wetlands sele cted for monitoring in the 1970s, plots and transects were installed at approximately the time of staff gage inst allation. Biological monitoring was conducted at least yearly from the 1970s through the 1990s. From 20002006 Well Marsh was monitored using the Wetland Assessment Procedure (WAP). The 1983 review report notes 17.2 mgd (yearly average) for 1982 was pumped from Morris Bridge. Vegetational trends through 1989 (as a result of reduced hydroperiods) ranged from the gradual invasi on and increase in cover by upl and/terrestrial plants and facultative wetland plants to the gradual and permanent loss of oblig ate aquatic and semiaquatic plants which are dependent on re gular, sustained seasonal inundation (Rochow, 1998). The 1983 Review report notes that le sser pipewort (Eriocaulon compressum) disappeared from the marsh edge in the early 1980s. In more interior areas of the marsh, floating hearts (Nymphoides aquatica) wa s reduced in abundance and pickerelweed (Pontederia cordata) disapp eared. Maidencane (Panicum hemitomon) increased considerably in interior marsh areas. Dogfennel (Eupatorium capillifolium) was a prominent invader of the marsh during the ea rly 1980s. The photo-file accompanying the history shows some of the tre nds described. Sustained gr ound-water production, coupled with several years of below normal rainfall conditions, were believed to be prime factors affecting wetland surface water levels. For the period 1986-1989 (and through 1993), coin cident with nearly a 40 percent reduction in overall average annual wellfield pumpage, wetland monitoring information suggested a stabilization of earlier vegetation trends. Principally, "dry wetland" vegetation trends did not con tinue to progress successionally to even more terrestrial (upland) conditions (Rochow, 1998). In the mid 1990s (i.e. 1993-1997), low wellfield pumpage rates accompanied favorable rainfall conditions. During this time We ll Marsh showed a trend toward improved wetland health compared with baseline 1977 conditions (Rochow, 1998). Observations through 2006 suggest a continuation of impr oved wetland health at Well Marsh although vegetational conditions still are not like they were in the 1970s.

PAGE 225

213 Appendix D: (Continued) References: Hancock, M.C., T. Rochow and J. Hood. 2005. Review of original wetland assessment procedure (WAP March 2000) an d test results of a proposed revision to the WAP, May 2004. Southwest Florida Water Manageme nt District, Brooksville, FL 121 pp. Hancock, M.C., T. Rochow, and J. Hood. 2005. Test results of a proposed revision to the Wetland Assessment Procedure (WAP), Oc tober 2004 and Development of the Final WAP Methodology Adopted in April 2005. Southwest Florida Water Management District, Brooksville, FL 147 pp. Lopez, M. 1980. Hydrobiological monitori ng of Morris Bridge Wellfield, Hillsborough County, Florida: 1978-1979 update. SWFWMD Environm ental Section Technical Report 1980-1. 68 pp. Lopez, M. 1983. Hydrobiological Mon itoring of Morris Bridge Well Field, Hillsborough County, Florida. A Review: 1977-1982. Environmental Section Technical Report 1983-5. 96 pp. Mumme, R.L. 1978. Hydrobiological m onitoring of Morris Bridge Wellfield, Hillsborough County, Florida. SWFWMD Environmental Section Technical Report 1978-3. 42 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 226

214 Appendix D: (Continued) Uid 388 Green Swamp Dome #2 Green Swamp Dome #2 along with five ot her Green Swamp domes was selected for monitoring in 1979 (S. 18, T. 24, R. 24). A staff gauge was in stalled and hydrologic monitoring began in May 1979. A stilling well was installed not too long after staff reading began. The depth of the stilling well is not available a nd a somewhat sketchy historical record indicates that the stilling well may have been deepened one or more times over the lengthy period of observation. A two-inch upland surficial well was installed in 1999 and a six-inch wetland surficial well in 200 1. Instrumentation on the stilling well was moved to the wetland surficial well after it was drilled. In 1981, three meter-square vegetational plots (A-C) were installed in vegetational zones from the center of dome (A) to the edge (C ). The graphical plots are in a file accompanying the history. The plots were in itially sampled for percent plant species coverage in May, 1981 and November, 1981 --at a later date sampling was changed to once per year in May-June. Yearly quant itative vegetational sampling continued until June, 2002. A report on 1979-1982 monitoring was completed in 1984. Many photographs were taken over the years of the exte rior and interior of the dome. Some of the photos are shown in the photo-file that accompanies this history. Hydrologic conditions over the years in the dome have been close to what is normally reported in the literature for isolated cypre ss domes. Dome #2 at the present time is in good condition. References: Berryman & Henigar, Inc. 2005. Vertical di stribution of vegetation species relative to normal pool elevations in ten isolated we tlands in the Northern Tampa Bay area. Prepared for Tampa Bay Wa ter, Clearwater FL Rochow, T.F. and M. Lopez. 1984. Hydrobiological monitoring of cypress domes in the Green Swamp area of Lake and Sumter c ounties, Florida 1979-1982. Environmental Section Technical Report 1984-1. Southwest Fl orida Water Management District. 79 pp.

PAGE 227

215 Appendix D: (Continued) Uid 407 Morris Bridge X-3 Marsh The Morris Bridge X-3 Marsh was added to mo nitoring program at the wellfield in 1985 with the installation of a staff gage. The m onitoring program at Morris Bridge began in the 1970s but the need was seen in the 1980s to expand geographic coverage of the program and hence several other wetlands were added in the 1980s. A shallow upland well was installed for hydrologic monitoring in 1999 and a shallow wetland well next to the staff in 2000. Hydrologic information is pa rt of the District's Hydrologic Data Base (HDB). Unlike wetlands selected for monitoring in the 1970s, the X-3 marsh was not monitored quantitatively using plots and transects. It has also not been mentioned in several comprehensive technical reports on Morris Bridge monitoring since the last such monitoring report on the wellfield in 1983. Biological informa tion on the dome has relied on photography, observations, and re peat site visitations from 1985 to 2005. Photographs of wetland conditions are include d in the photo-file that accompanies the history. Since 2000 the Wetland Assessment Procedure (WAP) has been used at the wetland to assess environmental conditions. Relatively high pumpage rates at Morris Bri dge in the early 1980s likely caused a drying of the marsh in the mid to late 1980s (Rochow 1998). In more recent years water levels appeared to have improved. References: Hancock, M.C., T. Rochow and J. Hood. 2005. Review of original wetland assessment procedure (WAP March 2000) an d test results of a proposed revision to the WAP, May 2004. Southwest Florida Water Manageme nt District, Brooksville, FL 121 pp. Lopez, M. 1983. Hydrobiological Mon itoring of Morris Bridge Well Field, Hillsborough County, Florida. A Review: 1977-1982. Environmental Section Technical Report 1983-5. 96 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 228

216 Appendix D: (Continued) Uid 489 J.B. Starkey #3 The J.B. Starkey #3 cypress dome (S. 27 T. 26, R. 17) was selected for monitoring in 1989 and a staff gauge placed in the dome at this time. A 2-inch shallow wetland well and a 2-inch shallow upland well were added in 2001. The #3 dome is surrounded by improved pasture that is heavily fertilized at times. Water levels are part of the District 's Hydrologic Data Base (HDB). J.B. Starkey dome #3 has predominantly been a hydrologic monitori ng site with ecologic monitoring of secondary importance until WAP monitoring began in 2000. Nevertheless, observations of conditions were made period ically and notes taken during the 1990s. Within a short time after the monitoring site was selected it became apparent that environmental conditions were deteriorating ra pidly in the dome. This is evident from the photographs taken during th e 1990s displayed in the photo-file accompanying the history. A large number of leaning cypress are shown in the photographs. Soil subsidence of 1-2 feet exposing cypress root s can be seen. Inspection of conditions around the dome did not reveal reasons for the deterioration in ecology of the dome. From 2000 to 2006 the J.B. Starkey #3 dome was monitored with the WAP. Extensive leaning of cypress and subsidence of soil around the bases of cypress continued to be noted during this time. Based on these imp acts the dome is regard ed as being in poor condition.

PAGE 229

217 Appendix D: (Continued) Uid 493 EWWF #1 History of the EWWF #1 cypress dome located just south of the wellf ield road and east of the water treatment plant at the Eldridge -Wilde Wellfield extends back to the 1970s (Courser 1972, 1973). In these reports th e dome was called the "0" Cypress Head. Notes provided indicate that the muck was pa rtially oxidized, there were scattered ferns, and some alligator weed. The EWWF #1 dome was visi ted a number of times during the period from 1982 to 1994 partly to collect environmental information that was provided for SWFWMD's consumptive use evidentiaries (CUP 202673) The dome was called TR-PMD #1 during this period of field evaluation. Soil subsid ence up to ft was noted with some cypress root exposure. The dome was observed to be dr ier than expected at visitation times. The possibility that the dome may at times receive water from augmented pasture areas was noted. More fallen cypress than normal were observed. A staff gauge was installed in the wetland in 1989. Wetland and upland shallow wells were installed in 1999 and 2001 respectively. The hydrological records are part of the WMDB. Three extensive examinations of aerial photographic sequences over Eldridge-Wilde wetlands have been performed (SWFWMD, 1982b; Rochow and Rhinesmith, 1991; and Rochow, 1998). The examinations covered year s prior to wellfield pr oduction as well as following wellfield production. Examinations generally showed that over the wellfield impacts were readily evident in the form of leaning and falling cypress trees within approximately 15 years following the init iation of wellfield production in 1956. Observations of EWWF #1 were continued and increased as part of the Northern Tampa Bay Water Resources Assessment Project (1996 ). Notes show the dome was visited and conditions observed on forty visitations from 1989 to 1999. Soil subsidence and treefall were noted a number of times during these site visitations. From 2000 to 2006 conditions at EWWF #1 were monitored once or twice yearly using WAP (Wetland Assessment Procedure) methodology. Greater than expected treefall and considerable soil oxidation were observed in the dome although th e canopy overall was in relatively good condition compared to ot her Eldridge-Wilde cypress domes. Cattle trampling of the understory was noted.

PAGE 230

218 Appendix D: (Continued) References: Courser, W.D. 1972. Investigations of the effect of Pinellas County Eldridge-Wilde Wellfield's aquifer cone of depression on cypress head water levels and associated vegetation. Southwest Florida Water Management District memorandum. July 14, 1972. Courser, W.D. 1973. Investigations of the effect of Pinellas County Eldridge-Wilde Wellfield's aquifer cone of depression on cypress pond water levels and associated vegetation --1973. Southwest Florida Wa ter Management District memorandum. October 10, 1973. Rochow, T.F. 1988. Eldridge-Wilde Well Field (CUP 202673) environmental evaluation. Southwest Florida Water Management District memorandum. August 24, 1988. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp. Rochow, T.F. 1998. Investigation of hi storic aerial photography in and around the Eldridge-Wilde Wellfield. SWFW MD Memorandum. August 31, 1998. Rochow, T.F. and P. Rhinesmith. 1991. Comp arative analysis of biological conditions in five cypress dome wetlands at the Starkey and Eldridge-Wilde well fields in southwest Florida. SWFWMD Environmental Section Technical Report 1991-1. 67 pp. Southwest Florida Water Management District 1982a. Evidentiary evaluation, CUP No. 202673, Eldridge-Wilde Wellfield, Renewal. February 24, 1982. Southwest Florida Water Management District. 1982b. Historic impact on wetlands within the Eldridge-Wilde We llfield, Work Order Number 238. April 13, 1982. Memorandum by Rock G. Taber. Southwest Florida Water Management District 1989. Staff Report. Consumptive Use Permit Application No. 202673.02, Eldridge-Wilde Wellfield, May 25, 1989. Southwest Florida Water Management Dist rict. 1996. Northern Tampa Bay Water Resources Assessment Project. Volume one. Surface-Water/Ground-Water Interrelationships. Resource Evaluation Section, SWFWMD

PAGE 231

219 Appendix D: (Continued) Uid 501 STWF South Central The Starkey South Central cypress marsh (S. 8, T. 26, R. 17) near the power lines in the central part of the Starkey Wellfield was first observed and notes recorded in 1984. A staff gauge was installed in 1986 and a reco rding gauge by 1987. A shallow upland well was added in 1999 and a shallow wetland well in 2001. The recording gauge was removed at a later time and water levels cu rrently come from the staff. Tampa Bay Water consultants monitor another part of the wetland as S-85. Observations and photographs in the mid-1980s show giant and lesser pipeworts in wet meadow edge. Water levels in the wetland ap peared normal compared to similar control wetlands outside the wellfield. By the late 1980s water levels appeared abnormally low in the wetland. Soil fissures were eviden t and young cypress were becoming established at the edge of the central ope n-water area. Signs of wate r level depression began to become evident at approximately the time water production was beginning in the central wellfield area and area along the power lines. Photographs in the photo-file a nd observations show that within about ten years of initial observation leaning and fallen cypress were evident along with considerable soil subsidence and cypress root exposure. Sand pine and wax myrtle invasion into edge cypress were noted and photographed. Fennel (Eupatorium spp.) and bluestem (Andropogon spp.) were at times much more a bundant in the wetland than expected. At least twenty visits to the wetland were made during this time to observe vegetational conditions. Starting in 2000, the Starkey South Cypress Marsh was monito red with the Wetland Assessment Procedure (WAP). General observa tions show that health of this cypress marsh has continued to be poor. Starting in 2005, the Starkey South Cypress Marsh was monitored with a WAP transect at Tampa Bay Water's S-85 site. WAP field data sheets, spreadsheets, and photographs should be consulted for environmen tal conditions at this wetland. Reference: Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 232

220 Appendix D: (Continued) Uid 505 Morris Bridge X-2 Dome The Morris Bridge X-2 Dome was added to mo nitoring program at the wellfield in 1985 with the installation of a staff gage. The m onitoring program at Morris Bridge began in the 1970s but the need was seen in the 1980s to expand geographic coverage of the program and hence several other wetlands were added in the 1980s. A shallow upland well was installed for hydrologic monitoring in 1999 and a shallow wetland well next to the staff in 2001. Hydrologic information is pa rt of the District's Hydrologic Data Base (HDB). The X-2 dome is also monitored by Biological Research Associates (BRA) as MBR-14 for Tampa Bay Water and therefore there are two sources of biological and hydrologic information for this dome. Unlike wetlands selected for monitoring in the 1970s, the X-2 dome was not monitored quantitatively using plots and transects. It has also not been mentioned in several comprehensive technical reports on Morris Bridge monitoring since the last such monitoring report on the wellfield in 1983. Biological informa tion on the dome has relied on photography, observations, and re peat site visitations from 1985 to 2000. Photographs of wetland conditions are included in the photo-file that accompanies this history. Since 2000 the Wetland Assessment Procedure (WAP) has been used at the wetland to assess environmental conditions. Relatively high pumpage rates at Morris Bri dge in the early 1980s likely caused a drying of this dome in the mid to late 1980s (R ochow, 1998). The dome supported understory weeds such as fennel (Eupatorium spp.) during th is time. There also appears to be some excess treefall in the dome. In recent years water levels in the dome appear to have improved. References: Lopez, M. 1983. Hydrobiological Mon itoring of Morris Bridge Well Field, Hillsborough County, Florida. A Review: 1977-1982. Environmental Section Technical Report 1983-5. 96 pp. Rochow, T.F. 1998. The effects of water table level changes on fresh-water marsh and cypress wetlands. SWFWMD Environmental Section Technical Report 1998-1. 64 pp.

PAGE 233

221 Appendix D: (Continued) Uid 541 Green Swamp Dome #5 Green Swamp Dome #5 along with five ot her Green Swamp domes was selected for monitoring in 1979 (S. 12, T. 24, R. 23). A staff gauge was in stalled and hydrologic monitoring began in May 1979. A stilling well was installed not too long after staff reading began. The depth of the stilling well is not available a nd a somewhat sketchy historical record indicates that the stilling well may have been deepened one or more times over the lengthy period of observation. A two-inch upland surficial well was installed in 1999 and a six-inch wetland surficial well in 200 1. Instrumentation on the stilling well was moved to the wetland surficial well after it was drilled. Five meter-square vegetational plots (A-E) were installed in vegetational zones from the center of dome (A) to the edge (E). The gr aphical plots are in a file accompanying the history. The plots were initi ally sampled for percent plant species coverage in May, 1981 and October, 1981 --at a later date sampling was changed to once per year in MayJune. Yearly quantitative vegetational sampling continued until June, 2002. A report on 1979-1982 monitoring was completed in 1984. Many photographs were taken over the years of the exterior and inte rior of the dome. Some of the photos are shown in the photo-file that accompanies the history. Starting in 2000 Green Swamp Dome #5 wa s assessed each year with the Wetland Assessment Procedure (WAP). In 2000 the USGS began a study of the bathymetry and vegetation in Northern Tampa Bay marsh and cypress wetlands. Dome #5 was one of the wetlands studied intensively (Haag et al., 2005). Monitoring plots we re set up at various elevations in the dome. Ve getational monitoring and bathym etry are reported in the publication. Based on all evidence, hydrologic conditions ov er the years in the dome have been close to what is normally reported in the litera ture for isolated cypress domes. Dome #5 has remained in good condition over the period of monitoring. References: Haag, K.H., Lee T.M., and Herndon, D.C. 2005. Bathymetry and vegetation in isolated marsh and cypress wetlands in the Northern Tampa Bay Area, 2000-2004. U.S. Geological Survey Scientific Inves tigations Report 2005-5109. 49 pp. Rochow, T.F. and M. Lopez. 1984. Hydrobiological monitoring of cypress domes in the Green Swamp area of Lake and Sumter c ounties, Florida 1979-1982. Environmental Section Technical Report 1984-1. Southwest Fl orida Water Management District. 79 pp.

PAGE 234

222 Appendix D: (Continued) Uid 544 STWF "GG" The STWF "GG" dome lies in S. 13, T. 26, R. 17 south of the Starkey Wellfield and the Anclote River and just west of the Suncoast Parkway. The wetland at various times has been surrounded by a combination of overgrown flatwoods and pasture. In recent years the Suncoast Corridor parkway was built to th e east of the dome. The dome lies about 1.5 miles from the easternmo st Starkey production well. A staff was installed in the dome in 1989 as part of the Northern Tampa Bay Water Resources Assessment Project (SWFWMD, 1996 ). A shallow upland well was installed in 2000 and a wetland well next to the staff in 2001. Hydrologic data are part of the Water Management Data Base. Quantitative meter-square monitoring has neve r been conducted in the "GG" dome but numerous observations with writ ten notes as well as photographs have been taken. The photographs are in the photographi c file that accompanies the hi story. Nearly forty site visitations with notes were recorded from 1989-1999. On site visitations the dome has often been reasonable well hydrated although notes reveal that water levels were often lowe r-than-expected when compared to control cypress dome wetlands. Pickerelweed has ge nerally occurred near the staff although at times fennel (Eupatorium spp.) has been observe d. The ecologic conditions in the dome are regarded as good although it is possible that some surface water depression has occurred at times. Lower-than-expected wate r levels sometimes lead to weediness in the dome. Since 2000, environmental conditions have been monitored with the WAP. Reference: Northern Tampa Bay Water Resources Assessm ent Project. 1996. Volume One. Surface-Water/Ground-Water Interrelationships. Resource Evaluation Section, Southwest Florida Water Manage ment District. March 1996

PAGE 235

223 Appendix D: (Continued) Uid 605 Green Swamp Marsh The Green Swamp Marsh is the only marsh the District currently monitors in the Green Swamp (S. 33, T. 24, R. 23). A 6-inch stilli ng well with recorder wa s installed in 1994. In 1999 a 2-inch upland suficial well was drilled --in 2001 a 6-inch wetland surficial was added. Instrumentation was moved to the wetland well after it was drilled. In 1995, six meter-square monitoring plots were installed at 30 meter intervals from the wetland edge to the deep-water center of the marsh with the "0" plot closest to the wetland edge. The plots were first sampled for species cover in May, 1995 and then yearly through 2002. The last meter-square sampling was performed in July, 2006. Graphical results of vegetational monitoring ar e in the wetland histor y file. Photographs taken during ten years of monitoring are in the photo-file accompanying the history. Since year 2000, the Green Swamp Mars h has been monitored with the Wetland Assessment Procedure (WAP). The Green Swamp Marsh has remained in good condition during the period of monitoring. However, from an examinati on of the vegetational plot data and photo record it is evident that there is more ma idencane than pickerelweed across the marsh center than in earlier years. Slash pines fr om saplings to moderately large trees occur through the WAP's Outer Deep Zone. An apparent borro w area is 100-200 feet from the marsh but is thought to have minimal effects on the hydrology of the marsh. Recently the USGS studied bathymetry and vegetation in the Green Swamp Marsh as part of a study on the effects of augmen tation on cypress and marsh wetlands in the Northern Tampa Bay area (Haag et al., 2005) Berryman & Henigar has studied the vertical distribution of vege tation species in the marsh. References: Berryman & Henigar, Inc. 2005. Vertical di stribution of vegetation species relative to normal pool elevations in ten isolated we tlands in the Northern Tampa Bay area. Prepared for Tampa Bay Wa ter, Clearwater FL Haag, K.H., Lee T.M., and Herndon, D.C. 2005. Bathymetry and vegetation in isolated marsh and cypress wetlands in the Northern Tampa Bay Area, 2000-2004. U.S. Geological Survey Scientific Inves tigations Report 2005-5109. 49 pp.

PAGE 236

224 Appendix D: (Continued) Uid 1316 Starkey Bay History SWFWMD's Starkey Bay monitoring station (S. 8, T. 26, R. 17) was started in 2000 with the installation of a staff gauge in the wetland. In 2000 a shallow upland well was installed and a year la ter a shallow wetland well next to the staff. Starkey Bay was known to SWFWMD's envi ronmental monitoring program for many years prior to setting up the monitoring station in the wetland. During the 1970s and 1980s, as monitoring was initiated in the central and eastern pa rts of the wellfield, the bay wetland was observed from the "Old Dade City Road", a sandy road that runs east-west across the wellfield. On occasion the wetland wa s entered with some difficulty due to a dense growth of bays. During the early ye ars the wetland was judge d too dense to locate a staff in the deep interior and therefore only occasional observations were made of the wetland up until 2000 when inclusion of a wide r diversity of wetland types became an objective of SWFWMD's monitoring program. Starkey Bay is a large wetland extending from S. 8, T. 26, R. 17 into S. 5, T. 26, R. 17. The southern part of the wetland where th e SWFWMD staff is located in the 1970s supported a dense healthy stand of bay (Gordonia lasianthus). It was believed that water levels in the wetland were supported by seepag e from the surrounding sand pine uplands. Observations in the 1980s indicate that fire impacted much of the wetland possibly due to dry conditions the wetland experienced. WA P monitoring was conducted in the area of the SWFWMD staff from 2000-2004. Photos in the photo-file and field data indicate an impacted wetland. Many bays had fallen, opening up the formerly dense canopy, and muscadine grape was expanding in treefall ar ea. Six inches of soil slump was noted. In 2005 WAP monitoring was moved from SWFW MD's staff location to the S-90 station of Tampa Bay Water in S. 5, T. 26, R. 17. The Starkey Bay wetland in this location can best be described as a cypress marsh. Si gns of impacts are also evident from 2005-2007 WAP monitoring at this location. Photos ta ken at the S-90 station appear in the S-90 photo-file. In the initial years of monitoring the bay wetland was assigned a uid of 1331 on WAP field sheets. In recent years the wetla nd was reassigned to uid 1316 on field sheets. 1317 Starkey Wet Prairie Starkey Wet Prairie is a small shallow wetla nd prairie in the north eastern part of the Starkey Wellfield (S. 1, T. 26, R. 17). The site was chosen for monitoring in 2000 and a staff gauge installed. An upland shallow well was drilled in 2 000 and a wetland shallow well next to the staff in 2001. The Wetland Assessment Procedure (WAP) ha s been used since Fall 2000 to monitor wetland conditions. Photographs appear in the photo-file accompanying the history.

PAGE 237

225 Appendix D: (Continued) Over the period of monitoring the wet prairie has appeared drier than normal. Broomsedge (Andropogon virginicus) a nd blue maidencane (Amphicarpum muhlenbergianum) have continually been pres ent in more interior wetland areas than expected. Sandweed (Hypericum fasciculatum) also seems to have encroached into the interior wetland area possib ly due to dry conditions. Overall, the wet prairie does not appear as healthy as similar wetlands in non-wellfield areas.

PAGE 238

226 Appendix D: (Continued) Uid 1319 New River Cypress New River Cypress lies several miles north of the Morris Bridge Wellfield and east of Morris Bridge Road (S. 12, T. 27, R. 20). A staff and shallow upland well were installed in the wetland in 2000 --a 6-inch shallow wetland well was added in 2001. The shallow wetland well is equipped with a recorder. Ecologic conditions at the wetla nd have been monitored for the last several years using WAP methodology. A photographic record accompan ies the history. The interior of the dome according to the 2006 WAP is sparsely vegetated with mostly chain fern. Ferns are often typical of heavily shaded domes. Sour paspalum (Paspalum conjugatum), common carpetgrass (Axonopus fissifolius), witchgrass (Dichanthelium sp.), and St. Andrew's-Cross (Hypericum hypericoides) were noted in the Transitional Zone at the edge of the dome. Fetterbush (Lyonia luci da) and wax myrtle (Myrica cerifera) are common within the dome with some sh rubs encroaching on the ground in the Transitional Zone. Buttonbush (Cephalanthus occidentalis) grows underneath the cypress canopy in the interior of the dome. The cypress canopy is healthy. The overall appearance of the New River Cypress dom e has not changed since monitoring was initiated. New River Cypress during the period of observation has been in good health. The cypress dome is well hydrated and water levels appear similar to other domes at some distance from groundwater pumping and development. Reference: Berryman & Henigar, Inc. 2005. Vertical di stribution of vegetation species relative to normal pool elevations in ten isolated we tlands in the Northern Tampa Bay area. Prepared for Tampa Bay Wa ter, Clearwater FL

PAGE 239

227 Appendix D: (Continued) Uid 1322 UHFDA Cypress #3 UHFDA Cypress #3 dome was added to SW FWMD's wetland monitoring network in 2000 with the installation of a staff gauge. The cypress dome is located in the Upper Hillsborough Flood Detention Area (S. 17, T. 26, R. 22). A surficial upland well was added in 2000 and a surficial wetland well near the staff in 2001. Hydrologic information is part of the District's Wate r Management Data Base (WMDB). Ecological information about the wetland has be en collected for the past several years with the Wetland Assessment Procedure (WAP). The most noteworthy ecologic observation about dome #3 is fall ing cypress which is clearly evident in photographs in the photo-file accompanying the history. Most cypress treefall occurred during the several years that the dome has been monito red. Compared to control cypress domes observations indicate that dome #3 experi ences greatly depressed water levels.

PAGE 240

228 Appendix D: (Continued) Uid 1323 UHFDA Cypress #2 The UHFDA Cypress #2 wetland (S. 17, T. 26, R. 22) was added to SWFWMD's wetland with the installation of a staff gauge in 2000. A shallow upland well was added in 2000 and a shallow wetland well next to the staff in 2001. The hydrologic record is part of the District's Water Manage ment Base (WMDB). Ecological information over the past several years has been collected with the Wetland Assessment Procedure (WAP) and is not de scribed in detail here. Cypress #2 has appeared abnormally dry whenever the dome has been visited in the last several years. At least one-half foot of soil subsidence is evident at the base of cypress causing root exposure in places. Considerable shrub and herbaceous understory invasion is evident. The wetland is close to a large borrow ar ea to the south which may be causing water levels to be depressed in the wetland. WA Ps should be consulted for more details on vegetational conditions.

PAGE 241

229 Appendix D: (Continued) Uid 1324 UHFDA Cypress #1 The UHFDA Cypress #1 wetland (S. 18, T. 26, R. 22) was added to SWFWMD's wetland monitoring network in 2000 with the installation of a sta ff gauge. A surficial upland monitoring well was added in 2000 and a shallow wetland well ne xt to the staff in 2001. The hydrologic record is part of the Dist rict's Water Management Base (WMDB). Ecological information over the past several years has been collected with the Wetland Assessment Procedure (WAP) and is not de scribed in detail here. Cypress #1 has appeared abnormally dry whenever the dome has been visited in the last several years. The canopy is stressed with dying cypress. WAPs should be consulted for further information on vegetational conditions.

PAGE 242

230 Appendix D: (Continued) Uid 1325 UHFDA Wet Prairie Ecologic monitoring of the Upper Hillsborough We t Prairie (S. 18, T. 26, R. 22) began with the installation of a staff gauge in 2000. A 2-inch upland surficial well was added in 2000 and a 6-inch wetland surficial well in 2001. The wetland well has a water level recorder. Water levels are pa rt of the District's Water Ma nagement Data Base (WMDB). Although considered a wet prairie base d on GIS mapping the wetland system is sufficiently deep to likely have been a marsh in the past. Ecologic information about the wet prai rie comes from the Wetland Assessment Procedure (WAP). The photographic record th at accompanies this history shows that at times the wet prairie has had dense dog fennel (Eupatorium capillifolium) in the central area near the recorder. Many slash pines (Pinus elliottii) have encroached into the wet prairie edge. Observations of water levels in the prairie indicate that surface waters are considerably lower than expected comp ared to control marsh-like wetlands.

PAGE 243

231 Appendix D: (Continued) Uid 1326 Alston Cypress #2 The Alston Cypress #2 dome lies on District-owned Alston lands (S. 22, T. 26, R. 22). A staff gauge and shallow upland well were in stalled in the wetland in 2000 --a shallow wetland well was added next to the staff in 2001. The shallow wetland well is equipped with a recorder. The water level record is pa rt of the District's Water Management Data Base. Ecologic conditions in the dome during the past several years have been monitored with the WAP (Wetland Assessment Procedure). A photographic record accompanies this history. It is evident from the photos that fireweed (Ere ctites hieraciifolius) and dog fennel (Eupatorium capillifolium) have been common at times in the dome. Water levels at times have appeared lower than expected.

PAGE 244

232 Appendix D: (Continued) Uid 1327 Alston Cypress #1 The Alston Cypress #1 dome lies on District-owned Alston lands (S. 34, T. 26, R. 22). A staff gauge and shallow upland well were in stalled in the wetland in 2000 --a shallow wetland well was added next to the staff in 2001. The water level record is part of the District's Water Management Data Base. Ecologic conditions in the dome during the past several years have been monitored with the WAP (Wetland Assessment Procedure). A photographic record accompanies this history. It is evident from the photos that fireweed (Ere ctites hieraciifolius) and dog fennel (Eupatorium capillifolium) have been common at times in the dome. Excess dead and dying cypress have been noted in the ca nopy and water levels have appeared lower than normal. Causes for low water levels in the dome are not known but a borrow area exists within a few hundred feet of the dome.

PAGE 245

233 Appendix D: (Continued) Uid 1329 Alston Wet Prairie The Alston Wet Prairie lies on District-owned Alston lands (S. 34, T. 26, R. 22). A staff gauge and shallow upland well were installed in the wetland in 2000 --a shallow wetland well was added next to the staff in 2001. The shallow wetland well is equipped with a recorder. The water level record is pa rt of the District's Water Management Data Base. Ecologic conditions in Alston Wet Prairie during the past several years have been monitored with the WAP (Wetland Assessmen t Procedure). A photographic record accompanies this history. 1332 STWF Wetland Coniferous Forest Monitoring of the Wetland Coniferous Forest (S. 17, T. 26S, R. 17) in the western part of the Starkey Wellfield along the power li nes was begun in 2000. The wetland was selected since it had a good re presentation of bay (Gordonia lasiathus ) which was not found in most other wetlands monitore d by SWFWMD. The we tland therefore was regarded as a worthwhile addition to SWFWMD's wetland monitoring network. A shallow upland well was added in 2000 and a sh allow wetland well next to the staff in 2001. Hydrological records are part of SWFWMDs Hydrologic Data Base. Biological data was collected with the Wetland Assessment Method (WAP) from 20002005. Up through 2004 the data was collected along a transect from the edge of the wetland to SWFWMD's staff. In Spring 2005 monitoring was moved to a nearby wetland monitoring station (S-112) in the same wetla nd being used for data collection by Tampa Bay Water's consultant. In the years from 2000-2004 SWFWMD noticed considerable treefal l and soil subsidence along the transect near the SWFWMD staff. WAP field sheets, spreadsheets, and photographs should be consulted for envir onmental conditions in this wetland.

PAGE 246

234 Appendix D: (Continued) Uid 1335 Green Swamp West Cypress The Green Swamp West Cypress dome was added to SWFWMD's wetland monitoring network in 2000 with the insta llation of a staff gauge (S. 2, T. 25, R. 22). A shallow upland well was added in 2000 along with a shallow wetland well in 2001. The West Cypress dome is typical of others in the Green Swamp but differs in the wide wet meadow that is part of the wetland assessment area. From 2000-2005 the West Cypress dome was monitored using WAP methodology. A br ief photo-file accompanies the history. The Green Swam p West Cypress dome is rega rded as generally healthy although the 2005 WAP indicates that ground-cove r plants such as Centella asiatica, Eleocharis baldwinii, and Eupatorium capillif olium have moved into the deep zone in small numbers. Similarly a few shrubs and sm all trees such as wax myrtle and slash pine were detected in 2005 in the deep zone. Vi sual hydrologic indications during relatively few site visitations seem to indicate that the hydrology of the West Cypress dome is typical of other Green Swamp domes. However, the hydrology needs further investigation considering some invasion of shallow-water species along the dome edge. Reference: Berryman & Henigar, Inc. 2005. Vertical di stribution of vegetation species relative to normal pool elevations in ten isolated we tlands in the Northern Tampa Bay area. Prepared for Tampa Bay Wa ter, Clearwater FL

PAGE 247

235 Appendix D: (Continued) Uid 1337 Green Swamp Bay A staff gauge was placed in the Green Swamp Bay (S. 6, T. 24, R. 24) in 2000 along with a shallow upland well. A shallow wetland we ll with a recorder was added in 2001. A considerable amount of Loblolly Bay (Gordonia lasian thus) occurs at th e edge of the bay stand. Water level data are part of the District's H ydrologic Data Base. Monitoring of the Green Swamp Bay during th e past several years has taken place using the WAP. Pine invasion at the edge of the bay stand as well as soil slumping between trees are conspicuous. Water levels apparently have not been reaching historic levels in the bay stand --soil slumping between trees may be one explanation for depressed water levels at the edge of the bay. Further comments on the Green Swamp Bay can be found on the field assessment sheets. Uid 1344 UHFDA Cypress #4 UHFDA Cypress #4 dome was added to SW FWMD's wetland monitoring network in 2001 with the installation of a staff gauge. The cypress dome in located in the Upper Hillsborough area (S. 28, T. 25, R. 22) betw een Berry Road (35A) and U.S. 98. A surficial wetland well was added near the sta ff in 2001 and a surficial upland well also in 2001. Hydrologic information is part of th e District's Water Management Data Base (WMDB). Ecological information about the wetland has be en collected for the past several years with the Wetland Assessment Procedure (WAP). The reader is referred to the WAPs for this information.

PAGE 248

236 Appendix D: (Continued) Uid 3713 Cypress Creek ELAPP Cypress The Cypress Creek ELAPP Cypress was adde d to the District's wetland monitoring network in 2002 with the insta llation of a staff gauge (S. 15, T. 27, R. 19). The wetland is on Hillsborough County ELAPP land --the Dist rict has a license agreement with the County to allow access to the wetland. Sha llow wetland and upland wells were added to the wetland in 2001. The wetland well is 6inches in diameter and has a recorder. Cypress Creek ELAPP Cypress has been monitored with the Wetland Assessment Procedure (WAP) methodology for the past severa l years. The wetland is typical of other cypress domes in the Northern Tampa Bay (N TB) area. The deep area is occupied by moderate amounts of Walter' s Sedge (Carex striata) and maidencane (Panicum hemitomon). The dome fringe has Walter's Sedge, maidencane, taperleaf waterhoarhound (Lycopus rubellus), Virgin ia buttonweed (Diodia virginiana), and falsefennel (Eupatorium leptophyllum). The 20 05 WAP reports that s lightly greater than expected amounts of buttonweed, falsefenne l, waterhoarhound and rosy camphorweed (Pluchea rosea) were found in the deep wetland zone. Such findings are not uncommon in naturally occurring wetlands in the NTB area. An indication of vegetational conditions in the cypress dome is seen in the photo-file. Cypress conditions during the period of monitoring are regarded as healthy.

PAGE 249

237 Appendix D: (Continued) Uid 3715 Cypress Creek ELAPP Marsh The Cypress Creek ELAPP Marsh was added to the District's wetland monitoring network in 2002 with the insta llation of a staff gauge (S. 15, T. 27, R. 19). The wetland is on Hillsborough County ELAPP land --the Dist rict has a license agreement with the County to allow access to the wetland. Sha llow wetland and upland wells were added to the wetland in 2001. The wetland well is 6inches in diameter and has a recorder. Cypress Creek ELAPP Marsh has been m onitored with the Wetland Assessment Procedure (WAP) methodology for the past seve ral years. The wetland is typical of other marshes in the Northern Tampa Ba y area. The deep area is occupied by maidencane (Panicum hemitomon), picker elweed (Pontederia cordata), swamp smartweed (Polygonun hydropiperoides) and minor amounts of sawgrass (Cladium jamaicense). Blue maidencane (Amphicarpum muhlenbergianum) a nd moderate amounts of falsefennel (Eupatorium leptophyllum), broomsedge (Andropogon virginicus), and sugarcane plumegrass (Saccharum giganteum) are found toward the marsh fringe. An indication of vegetational condi tions in the marsh can be seen in the photo-file. Marsh conditions during the period of m onitoring have been healthy.

PAGE 250

238 Appendix D: (Continued) Uid 4184 Cone Ranch (CR-3) Cypress The Cone Ranch (CR-3) Cypress wetland (S. 28, T.27, R. 22) was selected as a Minimum Level Wetland by District Governing Board acti on in 1998 (Chapter 40D-8, F.A.C.). As with other MFL wetlands, a Minimum Le vel in feet NGVD was specified for the wetland. The CR-3 wetland as with other MFL wetlands at Cone Ranch was part of Tampa Bay Water's wetland monitoring netw ork on the Ranch and hence historical information exists on water levels and environmental conditions in the wetland. The District upgraded water le vel installations at CR-3 in 2003 with the installation of a District staff and surficial wetland and upl and wells. Wetland Assessment Procedure (WAP) evaluations have been conducted at the wetland for the past several years. Photographs of the wetland are included in th e photo-file that accompanies the historical description. WAP assessment notes indicate pasture at the edge of the dome as well as cattle and hog signs within the dome. The cypress canopy is in good condition. Soils in the dome are mucky.

PAGE 251

239 Appendix D: (Continued) Uid 4187 Cone Ranch (CR-6) Cypress The Cone Ranch (CR-6) cypress wetland (S. 22, T.27, R. 22) was se lected as a Minimum Level Wetland by District Governing Board acti on in 1998 (Chapter 40D-8, F.A.C.). As with other MFL wetlands, a Minimum Le vel in feet NGVD was specified for the wetland. The CR-6 wetland as with other MFL wetlands at Cone Ranch was part of Tampa Bay Water's wetland monitoring netw ork on the Ranch and hence historical information exists on water levels and environmental conditions in the wetland. The District upgraded water le vel installations at CR-6 in 2003 with the installation of a District staff and surficial upland and we tland wells. Wetland Assessment Procedure (WAP) evaluations have been conducted at the wetland for the past several years. Photographs of the wetland are included in th e photo-file that accompanies the historical description. Recent WAP assessments have noted bahia grass (Paspalum notatum) at the edge of the dome and some dog fennel (Eupatorium cap illifolium) in the Deep Zone. Cattle trampling has been extensive at the dome edge. Observations have shown some excess cypress canopy stress along with noticeable dead and leaning cypress in the dome.

PAGE 252

ABOUT THE AUTHOR Kenneth Allan Nilsson received a Bachelors Degree in Civil Engineering from the University of Cincinnati in 2002 and a Masters of Science in Environmental Engineering from the University of Cincinnati in 2004. He entere d the Ph.D. program at the University of South Florida in the fall of 2004. While at the University of South Florida he became very involved with the Florid a Section of the American Water Works Association, an international society dedicated to the im provement of drinking water quality and supply.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003458
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Nilsson, Kenneth.
0 245
Improved methodologies for modeling storage and water level behavior in wetlands
h [electronic resource] /
by Kenneth Nilsson.
260
[Tampa, Fla] :
b University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: Wetlands are important elements of watersheds that influence water storage, surface water runoff, groundwater recharge/discharge processes, and evapotranspiration. To understand the cumulative effect wetlands have on a watershed, one must have a good understanding of the water-level fluctuations and the storage characteristics associated with multiple wetlands across a region. An improved analytical method is presented to describe the storage characteristics of wetlands in the absence of detailed hydrologic and bathymetric data. Also, a probabilistic approach based on frequency analysis is developed to provide insight into surface and groundwater interactions associated with isolated wetlands. The results of the work include: 1) a power-function model based on a single fitting parameter and two physically based parameters was developed and used to represent the storage of singular or multiple wetlands and lakes with acceptable error, 2) a novel hydrologic characterization applied to 56 wetlands in west-central Florida provided new information about wetland hydroperiods which indicated standing water was present in the wetlands 62% of the time and these wetlands were groundwater recharge zones 59% of the time over the seven year study, 3) the smallest extreme value probability distribution function was identified as the best-fit model to represent the water levels of five wetland categories in west-central Florida, 4) representative probability models were developed and used to predict the water levels of specific wetland categories, averaging less than 10% error between the predicted and recorded water levels, and 5) last, based on this probability analysis, the various wetland categories were shown to exhibit similar means, extremes and ranges in water-level behavior but unique slopes in frequency distributions, a here to for new finding. These results suggest that wetland types may best be differentiated by the regular variability in water levels, not by the mean and/or extreme water levels. The methods and analytical techniques presented in this dissertation can be used to help understand and quantify wetland hydrology in different climatological or anthropogenic stress conditions. Also, the methods explored in this study can be used to develop more accurate and representative hydrologic simulation models.
590
Advisor: Mark A. Ross, Ph.D.
653
Analytical techniques
Bathymetry
Frequency analysis
Hydrologic models
Water storage
Wetlands
690
Dissertations, Academic
z USF
x Civil & Environmental Engineering
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.3458