USF Libraries
USF Digital Collections

Adapting the SCS method for estimating runoff in shallow water table environments

MISSING IMAGE

Material Information

Title:
Adapting the SCS method for estimating runoff in shallow water table environments
Physical Description:
Book
Language:
English
Creator:
Masek, Caroline Humphrey
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Runoff -- Measurement   ( lcsh )
Water table   ( lcsh )
air encapsulation
variable source area
soil storage
saturation excess
Dissertations, Academic -- Civil Engineering -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Rainfall-runoff modeling in the United States has made extensive use of the Soil Conservation Service (SCS) curve number method for computing infiltration losses from rainfall. Even though the method is well established and may be applied to a wide range of environments, it often results in highly erroneous runoff estimates for shallow water table environments. Flat topography, wetlands, and fine sands are characteristics that make places like Florida very different from the environments where the SCS method was originally developed. The SCS method arose from experiments with soils that are dominated by infiltration excess (Hortonian mechanism), where runoff occurs after rainfall intensity exceeds the infiltration capacity of the soil. In contrast, Florida is likely dominated by saturation excess runoff (Dunne mechanism), where the soil storage capacity between a shallow water table and the ground surface is filled, and all remaining rainfall becomes runoff. The sandy^soils of Florida have very high infiltration capacities, and thus infiltration excess is less likely than saturation excess. As a consequence of the saturation-excess mechanism, wetlands expand in the wet season as the soil moisture storage around the perimeter is filled.A modified form of the SCS method is proposed with the objective that it is more suitable than the current method in flatly sloped, humid environments. Initial conditions, such as the pre-storm soil moisture profile and depth to water table, are critical when predicting runoff in these areas. Air encapsulation is addressed because its presence causes the soil storage capacity to be filled significantly faster than in its absence. Equations are presented that provide an estimate of the average depth to water table and average soil storage capacity in a catchment.Two Florida catchments and one runoff test bed were selected for testing the new methodology. The runoff test bed demonstrated the saturation-excess mechanism while the catchments provided larger-scale testing of the method. Though more data is needed to fully assess the performance of the method, the approach offers a more plausible mechanism for runoff estimation in shallow water table environments with sandy soils.
Thesis:
Thesis (M.S.)--University of South Florida, 2002.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Caroline Humphrey Masek .
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 120 pages.

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:
aleph - 001413343
oclc - 51164763
notis - AJJ0759
usfldc doi - E14-SFE0000040
usfldc handle - e14.40
System ID:
SFS0024736:00001


This item is only available as the following downloads:


Full Text

PAGE 1

ADAPTING THE SCS METHOD FOR ESTIMATING RUNOFF IN SHALLOW WATER TABLE ENVIRONMENTS by CAROLINE HUMPHREY MASEK A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Civil Engineering Department o f Civil and Environmental Engineering College of Engineering University of South Florida October 4, 2002 Major Professor: Mahmood Nachabe, Ph.D. Keywords: Air Encapsulation, Saturation Excess, Soil Storage, Variable Source Area

PAGE 2

DEDICATION Th is work is dedicated to my parents, Dr. and Mrs. Donald Humphrey. Thank you for all of the support that you have given me in my academic and professional pursuits.

PAGE 3

ACKNOWLEDGEMENTS I would like to thank Patrick Tara and Renee (Rokicki) Murch for their f requent assistance in this research, and for their humor and advice. I extend special gratitude to Jeffrey Vomacka and Don Thompson, who worked diligently to collect and process much of the data I needed. I also appreciate the work of Xiongfei Xie, who p rovided me with necessary laboratory data. Thank you to Dr. Mahmood Nachabe for his constant guidance and commitment to quality research. Thank you also to Dr. Mark Ross and Dr. Jeffrey Geurink for meeting with me and providing helpful insights. Finally, I would like to acknowledge Dr. Jayantha Obeysekera for his support and his willing ness to help with this research which was funded by C ontract C 13382 with the Sou th Florida Water Management District.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ............. iii LIST OF FIGURES ................................ ................................ ................................ ............. v ABSTRACT ................................ ................................ ................................ ..................... viii CHAPTER 1. INTRODUCTION ................................ ................................ ....................... 1 1.1 Background ................................ ................................ ................................ ............... 1 1.2 Objectives and Scope ................................ ................................ ................................ 2 1.3 Need for an Improved Runoff Estimation Method ................................ ................... 3 1.4 Review of Existin g Literature ................................ ................................ ................... 7 1.4.1 Development of the SCS Method ................................ ................................ .... 7 1.4.2 Runoff Mechanisms ................................ ................................ ....................... 10 1.4.3 Runoff Studies in Florida ................................ ................................ ............... 12 CHAPTER 2. MATERIALS AND METHODOLOGY ................................ ................... 14 2.1 Data Collec tion ................................ ................................ ................................ ....... 14 2.1.1 Runoff Test Bed ................................ ................................ ............................. 15 2.1.2 Long Flat Creek ................................ ................................ ............................. 20 2.1.3 West Fork Horse Creek ................................ ................................ .................. 23 2.2 Methodology ................................ ................................ ................................ ........... 26 2.2.1 The Bucket Model ................................ ................................ .......................... 26 2.2.2 The Soil Storage Capacity Equatio n ................................ .............................. 32 CHAPTER 3. THE SCS SFWMD METHOD ................................ ................................ 35 3.1 Soil Storage Capacity ................................ ................................ .............................. 35 3.1.1 Soil Characterization ................................ ................................ ...................... 35 3.1.2 Using the Soil Storage Capacity Equation ................................ ..................... 36 3.1.3 Air Encapsulation ................................ ................................ ........................... 3 7 3.1.4 Soil Storage Capacity for Three Layered Soils ................................ .............. 46 3.2 Initial Conditions ................................ ................................ ................................ .... 53 3.2.1 Depth to the Water Table ................................ ................................ ............... 54 3.2.2 Baseflow ................................ ................................ ................................ ......... 61 3.3 Runoff Estimation ................................ ................................ ................................ ... 64 3.3.1 Hydrograph Sep aration ................................ ................................ ................ 64

PAGE 5

ii 3.3.2 Surface Storage ................................ ................................ .............................. 66 CHAPTER 4. APPLICATION OF THE SCS SFWMD METHOD ................................ 69 4.1 Introduction ................................ ................................ ................................ ............. 69 4.2 Runoff T est Bed ................................ ................................ ................................ ...... 69 4.2.1 Application of the Method ................................ ................................ ............. 69 4.2.2 Results ................................ ................................ ................................ ............ 71 4.3 Long Flat Creek ................................ ................................ ................................ ...... 75 4.3.1 Application of the Method ................................ ................................ ............. 75 4.3.2 Results ................................ ................................ ................................ ............ 8 4 4.4 West Fork Horse Creek ................................ ................................ ........................... 88 4.4.1 Application of the Method ................................ ................................ ............. 89 4.4.2 Results ................................ ................................ ................................ ............ 90 CHAPTER 5. CONCLUSIONS ................................ ................................ ....................... 95 5.1 Factors Affecting Performance ................................ ................................ ............... 95 5.2 Summary and Conclusion ................................ ................................ ....................... 9 9 REFERENCES ................................ ................................ ................................ ............... 102 APPENDICIES ................................ ................................ ................................ ............... 108 Appendix A: Soils Data ................................ ................................ .............................. 109 Appendix B: Long Flat Creek Data ................................ ................................ ............ 114 Appendix C: Brooks and Corey Model of Soil Water Retention ............................... 118

PAGE 6

iii LIST OF TABLES Table 1. Encapsulated Air Volume as a Function of Depth ................................ ............ 40 Table 2. Average Air Encapsulation in the Runoff Test Bed ................................ .......... 45 Table 3. Average Air Encapsulation in the Long Flat Creek Subbasin ........................... 45 Table 4. Soil Storage Calculation for Immokalee Fine Sand ................................ ........... 49 T able 5. Soil Storage for Three Soils as a Function of Water Table Depth .................... 50 Table 6. Results for the Runoff Test Bed ................................ ................................ ........ 74 Table 7. Area Weights and Soil Moisture Storage at Long Flat Creek for 6/25/02 ........ 79 Table 8. SCS SFWMD Method Parameters for June 25, 2 002 at Long Flat Creek ........ 83 Table 9. Results for the Long Flat Creek Subbasin ................................ ......................... 88 Table 10. Soil Storage at West Fork Horse Creek ................................ ........................... 91 Table 11. Results for West Fork Horse Creek ................................ ................................ 91 Table 12. Soil Properties of Myakka Fine Sand ................................ ............................ 109 Table 13. Soil Propertie s of Immokalee Fine Sand ................................ ....................... 109 Table 14. Soil Properties of Smyrna Sand ................................ ................................ ..... 110 Table 15. Water Content for Selected Capillary Pressures for Myakka Fine Sand ....... 110 Table 16. Water Content for Selected Capillary Pressures for Immokalee Fine Sand .. 111 Table 17. W ater Content for Selected Capillary Pressures for Smyrna Sand ................ 111 Table 18. Brooks and Corey Model Parameters ................................ ............................ 112 Table 19. Soil Moisture Capacity Equation Regression Constants for Units in Inches 112 Table 20. Saturated Hydraulic Conductivity Data ................................ ......................... 113

PAGE 7

iv Table 21. Soil Moisture Probe Water Content Before Storm on June 25, 2002 ............ 114 Table 22. Laboratory Soil Texture and Porosity Data for Long Flat Creek .................. 115 Table 23. Air Encapsulation Data for the Runoff Test Bed ................................ ........... 116 Table 24. Air Encapsulation Data for Long Flat Creek ................................ ................. 117 Table 25. Pre Storm Water Table Depths in the Long Flat Creek Subbasin ................. 117 Table 26. Brooks and Corey Model for Immokalee Fine Sand ................................ ..... 119

PAGE 8

v LIST OF FIGURES Figure 1. The SCS Runoff Equation Solved for Selected Curve Numbers ....................... 5 Figure 2. Proposed Factors Influencing Runoff Mechanisms ................................ ......... 11 Figure 3. Location of Selected Study Basins ................................ ................................ ... 14 Figure 4. Instrumentation at the Runoff Test Bed ................................ ........................... 16 Figure 5. Photograph of the Runoff Test Bed ................................ ................................ .. 17 Figure 6. Photographs of the Test Bed Weir Box ................................ ............................ 18 Figure 7. Photograph of a Soil Moisture Probe at Long Flat Creek ................................ 19 Figure 8. The Long Flat Creek Subbasin ................................ ................................ ......... 22 Figure 9. Photograph of a Complex V notch / Rectangular Weir ................................ ... 23 Figure 10. The West Fork Horse Creek Basin ................................ ................................ 25 Fig ure 11. Runoff Response in a Single Bucket Model for a Uniform Watershed ......... 28 Figure 12. Two Buckets Representing Two Partial Areas Comprising a Watershed ...... 29 Figure 13. Two Buckets Initiating Runoff at Different Times ................................ ........ 29 Figure 14. Soil Moisture Storage Above a Shallow Water Table ................................ .... 3 4 Figure 15. Air Encapsulation at a Submerged Soil Moisture Sensor .............................. 42 Figure 16. Soil Moisture Profile Near USF 1 for April 12, 2002 ................................ .... 43 Figure 17. Air Encapsulation as a Function of Pressure Head ................................ ........ 44 Figure 18. Average Air Encapsulation at the Test Bed and Long Flat Creek ................. 45 Figure 19. Soil Moisture Profile for Myakka Fine Sand ................................ ................. 47

PAGE 9

vi Figure 20. Soil Moisture Profile for Immokalee Fine Sand ................................ ............. 47 Figure 21. Soil Mois ture Profile for Smyrna Sand ................................ .......................... 48 Figure 22. Comparison of Three Soil Storage Calculations for Myakka Fine Sand ....... 51 Figure 23. Comparison of Three Soil Storage Calculations for Immokalee Fine S and .. 51 Figure 24. Comparison of Three Soil Storage Calculations for Smyrna Sand ................ 52 Figure 25. Soil Storage Results for Three Soils ................................ ............................... 52 Figure 26. Saturated Hy draulic Conductivity for Myakka Fine Sand ............................. 56 Figure 27. Small Water Table Gradient Near Long Flat Creek ................................ ....... 59 Figure 28. Depth to Water Table Distribution for West Fork Horse Creek .................... 61 Figure 29. Storm Separation Method for West Fork Horse Creek ................................ .. 65 Figure 30. Rating Curve for the Test Bed Weir ................................ ............................... 71 Figure 31. Rainfall and Runoff at the Test Bed for J une 25, 2002 ................................ .. 72 Figure 32. Soil Moisture and Water Table Data at USF 1 for June 25, 2002 ................. 73 Figure 33. Soil Moisture and Water Table Data at USF 3 for June 25, 2002 ................. 73 Figure 34. Water Table Gradient from PS 5 to PS 1 for June 25 26, 2002 .................... 77 Figure 35. Water Table Gradient from PS 39 to PS 43 for June 25 26, 2002 ................ 77 Figure 36. Delineation of Are as for Assignment of Soil Moisture Storage ..................... 78 Figure 37. Digital Elevation Model of the Long Flat Creek Subbasin ............................ 81 Figure 38. Water Table Distribution in the Long Flat Creek Subbasin ........................... 82 Figure 39. Rainfall and Runoff at Long Flat Creek for June 25, 2002 ............................ 84 Figure 40. Soil Moisture and Water Table Data at PS 43 for June 25, 2002 .................. 86 Figure 41. So il Moisture and Water Table Data at PS 42 for June 25, 2002 .................. 86 Figure 42. Soil Moisture and Water Table Data at PS 41 for June 25, 2002 .................. 87

PAGE 10

vii Figure 43. Soil Moisture and Water Table Data at PS 40 for June 25, 2002 .................. 87 Figure 44. Digital Elevation Model of West Fork Horse Creek ................................ ...... 90 Figure 45. Runoff Prediction for West Fork Horse Creek ................................ ............... 93 Figure 46. Brooks and Corey Model Fitting for Immokalee Fine Sand ........................ 120

PAGE 11

viii ADAPTING THE SCS METHOD FOR ESTIMATING RUNOFF IN SHALLOW WATER TABLE ENVIRONMENTS by CAROLINE HUMPHREY MASEK An Abstract of a thesis submitted in partial fulfillment of the req uirements for the degree of Master of Science in Civil Engineering Department of Civil and Environmental Engineering College of Engineering University of South Florida October 4, 2002 Major Professor: Mahmood Nachabe, Ph.D.

PAGE 12

ix Rainfall runoff modeling in the United States has made extensive use of the Soil Conservation Service (SCS) curve number method for computing infiltration losses from rainfall. Even though the method is well established and may be applied to a wide range of environments, it often results in highly erroneous runoff estimates for shallow water table environments. Flat topography, wetlands, and fine sands are characteristics that make places like Florida very different from the environments where the SCS method was original ly developed. The SCS method arose from experiments with soils that are dominated by infiltration excess (Hortonian mechanism), where runoff occurs after rainfall intensity exceeds the infiltration capacity of the soil. In contrast, Florida is likely dom inated by saturation excess runoff (Dunne mechanism), where the soil storage capacity between a shallow water table and the ground surface is filled, and all remaining rainfall becomes runoff. The sandy soils of Florida have very high infiltration capacit ies, and thus infiltration excess is less likely than saturation excess. As a consequence of the saturation excess mechanism, wetlands expand in the wet season as the soil moisture storage around the perimeter is filled. A modified form of the SCS meth od is proposed with the objective that it is more suitable than the current method in flatly sloped, humid environments. Initial conditions, such as the pre storm soil moisture profile and depth to water table, are critical when predicting runoff in these areas. Air encapsulation is addressed because its presence causes the soil storage capacity to be filled significantly faster than in its absence. Equations are presented that provide an estimate of the average depth to water table and average soil stor age capacity in a catchment.

PAGE 13

x Two Florida catchments and one runoff test bed were selected for testing the new methodology. The runoff test bed demonstrated the saturation excess mechanism while the catchments provided larger scale testing of the method. Though more data is needed to fully assess the performance of the method, the approach offers a more plausible mechanism for runoff estimation in shallow water table environments with sandy soils. Abstract Approved: _____________________________________ _____________ Major Professor: Mahmood Nachabe, Ph.D. Assistant Professor, Department of Civil and Environmental Engineering Date Approved: ___________________________________

PAGE 14

1 CHAPTER 1. INTRODUCTION 1 1 Background Rainfall runoff modeling in Flor ida has historically relied on the Soil Conservation Service (SCS) curve number method for computation of losses from rainfall, by using the method either in its original form or in a modified form designed for hydrologic models in agriculture. The shallo w water tables, highly conductive sandy soils, and flat topography of Florida are quite different from the hydrogeologic characteristics of the regions where the SCS method was developed. The methodology arose from experiments with soils that are dominate d by infiltration excess (Hortonian mechanism) instead of by saturation excess (Dunne mechanism). Observations of catchment response to wet season rainfall events in Florida suggest that runoff is generated when the soil storage between the ground surface and the capillary fringe above the water table is filled. Once the available soil storage space is occupied, stormwater collects at the ground surface and eventually travels as overland flow to a stream. Saturated soil conditions govern the production o f saturation excess whereas infiltration capacity controls infiltration excess. Hortonian overland flow is known to occur in semiarid regions and agricultural lands similar to those in the midwestern United States, where the infiltration capacity is usual ly less than the rainfall intensity (Dunne et al. 1975). The high hydraulic conductivity of uniform sand reduces the likelihood of the rainfall intensity exceeding infiltration capacity; therefore, the Hortonian mechanism is

PAGE 15

2 likely less prominent in Flori da. For example, the lower limit of the infiltration rate for infiltration excess is the saturated hydraulic conductivity of the soil. Laboratory data for widespread Florida sands, such as Myakka fine sand, show that typical vertical conductivities range between 10 and 40 cm/hr, with an average of about 18 cm/hr (Carlisle et al. 1989). These conductivity values were compared with one minute time resolution rainfall data recorded on top of one of the engineering buildings at the University of South Florid a, as documented by Hernandez (2000). The average rainfall intensity out of 316 convective storms over a four year period was 4.7 cm/hr with a standard deviation of 4.3 cm/hr, a value much less than the infiltration capacity of Myakka fine sand. Saturati on excess runoff is likely to occur in a catchment composed of this sand. This study proposes a modified form of the SCS method more applicable to flatly sloped, humid environments with sandy soils. Initial conditions, such as the pre storm soil moisture profile and depth to water table, are crucial when predicting runoff in these areas. Equations are presented that estimate the average depth to the water table and the average soil moisture storage available to be filled in a catchment. 1 2 Objectives and S cope The purpose of this research is to develop a modification of the popular SCS method more suitable for environments with shallow water tables, sandy soils, and flat terrain. The primary objective is to improve estimates of runoff volume with a more ph ysically based method that emphasizes antecedent water table depths. The spatial

PAGE 16

3 distribution of rainfall and water table elevations is the key to success in such an endeavor. The results of this research offer the hydrologic modeling community an alt ernate application of the SCS method for runoff estimation in shallow water table environments. The need for better runoff prediction in Florida has persisted for decades now. Design decisions are made every day by engineers using the traditional SCS met hod, a method created in environments subject to infiltration excess. The hope is that the new SCS SFWMD method will provide an alternative to the old method. This work is divided into five chapters. Chapter 1 reviews background information on the SCS me thod, current need for an improved method, well known runoff mechanisms, and past studies of the method conducted in Florida. Chapter 2 describes the materials and general methodology used in this research. Chapter 3 explains the details of the proposed SCS SFWMD method. Actual application and results of the proposed method are presented in Chapter 4. Finally, Chapter 5 discusses factors affecting the performance of the method and summarizes the important conclusions of this research. 1 3 Need for an Imp roved Runoff Estimation Method The SCS method is widely used in stormwater design analysis in Florida, for both urban and natural basins. The method is based on a few simple equations. Storm runoff depth, Q for a particular catchment is estimated by

PAGE 17

4 ( ) ( ) S I P I P Q a a + = 2 (1) where: P equals the storm rainfall depth, S equals the maximum potential retention, and I a equals the initial abstraction (usually 0.2 S ) The equation has a single empirical parameter, S which is related to a curve number (CN) by 10 1000 = CN S (2) where S is in inches. Curve numbers (CNs) are found in widely published tables, and the user typically selects one for a catchment based on soil type, land cover, hydrologic condition, and antecedent moisture. Equation (1) is sensitive to the selection of an appropriate CN because it represents the division of rainfall into losses and rainfall excess (Hawkins 1980). Knowing the rainfall depth, P the runoff depth is determined from a published r ainfall runoff plot, such as the one shown in Figure 1. Springer et al. (1980) stated that one of the major weaknesses of the CN method is the absence of local calibration using experimental watershed data, i.e., CNs obtained from the authorized handbook did not equal those from their derivation by local calibration using rainfall runoff data from various watersheds. The ratio of 0.2 used in the initial abstraction I a was

PAGE 18

5 also not corroborated by the least squares fitting routine performed by the investig ators, for either humid or arid watersheds. Hawkins et al. (2002) found that a ratio of 0.05 resulted in a better fit to event rainfall runoff data from over 307 watersheds and plots. 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 Rainfall, P (in.) Direct runoff, Q (in.) 30 40 50 60 70 80 90 100 Figure 1. The SCS Runoff Equation Solved for Selected Curve Numbers The SCS method is simple to apply to a variety of basins and yields consistent results for particular land use categories; consequently it is popular among regulatory agencies. However, the methods use in Florida has been criticized, especially when it has been applied in situations for which it was not intended. Urban development is spreading across the state, and many mistakes are made due to user subjectivity in

PAGE 19

6 determining the curve number (Golding 1997). Trommer et al. (1996) applied the Natural Resources Conservation Service (NRCS) TR 20 Model (based on the SCS method) to west central Florida basins, and the error in runoff volume for individual urban basins ranged from 84.5% to 156% while error in natural basins ranged from 97% to 318% The average CN for five natural watersheds in the study as determined by SCS standard procedures was 72.4. Some water resources professionals do not support the curve number method at all. Smith (1997) and Willeke (1997) believe that there is a desper ate need to advance the hydrologic community beyond the SCS method. Yet, there is strong resistance from some practicing engineers and regulatory agencies to eliminate the method altogether, and thus a suitable interim solution is to develop a reliable mo dification to the method wherein the storage parameter, S is more constrained and is based on the concept of saturation excess. Capece (1984) and Konyha et al. (1982) found that the best results for estimating runoff in Florida flatwoods with a modified S CS equation included an antecedent depth to the water table. Water table depth controls the formation of variable source areas, which play a significant role in the production of runoff, especially toward the end of the wet season. Variable source areas are regions that generate runoff before other regions of the catchment as a result of having soil that fills to capacity more quickly than at other locations. The water table depth is the indicator of available soil storage. As the water table rises in t he wet season, the soil storage decreases and the total area of saturation increases, causing the source of runoff to increase. Spier et al. (1969) observed that there was no runoff with water table depths larger than 0.76 m (2.5 ft), meaning that the soi l storage accommodated the rainfall depth. The South Florida Water Management District

PAGE 20

7 (SFWMD) Environmental Resource Permit Information Manual includes a procedure where the storage parameter, S in the SCS method is a direct function of the depth to the water table, and a curve is provided for the user to select an S based on water table levels (SFWMD 2002). Heatwole (1986) analyzed the CREAMS WT model, which simulates a dynamic water table with limited or no deep seepage and estimates available soil st orage. One problem often encountered when using the curve number method is that the spatial and temporal variability of a storm, the quality of measured data, and the effect of antecedent soil moisture conditions result in a set of curve numbers for the s ame watershed instead of a single CN (Ponce and Hawkins 1996). This demonstrates that lumping all of the many properties of a storm and a watershed into a single parameter, S produces a runoff value that is not physically based. 1.4 Review of Existing L iterature A review is provided in this section about the origin of the SCS method and runoff mechanisms as they pertain to shallow water table environments, such as that which exists in the state of Florida. 1.4.1 Development of the SCS Method The Soil C onservation Service (SCS) curve number method of surface runoff prediction can be traced back to the work of Mockus (1949) and Andrews (1954), among others. Thousands of infiltrometer tests, mostly using a sprinkling type infiltrometer, were conducted on agricultural plots of land in the late 1930s and early 1940s, mainly in the midwestern United States, to determine the effects of soil conservation procedures on

PAGE 21

8 rainfall runoff mechanisms (Ponce and Hawkins 1996, Rallison 1980). Mockus used data on soil, land use, antecedent rainfall, storm duration, and average annual temperature to estimate surface runoff in ungaged catchments (Mishra and Singh 1999). This work, combined with a graphical procedure developed by Andrews for predicting runoff from the soi l vegetation land complex, was generalized and named the SCS CN method as found in the Soil Conservation Service National Engineering Handbook Section 4: Hydrology (NEH 4) (USDA 1985). The first version of the handbook was printed in 1954, with subsequen t revisions in 1956, 1964, 1965, 1971, 1972, 1985, and 1993 (Ponce and Hawkins 1996). Catchment scale hydrological models may be classified according to their place in a wide ranging spectrum, with lumped conceptual models on one end and distributed para meter, more physically based models on the other end. Sivapalan et al. (1997) reviewed these model classifications and their characteristics. The SCS method is an example of a lumped parameter model because it ignores the spatial and temporal variability of catchment responses. The proposed method in this study uses physically based catchment properties to predict a catchment scale lumped parameter. A lumped model has the advantage of simplicity, which makes the SCS method popular. The original SCS met hod has been the subject of much review and discussion, and there have been many suggestions as to how to modify it for various scenarios (see Bosznay 1989, Mishra and Singh 1999, Perrone and Madramootoo 1998, and White 1988 as examples). Even though the method is simple, well established, and well documented, modifications have been attempted because the method: originates from data representing mainly the Midwest

PAGE 22

9 may be very sensitive to choice of curve number and antecedent moisture conditions performs best in agricultural sites and poorly in forested sites does not explicitly consider spatial scale effects, and fixes the initial abstraction ratio to 0.2 (Ponce and Hawkins 1996) User subjectivity in land use and soil type interpretation further contri butes to runoff estimation uncertainty. The method was intended for smaller catchments with runoff events of a significant magnitude; the success of the SCS method is limited by these factors (Yu 1998). The model watersheds used to develop the SCS method are quite different from the watersheds found in Florida, which have very flat slopes, highly permeable sands, shallow water tables, and numerous wetlands. Additional error arises when the SCS method is applied to Floridas frequent showers instead of ex treme (design) events for which the original method was intended (Capece 1984). Despite criticism of the method, Heatwole (1986) stated that since the curve number method is storage based, it should be a good model for analyzing the flatwoods watersheds i n Florida, if an appropriate estimate of the S parameter can be determined. The idea is that the concepts of the SCS method are useful and could be adjusted to suit a different interpretation of the original methodology. Indeed, Yu (1998) presented the m uch needed theoretical justification of the SCS method assumption that the ratio of the actual retention to the potential retention is the same as the ratio of the actual runoff to potential runoff. Using the basic concepts of the SCS method, a modificati on in its application may improve runoff prediction in Florida.

PAGE 23

10 1.4.2 Runoff Mechanisms Humid vegetated areas may generate several forms of runoff, including partial area Hortonian overland flow, saturation overland flow, and subsurface flow, with the latt er being further subdivided into rapid throughflow due to macropores and displacement of pre storm water (old water). Pearce et al. (1986) reviewed these forms of runoff and made the important observation that they all originate from seasonal expanding and contracting zones of contribution, or variable source areas. The term variable source area in this study on saturation excess in Florida basins will only refer to areas producing saturation overland flow. There is general agreement regarding Dunne s theory (1983) that gently sloping humid catchments with thin soils and wide valley bottoms will produce hydrographs dominated by saturation overland flow. Subsurface stormflow cannot be separated from saturation excess because they are often linked, tho ugh the extremely flat slopes in Florida probably retard flow beneath the surface. Instead, return flow is likely to occur, where subsurface water returns to the surface to run overland at a high velocity, an event caused by the inability of the subsurfac e stormflow to remove incoming rainwater (Dunne et al. 1975). Return flow and direct precipitation on saturated areas are grouped under the category of saturation overland flow. Figure 2 shows the proposed influences of climate and vegetation on this and other types of runoff producing processes. Hortonian overland flow occurs when rainfall intensity exceeds the infiltration rate of the soil, causing resistance to downward flow and ponding at the surface. Because many Florida soils have saturated hydrau lic conductivities of nearly 40 cm/hr, with a typical value of 20 cm/hr (Carlisle et al. 1989), only the most intense convective storms should produce Hortonian runoff. Saturation

PAGE 24

11 Figure 2. Proposed Factors Influencing Runoff Mechanisms (Dunn e 1983) excess is likely to occur instead, wherever the soil storage is filled and the rainfall can no longer infiltrate the soil. However, the presence of air in the soil matrix may impede infiltrating water, lowering the infiltration rate during a stor m to a point where Hortonian runoff could occur. The concepts of soil storage capacity and air trapped in the soil matrix are further discussed in Chapter 3. Due to spatial variability in rainfall and catchment characteristics, it is unlikely that larger basins produce runoff with just one mechanism. Simulations have shown that Hortonian and Dunne runoff processes can occur simultaneously at different locations or switch from one process to the other at the same location, depending on initial conditions and characteristics of the rainfall event (Loague and Abrams 2001). Storm response models have been developed that include Arid to Subhumid Sparse Vegetation Humid Dense Vegetation Climate and Vegetation Soils and Topography Thin Soils High to Low Permeability Flatter Slopes Deep Soils High Permeability Steep Slopes Hortonian Overland Flow Dominant Subsurface Stormf low Dominant Saturation Overland Flow Dominant

PAGE 25

12 both the Hortonian and Dunne mechanisms, such as the dimensionless flood frequency model by Sivapalan and Wood (1990). 1.4.3 Runoff Studies in Florida Florida flatwoods have been the subject of a variety of runoff experiments, even though the term flatwoods is also used for areas where the pine forests (the woods) have been removed for range and pasture (Heatwole 1986). Lower lying areas of the flatwoods begin generating runoff before an entire catchment is saturated, and thus there is no strict storage relationship. Capece et al. (1987) concluded that estimates of runoff volume are sensitive to errors in data collection and basin d elineation, both of which are difficult in a flatwoods watershed. Past studies of rainfall runoff relationships in Floridas flatwoods have revealed that hydrology in environments with sandy soils and shallow water tables exhibit three main types of flow: slow, intermediate, and rapid (Spier et al. 1969). Slow flow is often associated with the groundwater contribution to a hydrograph, or baseflow. Long recessional limbs are typical of hydrographs in Florida because of the vast surface storage in wetland s and other depressions. Wetland storage slows or completely prevents a portion of the runoff from a storm from reaching a stream. During the wet season, when a wetland is nearly full, a rainfall event may raise the water level just enough to reach the i nvert of the outlet of the wetland. If this happens, a slow stream of runoff will flow from the wetland to the nearest channel and eventually contribute to the measured total flow at the outlet. In addition, directly connected wetlands fringing a stream have been observed draining into the channel year round. This is in contrast to rapid runoff resulting from overland flow from a variable source area.

PAGE 26

13 Rapid flow runoff occurs infrequently because of the high infiltration rates of sandy soils, and it has been found to occur mainly in September and October, when the water table is high (Spier et al. 1969). Intermediate flow is not easily distinguishable, but lies somewhere between rapid overland flow and groundwater baseflow, perhaps associated with subsur face stormflow during a rainfall event. Subsurface flow is generated by rapid infiltration (in sandy soils) and the resulting increase in hydraulic conductivity of the upper soil layers (Pearce et al. 1986). A significant source of runoff in Florida may come from bank storage release. The water table depth plays a key role in this mechanism, when during a storm, the water table gradient in the stream banks rises to a higher elevation than the stream water level. Once the flood wave passes through the rea ch, the bank storage is released back into the stream. When performing a storm analysis, Ponce and Hawkins (1996) proposed that infiltration is the most important hydrologic abstraction (a short term process) whereas interception and surface storage are often of lesser importance. The analysis in this study attempts to account for wetland surface storage and saturated areas by dividing the catchment into unsaturated areas of possible future saturation and initially saturated areas immediately available t o contribute runoff. Saturation overland flow is considered to be the only mechanism of runoff, and wetlands are assigned zero soil storage capacity.

PAGE 27

14 CHAPTER 2. MATERIALS AND METHODOLOGY 2.1 Data Collection Figure 3 shows the location of the two bas ins in west central Florida that were selected for study: West Fork Horse Creek (WFHC) and Long Flat Creek (LFC). A well instrumented runoff test bed in the LFC basin was added for close inspection of the saturation excess mechanism. Figure 3. Locat ion of Selected Study Basins

PAGE 28

15 The WFHC basin had the least amount of available data, while the LFC basin and runoff test bed contained a variety of instruments that recorded data at small time steps. The Center for Modeling Hydrologic and Aquatic Systems ( CMHAS) at the University of South Florida (USF) was responsible for the data collection effort at LFC. All data for the WFHC basin came from the United States Geological Survey (USGS). 2.1.1 Runoff Test Bed A schematic of the runoff test bed is in Figu re 4. The purpose of the test bed was to test the methodology at a small scale. Located on the west side of LFC (Figure 8), its proximity to the stream made it a prime candidate as a variable source area. Two continuously recording surficial wells, USF 1 and USF 3, provided water table elevations every five minutes. As with all recording wells in the LFC basin, USF 1 and USF 3 housed Instrumentation Northwest 0 5 psi submersible pressure transducers, accurate to 0.005 psi. Two non recording wells, USF 2 and USF 4, were available for supplementary water table data when needed. All test bed wells had a total depth of 4.6 m. Well construction at LFC was done by the Southwest Florida Water Management District (SWFWMD). A typical well was made with 5.08 cm (2 in.) PVC pipe, with a slotted PVC screen extending below a bentonite clay seal. Silica sand was packed around the screen to allow only the passage of water. The test bed was 30.5 m (100 ft) long and 6.1 m (20 ft) wide with a slope of 2%, i.e., there was a 0.6 m (2 ft) vertical drop along the 30.5 m (100 ft) length of the bed. Surface and subsurface stormwater was prevented from flowing laterally out of or into the bed by inserting aluminum flashing 10.2 cm below the surface along the sides,

PAGE 29

16 Figure 4. Instrumentation at the Runoff Test Bed creating a 20.3 cm wall aboveground (Figure 5). A trench was also excavated around the perimeter of the bed for the same purpose. A Unidata tipping bucket rain gage (accurate Well USF 1 Soil Moisture Probe Weir Collection Trough Well USF 2 Well USF 4 Soil Moisture Probe Well USF 3 2 Elevation Above Weir (ft)

PAGE 30

17 to 0.025 cm) was installed on the downstream end, as well as a PVC water collection trough and an aluminum 90 o V notch weir box. The V notch on the weir was 15.2 cm high and the length, width, and height of the box was 30.5 cm on each side of the weir plate A 7.6 cm outlet pipe was positioned near the bottom of the downstream side of the box. Figure 6 shows how the water level in the weir box was measured with a PVC stilling well and Unidata water level instrument, model #6531 (accurate to 0.75 mm), the s ame method used to measure stream stage at LFC. Beaded float lines (125 mm) and 12.7 cm (5 in.) floats were used in the assembly. The rain gage recorded at a 5 minute frequency to document the changes in rainfall intensity, which was especially useful dur ing convective storms. Two Sentek EnviroSCAN soil moisture probes (Figure 7) recorded soil moisture concentration at different depths in the soil horizons beneath the land surface. Before May 3, 2002, sensors were spaced at 0 cm, 10 cm, 20 cm, 30 cm, 40 cm, 70 cm, 100 cm, and 150 cm. Figure 5. Photograph of the Runoff Test Bed

PAGE 31

18 Figure 6. Photographs of the Test Bed Weir Box The arrangement of the sensors after May 3, 2002 was 10 cm, 20 cm, 30 cm, 50 cm, 70 cm, 90 cm, 110 cm, and 150 c m below the surface. The sensors, using the measuring principle of capacitance via the dielectric effect, have the capability of measuring volumetric water content in soil ranging from saturation to oven dry with a resolution of 0.1% (Buss 1993). These p robes were used to determine soil storage capacity. Fares and Alva (2000) tested the probes in central Florida and found that for Candler fine sand, there was no significant difference in water content as measured by the probes and the gravimetric method. Morgan et al. (1999) found that the calibration curve supplied with the probes underestimated water content in the plant available range for three fine sands in Florida. Use of the probes for irrigation planning in the dry season requires adequate cali bration of very low water content values. However, this study on saturation excess depends on higher water content values because the phenomenon typically occurs in the wet season. In general, researchers have found the capacitance probes to be reasonabl y accurate.

PAGE 32

19 Figure 7. Photograph of a Soil Moisture Probe at Long Flat Creek The physical properties of the soils were determined by sampling and standard laboratory procedures, such as wet ASTM sieve analysis and falling head permeability tests. Soil samples were extracted where soil moisture probes were installed and were given the name of the nearest well. The porosity of the samples by layer was the most important parameter for calculating soil storage. Porosity, defined as the volume of voids di vided by total sample volume, was determined by measuring the difference between the mass of a fully saturated and a completely dry soil sample and then applying density relationships. Because the soil samples were subject to settling and collapsed pores, the cumulative particle size distribution was entered into MVASKF, a public domain program that uses empirical equations to calculate parameters such as conductivity and porosity. If the porosity determined in the laboratory was smaller than the maximum

PAGE 33

20 observed water content in the field for the same soil layer, then the field measured water content or the MVASKF determined porosity was used, whichever was highest. Two Unidata STARLOGGER data loggers (model #7001A) at the test bed site stored rainfall, s oil moisture concentration, water table elevation, and runoff over the weir. These detailed data, along with the small size of the test bed, minimized spatial variation in rainfall, water table elevation, and soil characteristics, which allowed for better analysis of the saturation excess mechanism. However, the maintenance of the equipment was a difficult task, especially for the test bed weir. The demonstration of saturation excess required adequate rainfall, a water table rise to the surface, and prope r equipment operation throughout the entire storm. 2.1.2 Long Flat Creek A subbasin of the Long Flat Creek catchment was useful in studying saturation excess on a larger scale. The area of the subbasin is 0.753 km 2 smaller than the WFHC basin. Long Fla t Creek (LFC) flows in a northwesterly direction and is part of the Alafia River basin. All equipment in the subbasin was of the same make as the equipment previously described for the runoff test bed. Numerous surficial wells, shown in Figure 8, were us ed to measure water table elevations, and two 0.64 cm (0.25 in.) thick aluminum complex weirs were installed in the creek to enhance measurement of streamflow at low flow conditions (Figure 9). The weirs were 2.13 m (7 ft) wide with a 0.30 m (1 ft) high 9 0 o V notch combined with a 0.46 m (1.5 ft) high rectangular weir. UNIDATA recording water level instruments with 12.7 cm (5.0 in.) floats and 20.3 cm (8 in.) stilling wells with 0.64 cm (0.25 in.) diameter holes on the downstream side were

PAGE 34

21 installed next t o the weirs to measure stream stage. Unidata STARLOGGER data recorders logged soil moisture concentration from soil moisture probes, rainfall from rain gages, water table elevation from pressure transducers, and stage from the shaft encoders near the stre am. The Unidata tipping bucket rain gages (0.025 cm resolution) were located near USF 1 and PS 1. Some pressure transducers were subject to drifting, where the manual depth to water measurement increasingly differed from the transducer reading. Depth to water readings were adjusted accordingly, and these transducers were sent back to the manufacturer for repair. The Sentek soil moisture probes were installed only on the west side of the creek and were associated by proximity to the following wells: USF 1, USF 3, PS 40, PS 41, PS 42, and PS 43. The spacing of the sensors was the same as for the runoff test bed probes, with the exception of PS 43, which had a sensor at 50 cm instead of 40 cm before May 3, 2002. After this date, all sensor spacing was un iform. The soil texture and porosity data at the soil moisture probes by depth are presented in Table 22 of Appendix B. Similar to the runoff test bed, it was rare to record a heavy rainfall event where all the needed equipment was operating properly for the entire duration of the storm. The downstream weir was not operated by CMHAS, but by Tampa Bay Water. Basin delineation and surface storage estimation above the upstream weir were problematic because the LFC basin includes an old phosphate mining area There were numerous ponds connected by pipe at high water levels and several culverts that directed flow to the creek from disturbed areas. The subbasin shown was delineated in an effort to minimize the inclusion of troublesome areas. Analysis of LFC was conducted in a manner similar to that of WFHC, described in the next section.

PAGE 35

22 Figure 8. The Long Flat Creek Subbasin (Contour Interval: 1 ft)

PAGE 36

23 Figure 9. Photograph of a Complex V notch / Rectangular Weir 2.1.3 West Fo rk Horse Creek One of the least developed basins in west central Florida, West Fork Horse Creek (WFHC) drains approximately 35 km 2 (13.5 mi 2 ) of pastureland in northeastern Manatee County and northwestern Hardee County (Figure 10), and flows in a southeas terly direction to Horse Creek, a major tributary of the Peace River. This basin was chosen because it is relatively undisturbed and was less heavily instrumented than LFC. Often a scientist or engineer has to work with a basin that is sparsely instrumen ted, if instrumented at all. WFHC represented a basin of this type and was used to demonstrate how the proposed method is applied in this situation. Hourly and daily rainfall and streamflow data were obtained from the water resources database of the USGS There was only one surficial well in the basin, which was critical for determining depth to the water table in the basin. The spatial distribution of rainfall during storms was determined with Thiessen polygons. Three USGS rainfall stations near the b asin were used in the

PAGE 37

24 analysis: Watkins Road, Mitchell Hammock, and West Fork Horse Creek. Their rainfall records were assigned to 37%, 8%, and 55% of the basin area, respectively, based on the Thiessen polygon technique. If one of the stations malfunc tioned while the other two registered rain, then the area weight of the bad gage was evenly split between the other two gages in order to assign the rainfall to 100% of the basin area. A digital elevation model of the catchment was generated using 1.52 m (5 ft) contour data and a TIN (triangulated irregular network). The 1.52 m (5 ft) contours were considered to have poor vertical resolution for Floridas flat topography; 0.30 m (1 ft) or less would have been the preferred interval, but this kind of data was not available at the time. Such high resolution contours were available only for LFC. ArcView 3.2, a popular geographic information system (GIS) software, was used for most of the analysis. A polygon shapefile of soils from SWFWMD was useful for div iding the basin up by soil type. Soil properties, such as saturated hydraulic conductivity and water content, were available from the Hardee County soil survey (Robbins et al. 1984) and the Institute of Food and Agricultural Sciences (IFAS) at the Univers ity of Florida (Carlisle et al. 1989). These properties are tabulated in Appendix A.

PAGE 38

25 Figure 10. The West Fork Horse Creek Basin (Contour Interval: 5 ft)

PAGE 39

26 2.2 Methodology This section describes the concept of variable source area and an associated el ementary bucket model. The proposed runoff estimation method in this study incorporates a new definition of S in the SCS method that better represents these ideas. 2.2.1 The Bucket Model A spatially non uniform catchment in a shallow water table environ ment does not generate runoff as a single unit. Instead it may be conceptualized as an ensemble of buckets, each containing its own soil storage capacity. The available soil storage capacity near a stream is typically less than that found in the uplands. An observed exception to this at LFC is where there is heavy streamside vegetation such as water oaks that use large amounts of water for transpiration. In this case, there is more soil storage capacity directly adjacent to the stream than in the upland s. As the water table rises during a storm, water occupies all of the available space in the soil matrix, saturating the soil profile. All subsequent rainfall becomes runoff once saturation of the soil occurs and the water table is at the surface. Satur ation excess runoff refers to the water that cannot infiltrate at this point, but instead travels to a stream as overland flow. Areas adjacent to wetlands are usually the first to be saturated, causing the perimeter of a wetland to expand as rainstorms co ntinue to fill available soil storage. The wetland is a variable source area because its area of saturation varies by season. It is a source of runoff whereas unsaturated uplands are not. During the wet season more and more areas of the catchment become saturated, allowing for increased contribution to runoff. Each area of equal contribution may be modeled as a bucket.

PAGE 40

27 In order to analyze the saturation excess areas, an elementary bucket model may be used where no runoff is produced until a specified rainfall depth has filled the soil storage capacity, I, of a bucket (i.e. soil), after which all rainfall becomes runoff. The bucket size is best measured by the water table depth, and since the water table changes with time, the bucket size will vary als o (Sivapalan et al. 1997). The following discussion about buckets and fractional area contributions is based on Boughtons (1987) idea of evaluating partial areas of runoff in a watershed. Figure 11 presents runoff, Q, plotted against effective rainfall (total rainfall minus initial abstraction), P e where the slope of the curve is zero for P e < I and one (45 degrees) for P e > I. This condition exists for a catchment modeled as a bucket with a spatially uniform depth to water table and antecedent wetnes s. There is no runoff until the storage space, I, is filled, and then all rainfall becomes runoff, resulting in a slope of one. In reality, a catchment has some areas contributing to runoff before other areas due to variations in available storage in the soil. Assume now that there are two buckets in the model, with one area fraction of the watershed containing a storage capacity of I 1 and the other area fraction containing a storage capacity of I 2 The two buckets representing fractions of the entire catchment area, A, in Figure 12 demonstrate that the bucket with the smaller storage will produce runoff before the bucket with more soil storage. As soon as the second bucket is full the whole catchment area will be generating runoff. In Figure 13, the s oil storage capacity, I 1 is satisfied before producing runoff. The slope of the line is determined by tan a e.g., if 30% of the area were simulated by

PAGE 41

28 bucket 1, then tan 1 0.30 would be angle a in the diagram. This relationship holds true because the runoff from the catchment is expressed as 2 1 A P A P QA R e e + = = (3) where: R equals runoff volume, Q equals runoff depth, A equals total catchment area, P e equals effective rainfall, A 1 equals area representing bucket 1, and A 2 equals area representing bucket 2 Figure 11. Runoff Response in a Single Bucket M odel for a Uniform Watershed I, soil storage capacity Pe, effective rainfall Q 45 o

PAGE 42

29 Figure 12. Two Buckets Representing Two Partial Areas Comprising a Watershed Figure 13. Two Buckets Initiating Runoff at Different Times I 2 I 1 Area Fractio n A 1 / A A 2 / A Runoff Pe, effective rainfall Q 45 o I 1 I I 2 line b line a a

PAGE 43

30 If bucket 1 is full and is generating runoff while bucket 2 is not yet saturated, the last term of (3) is zero. Rearranging the equation at this point gives e P A A Q 1 = (4) which is in the form of the equation of a line, y = mx. The slope, m, of line a is the area fraction of the catchment producing runoff. This result is in agreement with an observation by Hawkins (2001), that within the partial area concept, the slope of the rainfall runoff curve is equivalent to the contributing area fraction. When bucket 2 is full, line b continues from line a, but at a slope equal to one because all areas are producing runoff. Two buckets result in two broken lines in the figure, but dozens of buckets would generate a smooth line, similar to the SCS method curve number p lot in Figure 1. Steenhuis et al. (1995) explained that trigonometry shows that the extension of line b to the x axis reveals I* the intersection being equal to ( ) ( ) + = + = e e e P S S S P S P I 2 1 2 (5) I* the average soil storage capacity of the runoff produci ng areas, approaches S (maximum retention) as P e approaches infinity, which in Figure 13 means that the entire catchment is producing runoff. As noted by Yu (2001) and others, the average soil

PAGE 44

31 storage capacity may be used for the S parameter in the SCS ru noff equation. The area weighted soil storage capacity represents a large number of storage buckets, and it is this value that may be used for a real catchment dominated by variable source areas. The S parameter may be calculated as = A Ida A S 0 1 (6) where: I equals the soil storage capacity, A equals the pre storm unsaturated area, and da equals a dummy variable of integration The dummy variable represents a single uniform area, such as a pixel in a GIS grid. As noted above, only the total unsaturated area in the catchment before a storm event is considered in the calculation because saturated areas have no soil storage capacity. Wetlands and recently saturated ground are incorporated separately and are discus sed in Chapter 3. Calculation of S requires the distribution of I over the catchment, which is derived from the spatial distribution of the water table depth and soil type. Areas in the catchment with homogeneous characteristics are identified so that a separate soil storage capacity may be calculated for each uniform area. An area weighted S is then determined for the catchment.

PAGE 45

32 2.2.2 The Soil Storage Capacity Equation The total rainfall depth needed to fill the soil storage capacity, I as a function of depth to water table, d may be expressed as = 1 1 ) ( ) ( ) ( 1 l l q q A A A g s h d h h d d I (7) where: q s equals the water content at natural saturation, q g equals the water content retained after gravity drainage (field capacity), h A equals the air entr y pressure, and l equals the pore size distribution index This equation was derived from the Brooks and Corey model of water retention (Eq. (20) in Appendix C) by first replacing h C capillary pressure, with z elevation above the water table, and then so lving for q so that the water content is a function of z q g was considered equivalent to q r and q s was substituted for f Next, the area between q s and the water content profile, as shown in Figure 14, was found by integration between the limits of 0 a nd d, using the new expression for q as derived from Eq. (20). The result of the integration was rearranged and simplified into the final form as presented above. Equation (7) is valid only for d > h A or when the depth to the water table is greater tha n the height of the capillary fringe, which is considered fully saturated. The equation is

PAGE 46

33 similar in form to the one given by Sivapalan et al. (1987) for total soil moisture deficit, but it has been modified to include the phenomenon of air encapsulation The limitations of (7) are a static water content distribution above the water table and a homogeneous soil profile. The Brooks and Corey soil model parameters ( q s q g h A l ) are found in many textbooks or derived from laboratory data. Figure 14 provides a visualization of (7), where I(d) is the depth of rainfall required to fill the hatched area representing soil storage capacity. The volume between q s and f t otal porosity, is the encapsulated air, a topic discussed in Chapter 3. An alternative to using (7) requires soil moisture characteristic curves above various depths to the water table. The calculation of soil storage is performed by summing the differen ces in all soil horizons between the effective porosity and measured water content at a particular depth. Either the equation or manual integration between water content profiles may be used to obtain the spatial average of soil storage capacity in the ca tchment, as shown in Eq. (6). To optimize Eq. (7), the average Brooks and Corey parameters as determined from laboratory data (Table 18, Appendix A) may be used as a first guess in a regression routine with the storages found from the water retention curv es (see Chapter 3).

PAGE 47

34 Figure 14. Soil Moisture Storage Capacity Above a Shallow Water Table Depth below ground surface Water table h A q s q g d I(d) l Water content (volume of water / volume of soil) f

PAGE 48

35 CHAPTER 3. THE SCS SFWMD METHOD 3.1 Soil Storage Capacity Soil characterization and properties are used to arrive at a physically based soil storage capacity value. Details of soil storage calculation for actual data from three soils are provided at the end of the section. 3.1.1 Soil Characterization The accurate identification of soils in a basin is crucial for a successful estimation of runoff. In the absence of field studies and soil sampling, county soil surveys are the best available information for dividing the basin into areas of homogeneous soil type. Bhaskar et al. (1992) stated that a GIS is a valuable tool for analyzing soil s and capturing variable source areas. If a GIS is available, soil polygons can be digitized from the county survey and then transformed into a grid. A high level of detail in soil characterization results in more accurate runoff estimation. Field studi es and soil sampling were available for the Long Flat Creek (LFC) subbasin and the runoff test bed, while only county soil surveys were available for West Fork Horse Creek (WFHC). A number of different flatwoods and depressional soils comprise the WFHC ba sin, but three representative soils were chosen to reduce computation and complexity. Soil textures of the three soils from the county survey were compared to textures of various soils sampled by the Institute of Food and Agricultural Sciences (IFAS) (Car lisle et al. 1989) to find the

PAGE 49

36 closest match. The final three soils were Myakka fine sand, Smyrna sand, and Immokalee fine sand, and one of these was assigned to each pixel in the soils grid, based on which soil was most similar to the actual soil. Water content data published by the IFAS were used to derive the Brooks and Corey parameters required in Eq. (7). These parameters are listed in Table 18 of Appendix A, and an example of fitting data to the Brooks and Corey model is in Appendix C. Porosity, t he total amount of pore space in a soil, was calculated from bulk density data also provided by the IFAS. 3.1.2 Using the Soil Storage Capacity Equation Once soil characterization is complete, the soil storage capacity equation may be prepared for use. Equation (7) was rearranged into the form D Cx Ax y B + + = (8) which contains the four constants, A B C and D that represent terms in ( ) + + + = l q q q q q q l q q l l 1 ) ( ) ( ) ( 1 ) ( ) ( 1 1 A g s A g s g s A A g s h h d d h h d I (9) so that: A equals the entire term in the leftmost set of brackets, B equals 1 l C equals ( q s q g ), and D equals the entire term in the rightmost set of brackets

PAGE 50

37 The independent variable, x is the depth to the water table, d and the dependent variable, y is the rainfall depth needed to fill the soil storage capacity, I A regre ssion was performed using (8) and soil storages as calculated manually from water retention curves. 3.1.3 Air Encapsulation Air encapsulation is an important consideration in environments with shallow water tables and sandy soils. It causes water table r ises in soils which are significantly faster and higher than those in soils without air encapsulation (Fayer and Hillel 1986). Water infiltrates soils with large pores quickly, leaving air in the soil profile little time to find a route of escape. As a s ide note, rapid water table rises observed in shallow water table environments have also been explained by the proximity of the capillary fringe to the ground surface. Gillham (1984) researched large, rapid water table rises and falls caused by either the addition or loss of a small amount of water, respectively, to a shallow capillary fringe. Myers (1999) studied the hydraulic properties of wetland peats in Florida and noted that the capillary behavior in the vadose zone can cause rapid water table rises These concepts should be kept in mind alongside the discussion of the effect of air encapsulation on a shallow water table. Seymour (2000) suggested that the large pores in sandy soils increase the likelihood of discontinuous air bubbles being trapp ed in the soil matrix. As the soil becomes wetter during infiltration, the hydraulic conductivity of the soil increases while the relative permeability of air decreases (Charbeneau 2000). The mobility of the air is reduced while interconnected pores are closed off by an advancing wetting front. Immobilized air bubbles are surrounded by water, occupying a fraction of the available

PAGE 51

38 pore space and reducing hydraulic conductivity. Runoff volumes are expected to be larger wherever there is a significant volu me of trapped air, simply because there is less available space in the soil matrix for infiltrating water to fill. Wangemann et al. (2000) studied decreased infiltration rates due to displaced air, noting that drier soil in particular would have more air to block conducting pores. Infiltration rates are impacted because air counterflow and compression ahead of the wetting front generate resistance to downward flow (Wang et al. 1997). Morel Seytoux and Billica (1985) found that the existence of an impervi ous bottom in a soil profile (e.g. shallow water table) greatly alters the infiltration process because of air compression. The infiltration rate may decrease to nearly zero if entrapped air is not able to erupt through the soil surface and escape to the atmosphere. In agriculture, it is critical that stormwater be able to infiltrate down into the soil for root water uptake. To solve the problem of reduced infiltration, Jarrett et al. (1980) installed a subsurface drain in a sandy soil above an impermeab le layer to provide a vent for entrapped air and enable more water to infiltrate. Air compression during infiltration causes sudden, artificial water level rises in water table wells because the fluid in the well is exposed to atmospheric pressure while the soil matrix is being pressurized (Freeze and Cherry 1979). The pressurization that occurs with air entrapment above a shallow water table causes an unstable wetting front that forms fingers, while also causing fluctuations in the infiltration rate d ue to pressure buildup and subsequent air eruptions (Wang et al. 1998). The soils in Florida flatwoods may be seasonally ponded as a result of periodic water table rises. Taboada et al. (2001) studied abnormal soil swelling processes caused by air trappe d between a ponded surface and a rising water table. They found that the pore air volume increased during soil

PAGE 52

39 wetting to a value as high as 35% of total soil volume, yet no air escaped to the surface. The more entrapped air there is in the vadose zone, the more runoff the catchment produces. The subjects of air encapsulation, hysteresis, and air compression are two phase flow problems that add to the complexity of runoff estimation. It is possible that the unstable infiltration rates in shallow water ta ble environments with sandy soils result in runoff generated by a combination of infiltration and saturation excess. To simplify the issue and to remain within the concept of saturation excess, a previous study involving field measurement of encapsulated air was used to determine the volume of air as a function of depth below the ground surface. Fayer and Hillel (1986) conducted field experiments with sprinklers to determine the encapsulated air content in a soil profile consisting of fine sandy loam that transitions to loamy sand near the bottom. Table 1 presents their measured air volumes as a function of depth with a sprinkling rate of 1.26 cm/hr (0.5 in/hr) and with the water table brought to the surface; these values were used to determine the volume of encapsulated air in the three representative flatwoods soils of Myakka fine sand, Smyrna sand, and Immokalee fine sand. At the time, these were the only known available estimates of air encapsulation as a function of depth. It was assumed there was n o air encapsulation deeper than 1.20 m in the soil profile because they reported no measurable encapsulated air at the depths of 1.35 m and 1.5 m below land surface. Wilson et al. (1982) approximated the volume of entrapped air for one type of loamy sand as 15% of the porosity, while Constantz et al. (1988) approximated the air volume in medium Olympic sand as 19% of the porosity. Most Florida sands have

PAGE 53

40 porosities near 0.40, and 15% and 19% of this value are 0.06 and 0.08, respectively, which are similar to the values found in Table 1. Table 1. Encapsulated Air Volume as a Function of Depth (Fayer and Hillel 1986) Depth (cm) 30 45 60 90 120 Volumetric Encapsulated Air (%) 4.8 5.0 6.3 4.7 1.7 The soil moisture probes in the runoff test bed at LFC pro vided direct measurement of encapsulated air. The sensors on the probes were positioned at selected depths in the soil matrix to monitor water content changes. When there is no bulk density data available to derive porosity, the porosity at a location in a soil may be taken as the highest water content ever measured at that location (Fayer and Hillel 1986). This maximum porosity value should be determined at a point that has been saturated for a long time, probably more than a month. This method was used at the LFC subbasin and runoff test bed whenever there was an observed water content value greater than the porosity determined in the laboratory. This was the case for PS 43, shown in Figure 15. The laboratory porosity was less than the water content a t the 150 cm sensor, which was under the water table before and after a February rainfall event. For this storm, it was assumed that the porosity was constant throughout the soil, an assumption not unfounded since the soil near PS 43 is mostly uniform san d (Table 22, Appendix B). The difference between the porosity and the final water content value one to two days after the storm was considered to be air encapsulation. The soil moisture profile one to two rainless days

PAGE 54

41 after the storm was used to allow f or transitions to smooth out, especially those associated with air compression. The figure shows that the soil sensor located at 100 cm below land surface near well PS 43 did not fill to maximum porosity, though it was submerged beneath the water table du ring a frontal storm. The difference between the submerged sensor depth and the final water table depth is the pressure head above the sensor. The pressure head may affect how much air remains trapped at a particular sensor location. Note also that the lines through the data points in the figure do not necessarily reflect the true moisture profiles; they were fitted through the points merely for visualization purposes. Figure 16 shows the water content with depth and water table position near well USF 1 at the runoff test bed before and after a storm occurred on April 12, 2002. In the figure, the area between the initial water content line (diamonds) and the natural saturation line (triangles) is the soil storage capacity, I The area between the porosi ty line (square markers) and the final profile is occupied by air. In this case, the laboratory determined porosity increases slightly with depth. Note that the laboratory porosity is less than the measured final water content value near the surface; thi s difference may be explained by the collapse of macropores during sample retrieval. The air encapsulation at the moisture probes 40 cm and 70 cm below the surface was determined by the difference between the final soil moisture profile after the storm en ded and the porosity. These probes were located between the initial and final water table positions and did not fill to porosity. The air encapsulation values from these two sensors were added to other data from other probes and storms (Appendix B) to ev entually find an average air encapsulation as a function of sensor depth.

PAGE 55

42 PS-43 Soil Moisture Profiles (Storm began 2/22/02 13:30 and ended 2/23/02 21:40) 0 25 50 75 100 125 150 175 0 0.1 0.2 0.3 0.4 Soil Moisture (cm 3 /cm 3 ) Depth (cm) 2/21/02 0:00 Initial Water Table 2/25/02 20:30 Final Water Table Porosity Encapsulated air Pressure head Figure 15. Air Encapsulation at a Submerged Soil Moisture Sensor LFC data was plotted to examine the relationship between air encapsulation and head above a moisture sensor locati on (Figure 17). When the water table rises during a storm, the head above the moisture sensors likely influences how easily air can escape. It was postulated that the higher the head (water table elevation minus sensor elevation) above a sensor, the m ore pressure there is to help squeeze the air out. However, the air encapsulation data did not exhibit a relationship to pressure head, but rather to depth below land surface. The soil closer to the surface contained less air, probably because macropores allow air to escape easily to the atmosphere. There was more air encapsulation from 40 cm to 100 cm below the surface because this is the range where the water table rises and falls quickly, especially during the wet season. In addition, there

PAGE 56

43 is less d irect connection to the atmosphere here. At depths greater than 100 cm, the air volume decreased because of the constant presence of the shallow water table replacing air filled pores with water over time. 0 20 40 60 80 100 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Initial Profile Final Profile Lab Porosity Initial Water Table Final Water Table Pressure head Figure 16. Soil Moisture Profile Near USF 1 fo r April 12, 2002 An adjustment was applied to LFC and runoff test bed soil storage data to correct for air encapsulation with depth. The average air encapsulation was used because the air volumes as a function of depth were similar and also because the a verage describes the generalized condition for a Florida flatwoods environment. In addition, the average helps to smooth out error in the estimates of individual porosity. In particular, the porosity values obtained for the upper soil layers were small er than expected because of

PAGE 57

44 macropore collapse during sampling. A smaller porosity translates to a smaller calculated air encapsulation value. R 2 = 0.0012 0 5 10 15 20 0 10 20 30 40 50 60 70 80 Pressure Head (cm) Encapsulated air (%) Figure 17. Air Encapsulation as a Function of Pressure Head Average air encapsulation for probes at the te st bed is shown in Table 2, while the LFC subbasin values are presented in Table 3. Both sets of data are plotted in Figure 18. These values are somewhat larger than those found in Table 1. The effective porosity (natural saturation) at each soil moistu re probe was found by subtracting the volumetric encapsulated air from the laboratory (or maximum field measured) porosity.

PAGE 58

45 Table 2. Average Air Encapsulation in the Runoff Test Bed Soil Sensor Depth (cm) 20 30 40 50 70 90 100 110 Volumetric E ncapsulated Air (%) 2.6 6.5 11.0 9.2 9.6 9.8 8.7 3.4 Table 3. Average Air Encapsulation in the Long Flat Creek Subbasin Soil Sensor Depth (cm) 20 30 40 50 70 90 100 110 Volumetric Encapsulated Air (%) 2.6 7.7 11.0 11.2 8.0 8.2 8.7 3.4 0 20 40 60 80 100 120 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Volumetric Encapsulated Air (%) Depth Below Land Surface (cm)) Test Bed Long Flat Creek Figure 18. Average Air Encapsulation at the Test Bed and Long Flat Creek

PAGE 59

46 3.1.4 Soil Storage Capacity for Three Layered Soils The three main soils used in the study Myakka fine sand, Immokalee fine sand, and Smyrna sand are layered and do not fit the homogeneous soil profile required by Eq. (7). This section presents the details concerning the soil storage analysis of the three layered soils used in the study. Appendix A lists the properties of the three soils by horizon. The first step was to d etermine the Brooks and Corey parameters for each layer separately. For example, Myakka fine sand had six horizons, and the water content at field capacity, pore size distribution index, and bubbling pressure were determined by the Brooks and Corey model for each. The average over the horizons provided the final values. The next step was to develop the soil moisture storage curves as a function of water table depth. For soil storage calculations, the water table depth was varied by 10 cm increments from ground surface to 200 cm, with the resulting soil profile and storage above the water table then determined. The porosity was calculated from the bulk density and then corrected for air encapsulation using Table 1; the subtraction of the selected air volu me from porosity was deemed natural saturation. The values in Tables 2 and 3 were not yet known at the time that the soil storage for Myakka fine sand, Immokalee fine sand, and Smyrna sand was being determined. Runoff estimates for these three soils woul d increase if more air were included in the soil storage calculation. Figures 19 to 21 depict the soil moisture profiles for the three soils at different water table depths before and after correction for air encapsulation. As an example of how the profi les look with multiple soil horizons, Myakka fine sand in Figure 19 is shown for a depth of 200 cm, Immokalee fine sand in Figure 20 is for a depth of 100 cm, and Smyrna sand in Figure 21 is for a depth of 50 cm. Stratification is visible by the presence of abrupt breaks in the

PAGE 60

47 0 40 80 120 160 200 0 0.1 0.2 0.3 0.4 0.5 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Water Content Natural Saturation Porosity Water Table Figure 19. Soil Moisture Profile for Myakka Fine Sand 0 20 40 60 80 100 0 0.1 0.2 0.3 0.4 0.5 0.6 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Water Content Natural Saturation Porosity Water Table Figure 20. Soil Moisture Profile for Immokalee Fine Sand

PAGE 61

48 0 10 20 30 40 50 0 0.1 0.2 0.3 0.4 0.5 0.6 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Water Content Natural Saturation Porosity Water Table Figure 21. Soil Moisture Profile for Smyrna Sand water content profiles. An example calculation for the Immok alee fine sand soil profile is in Table 4; the total soil storage in all layers for the 100 cm water table depth summed to 18 cm. Individual storage values were found by the trapezoidal rule of integration. Final results for storage in the soils used fo r the proposed SCS SFWMD method are in Table 5 and are plotted in Figures 22 to 24. The figures show the result of three different storage calculations for each of the three soils. The soil storage capacity equation (7) was optimized by calculating the p arameters A B C and D (Table 19, Appendix A) from (8) that minimized the square of the difference (least squares) between the storage as determined by the rearranged I(d) equation (9) and the storage as determined manually by integration under the publi shed soil moisture characteristic curves (Table 4) from the IFAS (Carlisle et al. 1989). The latter storage values were

PAGE 62

49 corrected for air with the values in Table 1, and then plotted as points with diamond markers. To begin the optimization, the average Brooks and Corey parameters from fitted laboratory data were used to provide a first estimate of the constants A B C and D Table 4. Soil Storage Calculation for Immokalee Fine Sand (DTW = 100 cm) Horizon Capillary Head (cm) Water Content (%) Natura l Saturation (%) Porosity (%) Soil Section Thickness (cm) Storage (cm) 1 100 17.3 50.5 55.5 10 3.3 1 90 18.0 50.5 55.5 3 1.0 1 87 18.3 50.5 55.5 0 0.0 2 87 8.3 40.7 45.7 5 1.6 2 82 8.5 40.7 45.7 2 0.6 2 80 8.6 40.7 45.7 10 3.0 2 70 12.1 40.9 45.7 1 0 2.7 2 60 15.6 40.7 45.7 10 2.2 2 50 21.3 40.7 45.7 10 1.5 2 40 27.9 39.4 45.7 4 0.4 2 36 30.8 40.7 45.7 6 0.5 2 30 35.2 40.7 45.7 6 0.3 2 24 35.3 40.7 45.7 4 0.2 2 20 35.3 40.7 45.7 10 0.5 2 10 36.8 41.0 45.7 1 0.0 2 9 37.0 40.7 45.7 9 0.2 2 0 40.7 40.7 45.7 0 0.0 Once the regression was completed and the constants were optimized, the final Brooks and Corey parameters to be used in the analysis were back calculated from the constants and are located in Table 18 of Appendix A. The South Flori da Water Management

PAGE 63

50 Table 5. Soil Storage for Three Soils as a Function of Water Table Depth Depth to Water (cm) Myakka Storage (cm) Immokalee Storage (cm) Smyrna Storage (cm) 200 43.0 29.6 190 40.1 27.7 180 37.4 25.7 170 34.5 23.4 160 31. 4 39.0 21.0 150 28.5 36.0 18.9 140 25.8 32.3 16.9 130 23.1 28.5 15.0 120 21.1 25.1 13.2 110 19.4 21.6 11.6 100 17.5 18.1 10.0 90 15.0 14.8 8.4 80 11.9 11.6 6.4 70 9.0 8.7 4.4 60 6.5 5.9 3.0 50 4.3 3.5 2.1 40 2.7 1.8 1.5 30 1.7 0.9 0.8 20 1.0 0.4 0.3 10 0.3 0.1 0.1 0 0.0 0.0 0.0 District (2002) publishes soil storage values for flatwoods and depressional soils (triangle markers in the figures) that are comparable to the independently calculated storages for the three representative soils. Myakka fine sand and Immokalee fine sand best fit the flatwoods soil storages while Smyrna sand best fits the depressional storage values. For comparison purposes soil storage values both without air and corrected for air for all three soils are plotted i n Figure 25. The equation for the optimized line was applied to

PAGE 64

51 0 5 10 15 20 25 30 0 25 50 75 100 125 150 Depth to Water Table (cm) Soil Storage (cm) Corrected for Air Fitted Regression SFWMD Flatwoods Figure 22. Comparison of Three Soil Storage Calculations for Myakka Fine Sand 0 5 10 15 20 25 30 0 25 50 75 100 125 150 Depth to Water Table (cm) Soil Storage (cm) Corrected for Air Fitted Regression SFWMD Flatwoods Figure 23. Comparison of Three Soil Storage Calculations for Immokalee Fine Sand

PAGE 65

52 0 5 10 15 20 25 30 0 25 50 75 100 125 150 Depth to Water Table (cm) Soil Storage (cm) Corrected for Air Fitted Regression SFWMD Depressional Figure 24. Comparison of Three Soil Storage Calculations for Smyrna Sand 0 10 20 30 40 50 0 50 100 150 200 Depth to Water Table (cm) Soil Storage (cm) Myakka With Air Myakka Without Air Immokalee With Air Immokalee Without Air Smyrna With Air Smyrna Without Air Figure 25. Soil Storage Results for Three Soils

PAGE 66

53 each 30 m x 30 m grid cell in the GIS analysis based on which of the three soils was represented at that location. 3 .2 Initial Conditions The successful prediction of saturation excess runoff depends on an accurate assessment of the initial conditions in the catchment, such as the initial water table distribution. For catchments where Hortonian runoff is the dominant runoff mechanism, Hernandez (2000) found that depth to the water table is not a significant factor in runoff estimation; instead, rainfall intensity is the determining factor. In contrast, if saturation excess is the dominant runoff mechanism, the depth t o the water table is a determining factor in runoff estimation while rainfall intensity is insignificant, because saturation excess depends on cumulative rainfall depth and not rainfall intensity. The depth to the water table is required information for t he proposed method, and the choice of data will depend on whether the analysis is for design or for multiple events. Design based modeling considers the worst case scenario and simply requires the seasonal high water table distribution. This type of data is usually available from soil surveys or from the local water management district. Event based modeling requires a day by day water table distribution obtained only by heavy instrumentation and constant monitoring. The LFC subbasin and runoff test bed partially met this demand. In the case of WFHC, where there were no such data, a theoretical equation had to be implemented. The depth to the water table was used in Eq. (7), which was applied to the various soils represented by 30 m x 30 m grid cells in a GIS. The resulting grid identified areas in the catchment that were most likely to be saturated during a rainfall event.

PAGE 67

54 3.2.1 Depth to the Water Table The determination of d depth to the water table, in Eq. (7) prior to rainfall at any point in tim e is a challenging task. Spatial variation in soil texture and soil layering in a catchment causes a nonuniform water table distribution. Adding to the complexity is a non steady state, dynamic water table that fluctuates during the actual rainfall event In the absence of a large number of shallow wells, physical properties of the landscape may be used to derive an estimate of water table depth. In particular, the topographic index, ln (a/ tan b ) has been widely used in hydrologic modeling to determine whether a uniform area like a cell in a grid based analysis is more or less likely to retain soil moisture. The TOPMODEL approach for variable source areas (Beven and Kirkby 1979) defines a as th e contributing upstream area per unit contour and tan b as the local slope. Willgoose and Perera (2001) noted that the predictive accuracy of this and other saturation excess runoff models depends on digital elevation model (DEM) grid resolution and the a nalysis procedures used on the DEM data. In this study, TOPMODEL concepts were used to find the depth to the water table. The topographic index acts as an index of hydrologic similarity, meaning that all grid cells with the same index value are thought to behave in a hydrologically similar manner. With this assumption, Beven (1997) concluded that calculations are not required for all points in a catchment, but only for different topographic index values. Equations derived from the TOPMODEL approach are provided by Sivapalan et al. (1987), who presented the following relationship:

PAGE 68

55 f Q Q z = 0 ) 0 ( ln (10) where: z equals the mean water table depth, Q (0) equals the baseflow at the outlet just prior to a storm event, Q o is a parameter of the baseflow recession curve, and f is a rate constant The rate constant, f is related to the exponential decrease in saturated hydraulic conductivity with depth in the soil profile. The settling of clays and silts in a sand profile t ends to decrease the hydraulic conductivity with depth due to lower porosity and poorly graded conditions. The county survey soils data in this study (Table 20, Appendix A) suggested a value for f of approximately 0.0125 cm 1 as determined from the expre ssion ) exp( ) ( fz K z K o s = (11) where: z equals the depth into the soil profile, K s equals the saturated hydraulic conductivity, and K o equals the saturated hydraulic conductivity at the soil surface

PAGE 69

56 Figure 26 shows an example conductivity profile for Myakka fine sand, a prolific sand found in Florida flatwoods, along with Eq. (11) plotted for f = 0.0125 cm 1 The conductivity data was taken from Robbins et al. (1984) because WFHC is located partly in Hardee County. 0 50 100 150 200 250 0 5 10 15 20 25 30 K s (cm/hr) Depth (cm) Lab Data Equation 11 Figure 26. Saturated Hydraulic Conductivity for Myakka Fine Sand Q o is given by Sivapalan et al. (1987) as ) exp( l = e o AT Q (12) where A equals the total catchment area and l and T e are expressed as

PAGE 70

57 dA a A A = b l tan ln 1 (13) = A o e dA T A T ln 1 ln (14) where: a equals the area draining through a location per unit contour length tan b equals the slope of the ground surface, and T o equals the transmissivity coefficient of an aquifer profile Equation (14) is an area average of the transmissivity coefficient, defined as f K T o o = (15) The distribution of the water table may be predicted by (Sivapa lan et al. 1987) { } e o i T T a z z f ln ln tan ln ) ( = l b (16) where z i is the depth to the water table, or d in (7), at a point in the catchment. In this study a point was represented by a 30 m x 30 m pixel in a DEM. Equation (16) represents, in dime nsionless form, the deviation of the local depth to water table from the

PAGE 71

58 average depth to water table in terms of the deviation of the logarithm of the transmissivity and local topographic index away from their respective area averages over the catchment. For any given z a value of z i less than or equal to the depth of the capillary fringe, or h A from (7), reveals the predicted area of saturation, or where a variable source area will appear. Equation (10) may be omitted if there is a known z i for the day of interest, such as from a shallow water table well reading. The average water table depth z may then be solved for in (16). The incorporation of observed data usually brings model results closer to actual res ults. For example, Seibert et al. (1997) improved poor simulated water table results by using spatially distributed groundwater observations to calculate new topographic soil indices. The relationships presented are based on several assumptions: the wa ter table is parallel to the ground surface the recession discharge prior to a storm results from a steady rate of recharge to the water table, and the saturated hydraulic conductivities within the soil profile exhibit an exponential decline with depth Th e first assumption is supported by Spier et al. (1969), who noted that the ground surface and water table surface mimic each other. Figure 27 shows a small water table gradient at Long Flat Creek, where the land surface and the water table are relatively parallel before and after a storm. The third assumption has been observed in studies where there is a retarding layer 2 to 3 feet below the surface of Florida soils (Capece

PAGE 72

59 1984). This can be seen to a degree in the saturated conductivity data presented in Figure 26 and Appendix A. The low conductivity layers cause poor drainage, resulting in the upper layers being quickly saturated. Water Table Response of Wells PS-5 to PS-2 to 5.18 cm of Rain Storm Began 2/22/02 10:05 and Ended 2/23/02 22:25 22 23 24 25 26 0 20 40 60 80 100 120 140 Distance from Creek (m) Elevation (m) Land Surface 2/22/2002 9:55 2/23/2002 22:15 Figure 27. Small Water Table Gradient Near Long Flat Creek As mentioned before, the expression ln ( a / tan b ) in (13) is the topographic index, a widely used geomorphologic parameter that determines the potential soil moisture at a location in the catchment. If the location has a large upstream contributing area and a flat slope, then that area will retain mor e soil moisture than a ridge area with a steep slope. The topographic index is useful in evaluating variable source areas because it indicates where saturated areas will occur. However, Jordan (1994) found that the use of the

PAGE 73

60 topographic index does not p rovide entirely adequate results because there is a lack of coherency between the topographic index and water table depths. Water table elevations show no systematic behavior and the only time a correlation between the topographic index and water table le vels can be expected is during saturated moisture conditions, when the water table reaches a stable condition. It may well be impossible to determine water table depths over a catchment, even if wells are available for direct observation at a few location s. Lamb et al. (1997) used a variation of the TOPMODEL approach, and the simulated water table depths only approximated the observed depths and did not reproduce the actual local variations. Moore and Thompson (1996) modified the TOPMODEL concept by usin g a linear model to express the time and location effects on water table depth independently, which provided a good fit to observed water table depths in a very small (0.04 km 2 ) forested catchment with shallow soil. In a future study it would be interesti ng to see if this linear model predicts the water table distribution of catchments in this research better than Eq. (16). A good critique of other studies that have applied an original and/or modified TOPMODEL approach is provided by Beven (1997). As exp ected, the studies had mixed results due to catchment heterogeneity and different assumptions. Figure 28 is an example of the application of Eq. (16) to WFHC. Unfortunately, only 1.52 m (5 ft) digital contours were available for the basin, which produced a digital elevation model with numerous flat areas after using a triangulated irregular network (TIN). The preferred vertical resolution would have been 0.30 m (1 ft). However, a water table distribution was needed and the errors produced by the 1.52 m (5 ft) contours, such as saturation near the eastern ridge, were ignored.

PAGE 74

61 Figure 28. Depth to Water Table Distribution for West Fork Horse Creek 3.2.2 Baseflow Baseflow is the component of streamflow that consists of groundwater i nflow. It is considered to be a process that acts on a slow time scale (Spier et al. 1969). Typically catchments with moderate to steep slopes and high water tables produce the highest baseflow because there is a large gradient to drive groundwater to th e streams. Because the TOPMODEL approach relies on the existence of steady baseflow from gravity drainage, it is best applied in catchments with these conditions. Nevertheless, an attempt

PAGE 75

62 was made to use Eq. (16) in a flatly sloped environment because th ere was no better alternative for deriving a physically based water table distribution at the time. The successful application of Eq. (16) is dependent on the estimation of baseflow at the outlet in (10). Knowledge of baseflow is also needed in hydrograph separation in order to extract the observed runoff. It is therefore unfortunate that baseflow is difficult to quantify. Filter programs or graphical methods may be applied to measured streamflow data, but these are often arbitrary. Floridas terrain is extremely flat; therefore, one would expect to use a low baseflow value in (10), because there is not much slope to drive groundwater to the streams. The approach often used to measure this baseflow is to observe streamflow in the dry season when there h as been no rainfall for an extended period of time. Baseflow does not remain constant, but tends to increase during a storm when a rising water table increasingly intersects with the channel. In this study, WFHC and LFC were observed to have large areas o f standing water, either in the form of natural wetlands or old phosphate pits. Capece (1984) was studying a similar flatwoods environment when he postulated that the vegetation and open water consume baseflow before it reaches the streams. Indeed, the f lat slope of the study area slows the movement of groundwater to the point that transpiring plants and evaporating ponds may remove would be baseflow. Figure 27 depicts a very low water table gradient (slope of 0.2%) between the two wells nearest the chan nel of LFC, which happen to be located in dense trees and shrubs. A feasible water table distribution for WFHC from (16) was achieved only with baseflow values that were less than 0.028 m 3 /s (1 cfs). As a check, Darcys Law was applied to WFHC for a range of values by the expression

PAGE 76

63 dx dh KA Q = (17) where: Q equals the groundwater discharge through both banks of the stream, K equals saturated hydraulic conductivity, A equals the area along the banks receiving groundwater, an d dh/dx equals the water table gradient (set equal to the land surface slope) The average K for the basin was approximately 22 cm/hr. The average slope was varied between 0.002 m/m and 0.006 m/m while also varying the depth of flow through the banks, and the value for groundwater inflow ranged from 0.014 m 3 /s (0.5 cfs) to 0.028 m 3 /s (1.0 cfs). The South Florida Water Management District (SFWMD) currently excludes baseflow from its hydrologic models because south Florida is extremely flat and has enormous areas of open water. To bypass this difficulty Eq. (10) was not used; instead, the initial water table depth from the well shown in Figure 28 was set equal to z i in order to solve for z in (16). Basing the entire water table distri bution on a well near the edge of the catchment adds to the uncertainty in the runoff estimate, but it is still superior to using an unknown baseflow value. Baseflow becomes more significant in WFHC and LFC during the late wet season because the observed runoff exceeds the rainfall if baseflow is set to zero in the streamflow series. Therefore a moving minimum filter program developed by CMHAS (Perry 1995) was applied to the data to estimate baseflow and arrive at an observed runoff value.

PAGE 77

64 3.3 Runoff Est imation The calculation of soil storage capacity based on initial depth to the water table and the initial soil moisture profile is necessary to reach the final goal, the estimation of runoff in a saturation excess environment. For comparison purposes, o bserved runoff from a streamflow gage at a catchment outlet may be examined alongside the simulated runoff generated by the proposed method. A streamflow hydrograph may be available to obtain the observed runoff, but the actual estimation of that runoff i s not an easy task. 3.3.1 Hydrograph Separation A streamflow hydrograph from a Florida catchment is composed of baseflow, direct runoff, and usually wetland attenuated runoff. As discussed in the previous section, baseflow in the observed hydrographs w as found by a filter program to expedite the runoff estimation process. The remaining two components, runoff and wetland storage release, are difficult to separate from one another. Typically the direct runoff is located in the area under the rising limb and the peak of the hydrograph while the volume of slow drainage from the wetlands or other surface storage is found under the long recession limb (tail) that is characteristic of a hydrograph in Florida. In order to avoid adding complexity with storage routing models, areas of surface storage may be viewed as a source of immediate saturation excess since the soil storage capacity has been filled and all rainfall on these areas are assumed to become runoff in the wet season. The combination of direct run off and surface storage release may therefore represent the total amount of saturation excess runoff recorded by the hydrograph. It is then necessary to assign which part of the streamflow record belongs to which storm. This assignment is

PAGE 78

65 not so difficul t in the dry season, when frontal storms are separated by days of no rain and individual hydrographs are distinguishable. The difficulty increases in the wet season, when it often rains everyday and multiple, closely spaced peaks appear in the streamflow hydrograph. Often an arbitrary line of separation must be drawn to separate storms. Figure 29 shows how individual storms were identified in this study for both the dry season and the wet season. A line was extended downward to the current baseflow valu e from the point on the recession limb just before the streamflow record started on the next rising limb. The slope of the line was determined by taking the logarithm of the recession limb values, plotting them versus time, and finding the slope of the li near Hydrograph Separation for January 8, 1993 Storm 0 0.1 0.2 0.3 0.4 0.5 0 30 60 90 Time (hours) Streamflow (m 3 /s) Second storm begins Figure 29. Storm Separation Method for West Fork Horse Creek

PAGE 79

66 trendline through these points. Hydrograph separation in the wet season is more difficult because there are often several peaks lumped together over a few days. The runoff for a part icular storm was identified in the manner described above, though two or three recession tails were often stacked on one another in the plot. 3.3.2 Surface Storage Florida basins often contain large depressional areas, such as natural wetlands and mars hes. Many sinkholes dot the landscape as well, forming lakes and ponds. Additionally there are extensive areas of phosphate mining, where there are old excavated holes that later become lakes. A good percentage of the drainage basin area is often occupi ed by these various types of depressions, and therefore they must be accounted for in any runoff estimation method. Technically, flow through a series of connected wetlands or lakes may be determined with a reservoir flood routing model. However this kind of routing requires knowledge of the storage in the reservoirs, which means that details about the area and stage of the wetlands or ponds would be needed. Relatively few of the vast number of depressional areas in Florida are monitored; those in LFC and WFHC are no exception. With the lack of wetland storage details, another approach may be used. As mentioned earlier, areas that are already saturated with water were deemed ready for immediate runoff production. Bedient et al. (1976) assumed otherwise in their HLAND model for Floridas depressional watersheds, which includes detention constants for a fraction of the excess rainfall that does not immediately become overland flow. Though it is true that Florida basins retain volumes of excess water, thi s was ignored to

PAGE 80

67 avoid added complexity and parameters. In the late wet season, it was assumed that most of the wetlands were near their maximum storage capacity and contributed runoff during a storm. These areas of zero soil storage capacity were viewed as impervious areas and were separated into two categories: directly connected and conditionally connected. Depressions near the stream were considered directly connected because it was observed that wetlands fringing WFHC were discharging through swale s into the channel after a storm. A land use coverage in a GIS is helpful in determining chains of connected lakes or wetlands that discharge directly to the stream. In the uplands there are sometimes conditionally connected depressions that simply store water until they overtop, after which water travels as overland flow to the next water body downstream. For the purposes of this study the catchment area was divided into unsaturated and saturated regions, depending on the status of a particular uniform area just before a storm. Saturated areas included all wetlands, marshes, ponds, and other impervious areas, such as unpaved grid cells with a depth to water less than the value of h A (approximately equal to the capillary fringe) as determined by (16) or by well readings. Unsaturated areas comprised the rest of the basin that still had available soil moisture storage. Because of (6), the S parameter only applies to areas with a positive value for I where some soil storage capacity remains to be filled. Therefore a value called Q scs was calculated from (1) using (6) for unsaturated areas and was added to the runoff for impervious areas expressed by imperv a imperv A I P Q ) ( = (18)

PAGE 81

68 where: P equals the storm rainfall depth, I a e quals the initial abstraction due to interception only, and A imperv equals the fraction of impervious (saturated) area The total runoff was found by ( ) imperv perv scs Q A Q Q + = (19) where A perv is equal to the fraction of the basin ar ea that is pervious. Although this approach was applied to rural catchments, Valeo and Moin (2001) discuss other variable source area models that separate impervious and pervious areas in an urban setting. These models used Eq. (1) to calculate runoff vo lume in the pervious areas and Eq. (18) for the impervious areas.

PAGE 82

69 CHAPTER 4. APPLICATION OF THE SCS SFWMD METHOD 4.1 Introduction The SCS SFWMD method was tested on a small scale and at a larger catchment scale. A runoff test bed in the Long Flat Cree k (LFC) basin was useful for study of the saturation excess mechanism on a small scale, while LFC and West Fork Horse Creek (WFHC) represented small and large basins at the catchment scale. 4.2 Runoff Test Bed The runoff test bed provided a close up vi ew of saturation excess at work. Two soil moisture probes and four wells in the 185.9 m 2 (2000 ft 2 ) area reduced uncertainty in the water table distribution and in the soil moisture storage capacity. A rain gage located on the downstream end recorded at a five minute frequency to retain accuracy in storm intensity. Chapter 2 describes the equipment in more detail. 4.2.1 Application of the Method Four data series were retrieved at the test bed from the data loggers for each storm: rainfall, water table elevation, soil moisture, and weir head. The weir head was converted into discharge representing rainfall excess from the test bed. Cumulative rainfall was calculated along with cumulative runoff. The relationship between weir head and discharge was re corded in the laboratory before the weir box was installed so that

PAGE 83

70 measured head could be converted to a runoff volume (Figure 30). Final soil moisture profiles after a storm were constructed by subtracting values from Table 2 from the porosity to adjust for air encapsulation. To find soil storage, the area between the effective porosity (natural saturation) line and the water content profile was calculated using the trapezoidal rule. The result from this method of calculation was compared to the storage as calculated from the difference between the actual recorded initial and final moisture profiles. It was necessary to use the former method of determining storage when the moisture probes failed to record the final profile. The test bed was divided int o two equal areas so that the soil storage values from the downstream probe were assigned to the downstream half and the values from the upstream probe were assigned to the upstream half. Water table elevations were assigned in a similar manner, so that e ach half of the test bed was assigned one elevation. The test bed is best represented by the two bucket model, shown before in Figure 12. The only equation needed was (6) to find the area averaged soil storage capacity. There were no impervious areas in the runoff test bed until the surface became saturated.

PAGE 84

71 Figure 30. Rating Curve for the Test Bed Weir 4.2.2 Results Figure 31 shows the rainfall and runoff at the test bed for a storm occurring on June 25, 2002. Figures 32 a nd 33 present the actual initial soil moisture profiles for USF 1 and USF 3, respectively, just before the storm. The final profile was estimated by subtracting the average air encapsulation from the porosity. The points at land surface and at the water table in the initial profile do not represent actual sensor readings, but were approximated in order to calculate soil storage. The soil moisture at the water table does not increase as it does at the other points because the location is assumed effective ly saturated at the start of the storm. Larger amounts of air encapsulation are visible deeper into the profiles. The initial water table reading is provided, as well as the final water table shown at the land surface. The behavior of the water level in the surficial wells was sometimes difficult to interpret. Besides the problem of drift in the transducer readings, y = 3.2746x + 0.9487 R 2 = 0.9908 -4 -3 -2 -1 0 -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 log Head log Discharge

PAGE 85

72 water levels continued to rise after a storm despite lack of rain, and even went into negative depth to water readings when the ground was originally close to saturation. Even though some water table readings for USF 1 and USF 3 indicated that by the end of the storm the water table had not reached the surface, field evidence of saturated soil and water levels at land surface in non recordin g wells suggested that the ground had indeed been saturated. The transducer readings that did not reflect a rise of the water table to land surface were likely influenced by pressurization in the soil matrix (see Chapter 3) and the effects of lateral flow through macropores in the top layer of the soil, which contains roots and organic material. 0 1 2 3 4 5 6 6/25/02 12:00 6/25/02 14:24 6/25/02 16:48 6/25/02 19:12 6/25/02 21:36 6/26/02 0:00 Cumulative Rainfall (cm))) 0 0.0004 0.0008 0.0012 0.0016 Runoff (m 3 /s) Cumulative Rainfall Runoff Figure 31. Rainfall and Runoff at the Test Bed for June 25, 2002

PAGE 86

73 USF-1 Soil Moisture Profiles Storm begins 6/25/02 14:35 and ends 18:00 0 10 20 30 40 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Initial Profile Final Profile Lab Porosity Initial Water Table Final Water Table Figure 32. Soil Moisture and Water Table Data at USF 1 for June 25, 2002 USF-3 Soil Moisture Profiles Storm begins 6/25/02 14:35 and ends 18:00 0 10 20 30 40 50 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Initial Profile Final Profile Lab Porosity Initial Water Table Final Water Table Fi gure 33. Soil Moisture and Water Table Data at USF 3 for June 25, 2002

PAGE 87

74 After the June event was recorded, the test bed weir was inoperable until September, when another storm was analyzed. The test bed weir continued to malfunction despite changes in th e design and setup. If the problem is resolved in the future, more events will be analyzed in an effort to further pursue the research. Table 6 presents the observed rainfall, area averaged soil storage, area averaged initial water table depth, and runo ff for the recorded storms. Simulated runoff was understood to be the average soil storage subtracted from the rainfall. If the soil storage was greater than the rainfall, then no runoff was expected. The simulated runoff shown in the table is based on the method of soil storage calculation where the average air encapsulation is subtracted from the porosity to obtain the final profile. The runoff as calculated from the difference between the actual initial and final soil profiles was 3.51 cm for June 25 and 0.0 cm for September 14. These rainfall events demonstrated the saturation excess mechanism because no Hortonian runoff was observed, i.e., there was no significant runoff unaccounted for since the simulated runoff was quite close to the observed run off. The runoff was almost exactly the rainfall minus the soil storage. Table 6. Results for the Runoff Test Bed Date of Storm Rainfall (cm) Average Soil Storage (cm) Average Water Table Depth (cm) Observed Runoff (cm) Simulated Runoff (cm) 6/25/02 5.13 1.18 35.7 3.99 3.95 9/14/02 2.03 0.0 0.0 2.03 2.03

PAGE 88

75 4.3 Long Flat Creek The LFC subbasin provided a small catchment to apply both the heavily instrumented approach used on the runoff test bed and the generalized topographic index approach used for WFHC. An event occurring on June 25, 2002 was particularly suited for the study of the saturation excess mechanism. Because the runoff test bed resided in the subbasin, its data was added to the data from the subbasin. 4.3.1 Application of the Method The first runoff estimation method tried for the June 25th event was similar to the test bed method. One major difference in its application was the assignment of soil storage capacity to areas that were non instrumented. Vegetation along LFC was thick and apparently used much of the local groundwater for transpiration. Wells located near trees usually had deeper water table depths than wells without trees, regardless of proximity to the creek. Figure 34 shows the water table on the east side of LFC be fore and after the June 25, 2002 storm. PS 5 and PS 4, the wells nearest the creek, had a similar or deeper water table than the water table at PS 2, located in the middle uplands. The west side demonstrates the deep riparian water table even better than the east because there are more trees on the west side of the creek. For example, the test bed wells USF 1 and USF 3, located in open pasture, recorded shallower water table depths and had correspondingly less soil storage than PS 40, PS 41, and PS 42 (T able 7 and Appendix B), even though the test bed was further from the stream than these other three wells. PS 40 and PS 41 in particular showed more available storage than at other sites because of their location in the thickest part of the vegetation. H eterogeneity in soils may account

PAGE 89

76 for the difference, but in Figure 35 the same trend was noticed in the upland wells, where the water table depth decreased as distance from the stream increased. This figure also reveals the presence of a baseflow sink ne ar LFC. The difference in soil moisture storage values, as listed in Table 7, is even more significant than the difference in the water table depths. The LFC subbasin was divided into east and west portions since there was one rain gage on each side (one near USF 1 and one near PS 1), and only the west side contained soil moisture probes. There was one soil moisture probe per well on the west side, except for PS 39. Thiessen polygons were used to assign soil moisture to regions around the various wells, while consideration was also given to similarity in vegetation type, land cover, distance from the creek, and soil texture (Figure 36). Soil moisture for areas with probes was estimated at the ground surface and at the water table because sensors were not at these locations. Soil moisture on the east side was calculated for each region assigned to the wells by inserting the observed water table depth on the day of interest into Eq. (7) for the particular soil where the well was located. The equation was used because there were no moisture probes on the east side. All wells and probes are located in Smyrna sand as defined by the simplified classification system in this study, except for PS 2, which is in Myakka fine sand. The reliability of the equation was first tested by applying it to each region on the west side in the same manner as the east side. Table 7 lists the area percentage of each portion of the two sides of the catchment and the corresponding soil storage for June 25, 2002. Using the equat ion on the west side produced reasonable results when compared to the storage as calculated from the soil moisture probe data. Appendix B provides supplementary soil

PAGE 90

77 21 22 23 24 25 26 27 28 0 40 80 120 160 200 240 280 Distance from Creek (m) Elevation (m) Land Surface 6/25/2002 14:35 6/26/2002 13:15 Figure 34. Water Table Gradient from PS 5 to PS 1 for June 25 26, 2002 20 21 22 23 24 25 26 27 0 50 100 150 200 250 300 Distance from Creek (m) Elevation (m) Land Surface 6/25/2002 14:35 6/26/2002 13:15 Figure 35. Water Table Gradient from PS 39 to PS 43 for June 25 26, 2002

PAGE 91

78 Figure 36. Delineation of Areas for Assignment of Soil Moisture Storage moisture and porosity data. The area fractions were multiplied by the observed probe storage for the west side and m ultiplied by the Eq. (7) storage for the east side. Individual areas on each side of the basin were summed to calculate the percentage of area the east and west sides occupied in the entire basin. The west side, excluding

PAGE 92

79 wetlands, comprised 72% of the e ntire basin, the east side comprised 25%, and wetlands on the west side comprised 3%. Two different rainfall amounts were recorded for the June 25, 2002 storm and were apportioned by location in the basin: 5.13 cm fell on the west side of the basin while 4.55 cm fell on the east side. Table 7. Area Weights and Soil Moisture Storage at Long Flat Creek for 6/25/02 Area (%) Equation (7) (cm) Probe Soil Storage (cm) Area Weighted Soil Storage (cm) West LFC PS 43 57 3.25 2.61 1.49 PS 42 4.4 5.02 4. 98 0.22 PS 41 7.5 6.73 7.42 0.56 PS 40 4.6 9.03 7.06 0.32 USF 1 13 0.97 1.76 0.23 USF 3 9.4 1.65 0.60 0.06 Wetlands 4.1 0.00 0.00 0.00 Sum 100 2.88 East LFC PS 5 24 1.78 0.43 PS 4 16 6.67 1.07 PS 3 18 9.59 1.73 PS 2 25 6.24 1.56 PS 1 17 17.27 2.94 Sum 100 7.73

PAGE 93

80 The east side, west side, and wetlands were considered to be separate units. From Table 7, the west side had an area averaged soil storage value of 2.88 cm while the east side had a value of 7.73 cm. The r unoff from these areas was the difference between their respective rainfall and storage. The wetlands were considered to have only an initial abstraction for storage, assuming they were full in the wet season. As noted before, the value for initial abstr action in flatwoods used by CMHAS is 0.25 cm. Therefore the total runoff for the June 25, 2002 storm was calculated as (0.72 (5.13 2.88) + 0.25 0 + 0.03 (5.13 0.25)), which equals 1.77 cm. The east side area percentage was multiplied by zero be cause the storage was greater than the rainfall. This value was compared with the observed runoff (2.36 cm), which was calculated as the difference in discharge between the upstream and downstream weirs at the LFC subbasin. The second runoff estimation method, the SCS SFWMD method, was also applied to WFHC, which is described in the next section. One advantage for LFC was that 0.305 m (1 ft) contours were available for the subbasin instead of 1.52 m (5 ft) contours to derive a DEM (Figure 37). Myakka f ine sand and Smyrna sand were the representative soil types used in the subbasin; properties of the soils are provided in Appendix A. The observed runoff is an estimate based on the assumption that the weir equation applied to the stage data from the weir s is correct. Stage associated with low flows was recorded by a V notch, and then at 0.30 m (1 ft) of head above the notch, the weir transitioned into a rectangular configuration to record high flows. Average rainfall for a storm was determined by area w eights from Thiessen polygons: 75% of the basin area was assigned to the west rain gage and 25% was assigned to the east rain gage.

PAGE 94

81 Figure 37. Digital Elevation Model of the Long Flat Creek Subbasin A grid was generated with soils information and depth to the water table as described in Chapter 3. Baseflow was not used to find z ; the result was unreasonable. Instead all the pre storm observed water table depths ( z i ) from the wells in the basin were used to calculate their respecti ve z and then these results were averaged to obtain a final value for z Figure 38 presents the final water table distribution. A value of z i equal to zero indicates that the water table is at the land surface, while a negative value was understood to represent standing water. Similar to the results for West Fork Horse Creek,

PAGE 95

82 there were areas of saturation near the ridges in the basin. A coarse grid resolution (30 m x 30 m) increases the topographic index in t he TOPMODEL approach by artificially increasing the upstream area at a location. A higher topographic index indicates a higher probability of saturation. The creek bed is correctly marked by saturated cells. The most apparent problem in the figure is th e presence of water table depths over 2 meters, which is not possible when compared to well readings for June 25, 2002. There were some edge effects brought over from Research Systems RiverTools, where the grid cells comprising the boundary around the bas in contained lower topographic index values, resulting in deeper water table depths. These cells decrease the average topographic index, which increases the average water table depth. Figure 38. Water Table Distribution in the Long Flat Creek Subbasi n

PAGE 96

83 Soils in the basin were renamed according to soil texture and were classified as either Myakka fine sand or Smyrna sand. The regions containing these two sands were treated separately during all steps of the process. For Myakka fine sand grid cells, Eq (7) was applied wherever water table depths were greater than 18.09 cm, which was the bubbling pressure found by the regression procedure discussed in the latter part of Chapter 3 (see also Table 18 of Appendix A). For pixels containing Smyrna sand, the bubbling pressure was 28.27 cm, and the soil storage equation was applied to pixels wherever the water table depth was larger. The I(d) value for each 30 m x 30 m grid cell was found in this manner. The sum of the storage in all of the pixels in the bas in for Myakka fine sand was 76.1 m and 274.3 m for Smyrna sand, as computed by the GIS. These two totals were added together to finally obtain 350.4 m for I in Eq. (6) Table 8 summarizes chronologically the process of runoff estimation from this point w ith the necessary equations and their associated parameters. Table 8. SCS SFWMD Method Parameters for June 25, 2002 at Long Flat Creek Equation (6) Value Equation (1) Value Equation (18) Value Equation (19) Value A (m 2 ) 417,600 P (cm) 4.99 P (cm) 4.99 Q imperv (cm) 2.08 I (m) 350.4 I a (cm) 0.25 I a (cm) 0.25 Q scs (cm) 0.28 da (m 2 ) 900 S (cm) 75.5 A imperv 0.44 A perv 0.56 S (m) 0.75 Q scs (cm) 0.28 Q imperv (cm) 2.08 Q (cm) 2.24

PAGE 97

84 It is apparent that the vast majority of the runoff originated from th e areas already saturated before the storm, such as the wetlands and uplands that had no remaining soil moisture storage. The estimated runoff in this case was 2.24 cm, which is quite close to the observed runoff of 2.36 cm. 4.3.2 Results Figure 39 shows the runoff hydrograph and cumulative rainfall at both rain gages for June 25, 2002. The hydrograph is not smooth because it represents the difference between two streamflow stations, which are not likely to respond in exactly the same manner at the same time. 0 1 2 3 4 5 6 6/25/02 12:00 6/25/02 14:24 6/25/02 16:48 6/25/02 19:12 6/25/02 21:36 6/26/02 0:00 Cumulative Rainfall (cm))) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Runoff (m 3 /s) USF-1 Rain PS-1 Rain Runoff Figure 39. Rainfall and Runoff at Long Flat Creek for June 25, 2002

PAGE 98

85 Figures 40 to 43 provide soil moisture profiles in addition to those for USF 1 and USF 3 previously shown in the results section for the runoff test bed. Table 9 presents the observed rainfall, area averaged initial water table depth, and runoff for the June 25, 2002 storm. The Method column indicates which of the two approaches produced the given results. The method similar to the test bed method that used all instrumenta tion and employed a soil moisture accounting approach was the most site specific and is labeled Site, while the SCS SFWMD method is labeled Proposed. The Site method average water table depth was found by multiplying the area weights from Table 7 fo r each side by the pre storm water table depths (Table 25, Appendix B), then multiplying the west side result by 75% and the east side result by 25%, and finally adding these two results. For the Proposed method, the z value of 141 .3 cm was multiplied by 56%, representing the unsaturated area. The other part of the basin was saturated and had a water table depth of effectively zero. The simulated runoff value for the site method was not as close as the SCS SFWMD (proposed) method t o the observed runoff. This was unexpected since there was more data available for the site method and the average water table was deeper for the proposed method. However, the vast majority of runoff in the proposed method came from saturated areas so th at correct soil storage calculation was not as important as in the site method. In addition, the rating curve used for the upstream and downstream weirs may be a source of error so that the observed runoff may in reality be closer to the site method value Nevertheless there is general agreement between both methods for this particular storm, which had wet initial conditions.

PAGE 99

86 PS-43 Soil Moisture Profiles Storm begins 6/25/02 14:35 and ends 18:00 0 20 40 60 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Initial Profile Final Profile Lab Porosity Initial Water Table Final Water Table Figure 40. Soil Moisture and Water Table Data at PS 43 for June 25, 2002 PS-42 Soil Moisture Profiles Storm begins 6/25/02 14:35 and ends 18:00 0 20 40 60 80 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Initial Profile Final Profile Lab Porosity Initial Water Table Final Water Table Figure 41. Soil Moisture and Water Table Data at P S 42 for June 25, 2002

PAGE 100

87 PS-41 Soil Moisture Profiles Storm begins 6/25/02 14:35 and ends 18:00 0 20 40 60 80 100 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Initial Profile Final Profile Lab Porosity Initial Water Table Final Water Table Figure 42. Soil Moisture and Water Table Data at PS 41 for June 25, 2002 PS-40 Soil Moisture Profiles Storm begins 6/25/02 14:35 and ends 18:00 0 20 40 60 80 100 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 Soil Moisture (cm 3 /cm 3 ) Depth (cm) Initial Profile Final Profile Lab Porosity Initial Water Table Final Water Table Figure 43. Soil Moisture and Water Table Data at PS 40 for June 25, 2002

PAGE 101

88 Table 9. Results for the Long Flat Creek Subbasin Date of Storm Method Rainfall ( cm) Average Water Table Depth (cm) Observed Runoff (cm) Simulated Runoff (cm) 6/25/02 Site 4.84 60.3 2.36 1.77 6/25/02 Proposed 4.84 79.1 2.36 2.24 For further comparison, the original SCS method was applied to the June 25, 2002 rainfall event. The re was over 5.0 cm of rainfall the day before the event, so the antecedent moisture condition (AMC) would be III, or wet. Various sources define AMC and curve number (CN) conversion differently; values used here are based on those published by McCuen (199 8). As noted before in Chapter 1, Trommer et al. (1996) listed CNs for five natural Florida watersheds, and the average of these is 72.4. Using this CN as a base, the CN adjusted for AMC III conditions becomes 89. From Eq.(2) S is found to be 3.14 cm, a nd after inserting into Eq.(1), Q is found to be 2.41 cm, which is extremely close to the observed runoff. It appears that the SCS approach in general works well for this storm. 4.4 West Fork Horse Creek The WFHC basin was the largest catchment used in the study. Details about the hydrologic conditions at WFHC are provided by Lewelling (1997). Spatial variation in rainfall and soils contributed to error and uncertainty in the analysis. Difficulty in estimating the water table distribution resulted in soil storages that were too high to

PAGE 102

89 produce enough runoff to equal the observed value. However the results may explain some of the variability in the catchment. 4.4.1 Application of the Method Unlike the test bed and the LFC subbasin, there was a lack o f instrumentation at WFHC. Streamflow was obtained from the USGS gage West Fork Horse Creek near Myakka Head. Baseflow was estimated with a filter program called BASEFLOW, developed by the Center for Modeling Hydrologic and Aquatic Systems (CMHAS) at USF Storms were selected and analyzed according to the methods described in Chapter 3. The initial abstraction in Eq. (1) and (18) for forested wetlands was set to 0.254 cm (0.1 in.), a value determined in Florida basins by CMHAS for hydrologic modeling. The original SCS method requires an initial abstraction that includes infiltration occurring before Hortonian runoff commences. In the proposed method for this study, the initial abstraction refers to interception and depressional storage only. Soil sto rage and the water table distribution were found by the methods in Chapter 3. The DEM in Figure 44 was generated using topographic shapefiles of 1.52 m (5 ft) contours from the Southwest Florida Water Management District (SWFWMD) and then was input into R iverTools to calculate the topographic index in every 30 m x 30 m cell. To calculate a in (13), the contributing area to a pixel was divided by the width of the cell, or 30 m. Equation (8) was applied with the appropriate constants (Appendix A) using ESR Is ArcView 3.2.

PAGE 103

90 Figure 44. Digital Elevation Model of West Fork Horse Creek 4.4.2 Results Table 10 lists the soil storage and S parameter as determined by GIS analysis for various storms in WFHC. Table 11 presents the corresponding runoff results. The selected storms had rainfall depths greater than 3.81 cm (1.5 in.) because of the inherent

PAGE 104

91 nature of the SCS method being designed for large events. A high water table usually developed in the wet season, and thus most saturation excess occurred dur ing the summer months. The simulated runoff value for the 9/15/94 9/16/94 event was the worst while the runoff estimation for the 4/1/93 event was quite close to the observed value. Table 10. Soil Storage at West Fork Horse Creek Date of Storm Fractio n of Unsaturated Area (%) Sum of Basin Storage, I (m) Average Soil Storage, S (cm) 1/25/93 1/27/93 61 8113 36.0 3/25/93 3/26/93 58 7059 33.2 4/1/93 56 6624 32.1 6/15/94 74 11723 43.3 8/28/94 65 8487 35.8 9/15/94 9/16/94 53 5767 29.8 9/27/94 9/28/94 38 2938 21.1 Table 11. Results for West Fork Horse Creek Date of Storm Rainfall, P (cm) Unsaturated Area Runoff, Q scs (cm) Saturated Area Runoff, Q imperv (cm) Simulated Total Runoff, Q (cm) Observed Runoff, Q obs (cm) 1/25/93 1/27/93 4.14 0.38 1.50 1.73 1.42 3/25/93 3/26/93 4.19 0.42 1.66 1.90 3.14 4/1/93 5.47 0.73 2.28 2.69 2.78 6/15/94 5.94 0.66 1.49 1.98 0.88 8/28/94 4.73 0.50 1.58 1.90 2.93 9/15/94 9/16/94 6.15 0.97 2.79 3.30 6.15 9/27/94 9/28/94 3.92 0.54 2.27 2.48 3.52

PAGE 105

92 The soil storage c urves used in the WFHC analysis are represented by the regression line in Figures 22 to 24, depending on the soil type located in individual pixels. Figure 45 compares the predicted runoff to the observed runoff. In general, the SCS SFWMD method underpre dicted runoff at WFHC. Of course, there were indeterminate sources of error such as estimation of runoff from the recorded hydrograph and lack of detail in the spatial distribution of storms. Another troublesome source of error was the prediction of the water table elevation over the whole catchment, the critical factor in estimating soil storage capacity. The TOPMODEL approach generated water table depths that were too deep and consequently an average soil storage capacity that was too high. One explan ation for this could be that the approach assumes the baseflow is a result of gravity drainage alone. In a flatwoods environment, the baseflow is partially influenced by a small water table gradient and largely influenced by evapotranspiration (ET). The TOPMODEL approach was designed for steep and moderate slopes, where there is a strong water table gradient to achieve gravity flow. With the aid of a thickly vegetated riparian zone, ET in a shallow water table environment depresses the water table near t he creek, resulting in the land surface and water table no longer being parallel. In addition to not satisfying all TOPMODEL assumptions, the method of determining the water table distribution relied heavily on the original digital elevation model, which had coarse vertical and horizontal resolution. The 5 ft contours contributed to the error in the vertical direction while the 30 m x 30 m grid cell size was the source of error in the horizontal direction. Correct calculation of the topographic index dep ends on the accuracy of the DEM and its derived flow grid, upstream area grid, and slope grid. Coarse resolution in the DEM transmits error throughout all of these grids.

PAGE 106

93 y = 0.2735x + 1.4695 R 2 = 0.6654 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Observed Runoff (cm) Predicted Runoff (cm)) Figure 45. Runoff Prediction for West Fork Horse Creek For comparison purposes, the events from WFHC were tested with the original SCS method. The first three events, occurring in January, March, and April, were assigned as AMC I because very little rainfall fell in the preceding five days. Using a base CN of 72.4, the adjusted CN becomes 54 and the resulting runoff for all three events was calculated as nearly zero. This is in contrast to the large observed runoff values. The June, August, and latter September events are AMC II and therefore keep the base CN of 72.4. Runoff valu es were calculated as 1.17 cm, 0.63 cm, and 0.34 cm, respectively. Only the June event value was closer to the observed value than the simulated runoff from the proposed method. The first September event is AMC III with a CN of 89 and a runoff value of 3 .52 cm, which is close to 3.30 cm, the simulated value in Table 11. Again the original SCS method for AMC III performed similarly to the proposed method. Unlike the LFC results, however, the simulated values for the

PAGE 107

94 proposed and original SCS method did n ot match the observed runoff. Nevertheless, there is an advantage in using the proposed method because the results from the original method depend heavily on the selected curve number.

PAGE 108

95 CHAPTER 5. CONCLUSIONS 5 1 Factors Affecting Performance Every rainfa ll runoff model is limited by its ability to account for the variability of parameters such as saturated hydraulic conductivity and depth to the water table throughout the basin. Dunne (1983) recognized the difficulty in measuring the conductivity of poro us organic soils and in describing flow through thick vegetation that is characteristic of the frequently flooded zones associated with saturation overland flow. In contrast to variable source area theory, soil storage capacity was largest near the stream in the Long Flat Creek (LFC) subbasin. Large water oaks and heavy vegetation depressed the water table significantly within the riparian zone, presumably by root water uptake and transpiration. There was often less soil moisture storage in the uplands, which apparently formed runoff generating zones before the region around the creek itself. The west side of LFC was somewhat more affected by this phenomenon than the east side, probably because of thicker vegetation. This occurrence is in opposition to TOPMODEL theory, which suggests that saturated areas form near the creek first and then expand into the uplands. If the uplands saturate first and generate runoff, there is a chance that the runoff will infiltrate in the downstream unsaturated areas on it s way to the stream. Indeed, it was discovered that integration of some of the downstream soil moisture profiles at LFC revealed quantities of infiltrated water that were higher than the causative rainfall. This discrepancy suggests that the locations of these moisture probes were in

PAGE 109

96 areas allowing infiltration of some of the runoff originating from upstream. However, the chronological order of areas that became saturated did not change anything in the application of the method in this study. Identifica tion of saturated areas was only for quantification purposes and did not depend on timing. For future applications, the behavior of runoff zones in a flatwoods environment should be taken into consideration. It was expected that the runoff predicted by the SCS SFWMD method would contain significant error due to the lack of recorded detail in the spatial distribution of rainfall and catchment properties. The method contains a single parameter, S that lumps all of the catchment characteristics into one simplistic calculation. In this study another physically based approach, the TOPMODEL concept, was combined with the proposed method to demonstrate how other existing models could be used in conjunction with the method. Application of TOPMODEL in Florida flatwoods is difficult, and a simpler method may be preferred by practicing engineers. One such method could use a yearly average water table distribution, as provided by the local water management district or other agency. If one surficial well were av ailable in the catchment, the fluctuation of the water table could be mimicked throughout the basin by finding the deviation in the well reading from the yearly average for the required day and time, and then applying that difference to all of the other ye arly average depths. Even though the water table in a flatwoods environment is unlikely to rise and fall uniformly in the basin, assuming it does so offers a simplified approach. Numerous sources of uncertainty make it difficult to determine which approac h is best. Runoff determined from convective storms is likely to be less accurate than runoff determined from frontal storms because current rain gages are too sparsely situated in the

PAGE 110

97 catchments to record the spatial distribution of a convective event. Frontal storms are recorded with less error because they tend to have low, uniform intensities spread out over large areas. Complex storage relationships in Florida hydrology add additional uncertainty to the runoff estimates. Wetlands and depressions st ore unknown volumes of water and eventually bleed off to nearby streams. The storage available in wetlands, an initial condition difficult to determine, affects the amount of runoff seen at the outlet of the basin. To simplify the matter, it was assumed that the wetlands were full and stored only a small initial abstraction before producing overland flow. The exact amount produced in each event is impossible to quantify without field measurements. Because the proposed method assumes that all runoff is s aturation excess and not infiltration excess, the water table distribution is the most critical factor. Without an extensive network of shallow wells, it is extremely difficult to describe the behavior of the water table at all points in a catchment. In this study the TOPMODEL approach using the topographic index was attempted and was not very successful. Indeed, Sivapalan et al. (1987) stated that their equation for the water table distribution was meant for moderate and steep slopes. The flat topograp hy in Florida is not conducive to accurate water table predictions using topographic index relationships. In addition, digital elevation models are almost necessary in the analysis of large basins, and working with coarse horizontal and vertical resolutio n grids increases error, especially when parameters such as the topographic index are derived from the grids. Some of the error was reduced by calibration to the many surficial wells in the Long Flat Creek subbasin, which improved the application of TOPMO DEL. However, the SCS SFWMD method should not have to rely on a basin having extensive water table data.

PAGE 111

98 The soil storage capacity equation should be tested further by comparing results to observed data for various water table depth conditions and differe nt locations of a particular soil type. The equation predicted soil storage relatively well for the west side of the Long Flat Creek subbasin, but more testing needs to be conducted. The reliability of the equation should be verified with many soil moist ure data sets. Despite all of the difficulties, modifying the original SCS curve number equation for use with the saturation excess runoff mechanism should be possible. Rallison (1980) reviewed the fact that the parameter S is the maximum difference of ( P Q ) for a particular storm and set of watershed conditions, and it is limited by either the infiltration rate at the soil surface or the soil moisture storage capacity, whichever results in the smaller S value. Though many practitioners assume an infiltra tion excess condition, the proposed method uses the saturation excess mechanism, which is limited by the storage in the soil profile. However, Rallison questioned whether or not an entirely new method should be devised since even the original authors of t he SCS method did not realize how widely it would be used on hydrologic problems it was not intended to solve. Even if the SCS SFWMD method were considered a new method, it would be limited in its application to areas prone to saturation excess. It is un likely that there will ever be just one method that may be used at all times in all conditions, but modifying a single method to suit specific conditions may increase its applicability. One remaining factor that should be considered in the performance of any new method is the inclusion of existing urban areas. Though the catchments in this study were rural, urban areas should not be a problem because they are impervious just like the saturated areas in the proposed method. It is important to include an a nalysis of

PAGE 112

99 urbanization because Florida is developing at a fast pace, and the concepts developed in this study should be able to be applied to developing areas as well as to rural areas. Valeo and Moin (2001) incorporated much of this studys approach in the model TOPURBAN, which determines runoff generated by saturation excess overland flow in urbanizing basins. This model, which also includes a linear reservoir representation of urban storage, underpredicted runoff like the proposed model discussed in t his study. 5.2 Summary and Conclusion The objective of this research was to work toward a more physically based runoff estimation technique for shallow water table environments using the SCS method. The proposed method has an advantage over the current SCS method because it does not rely on the subjective selection of a curve number. Instead, an area averaged soil storage parameter is calculated from soil storage values for common flatwoods soils for the required water table depth and then inserted int o the SCS equation. The soil storage curves account for air encapsulation and could be compiled and distributed for common use. It is important to estimate air encapsulation to better characterize the appropriate soil storage. Runoff volumes are increas ed as a result of the reduced storage in the soil. The key is determining the correct water table depth at locations in a catchment, perhaps by using a technique like TOPMODEL or even groundwater model results. In this study, runoff was often underpredic ted when the SCS SFWMD method was used in conjunction with TOPMODEL. Even though the approach is physically based, the predicted water table depths were too deep and the resulting soil storage estimates too large for the West Fork Horse Creek basin. Over estimation of soil storage resulted in low runoff rates. The

PAGE 113

100 presence of numerous surficial wells in the Long Flat Creek subbasin allowed for better calibration of the average water table depth, which produced a good simulated runoff result from the propo sed method. Pressurization effects and certain installation techniques of the wells may have caused recorded water levels to be different than the actual water table in the soil matrix. Air encapsulation and air compression in the soil profile slow the p rocess of infiltration and generate doubt as to whether the soil parcel is subject to infiltration or saturation excess, or both. In addition, the Long Flat Creek subbasin, which had heavy vegetation lining the stream channel, did not follow the classical variable source area concept. Areas further upland seemed to be saturated before those near the creek because of a shallower water table and less soil storage. The outcome of the proposed method was not affected by this occurrence because only the total saturated area was required, not the location or direction of expansion of individual saturated areas. Further study is required to develop a technique to describe the water table distribution in environments with mild slopes. The TOPMODEL approach requi red only elevation and common soils data, but using a grid did not allow for a smooth continuous water table. Rather, neighboring grid cells had different water table depths, resulting in a disjointed pattern. However, because the SCS and SCS SFWMD method s are lumped parameter models, the average condition in the catchment is the goal, not the description of minute details. There was probably less gravity drainage to the study creeks through baseflow than assumed by TOPMODEL, because the slopes are not st eep enough to direct groundwater quickly to the stream. Instead, water is retained longer and closer to the surface in the uplands. Additionally, streamside trees are heavy water users and often

PAGE 114

101 act as a sink near the creek channel. Modification of the TOPMODEL concept for these flatwoods conditions may help to better predict baseflow and water table depth throughout a catchment. The SCS SFWMD method showed some promise in runoff prediction. The saturation excess mechanism was observed at the runoff tes t bed, and an event in the larger Long Flat Creek subbasin was closely simulated. More data is needed to further test the method. If accurate plots of the soil storage water table depth relationship for different Florida soils were made available, then s aturated and unsaturated areas could be predicted with only a water table distribution. Though the method is limited to shallow water table environments subject to saturation excess runoff, it has potential for widespread use because of the need to expand the original SCS method beyond infiltration excess applications.

PAGE 115

102 REFERENCES Andrews, R.G. (1954). The Use of Relative Infiltration Indices in Computing Runoff. Cited in Rainfall Runoff Relationship V.P. Singh, ed., Water Resources Publication s, Littleton, Colorado. Bedient, P.P., Huber, W.C., and Heaney, J.P. (1976). Modeling Hydrologic Land Use Interactions in Florida. Proceedings of the Conference on Environmental Modeling and Simulation Environmental Protection Agency Paper No. 600/9 76 016, Washington, D.C. Beven, K.J. and Kirkby, M.J. (1979). A Physically Based, Variable Contributing Area Model of Basin Hydrology. Hydrological Sciences Bulletin 24(1): 43 69. Beven, K.J. (1997). TOPMODEL: A Critique. Hydrological Process es 11(9): 1069 1085. Bhaskar, N.R., James, W.P., and Devulapalli, R.S. (1992). Hydrologic Parameter Estimation Using Geographic Information System. Journal of Water Resources Planning and Management ASCE, 118(5): 492 512. Bosznay, M. (1989). Generalization of SCS Curve Number Method. Journal of Irrigation and Drainage Engineering ASCE, 115(1): 139 144. Boughton, W.C. (1987). Evaluating Partial Areas of Watershed Runoff. Journal of Irrigation and Drainage Engineering ASCE, 113(3): 3 56 366. Buss, P. (1993). The Use of Capacitance Based Measurements of Real Time Soil Water Profile Dynamics for Irrigation Scheduling. Proceedings of the National Conference of the Irrigation Association Irrigation Association of Australia, Homebush NSW, Australia Capece, J.C. (1984). Estimating Runoff Peak Rates and Volumes from Flat, High Water Table Watersheds. Masters Degree Thesis University of Florida, Gainesville, Florida. Capece, J.C., Campbell, K.L., Baldwin, L.B., and Konya, K.D (1987). Estimating Runoff Volumes from Flat, High Water Table Watersheds. Transactions of the American Society of Agricultural Engineers 30(5): 1397 1402.

PAGE 116

103 Carlisle, V.W., Sodek, F., Collins, M.E., Hammond, L.C., and Harris, W.G. (1989). Characte rization Data for Selected Florida Soils. Institute of Food and Agricultural Sciences, University of Florida, Gainesville, Florida. Charbeneau, R.J. (2000). Groundwater Hydraulics and Pollutant Transport Upper Saddle River, New Jersey: Prentice Hall Inc. Constantz, J., Herkelrath, W.N., and Murphy, F. (1988). Air Encapsulation during Infiltration. Soil Science Society of America Journal 52(1): 10 16. Dunne, T., Moore, T.R., and Taylor, C.H. (1975). Recognition and Prediction of Runoff Pro ducing Zones in Humid Regions. Hydrological Sciences Bulletin 20(3): 305 327. Dunne, T. (1983). Relation of Field Studies and Modeling in the Prediction of Storm Runoff. Journal of Hydrology 65(1 3): 25 48. Fares, A., and Alva, A.K. (2000). Eva luation of Capacitance Probes for Optimal Irrigation of Citrus through Soil Moisture Monitoring in an Entisol Profile. Irrigation Science 19: 57 64. Fayer, M.J., and Hillel, D. (1986). Air Encapsulation: I. Measurement in a Field Soil. Soil Scienc e Society of America Journal 50: 568 572. Freeze, A.F., and Cherry, J.A. (1979). Groundwater Englewood Cliffs, New Jersey: Prentice Hall, Inc. Gillham, R.W. (1984). The Capillary Fringe and its Effect on Water Table Response. Journal of Hydrology 67(1 4): 307 324. Golding, B.L. (1997). Discussion of Runoff Curve Number: Has it Reached Maturity? by V.M. Ponce and R.H. Hawkins. Journal of Hydrologic Engineering ASCE, 2(3): 145 148. Hawkins, R.H. (1980). Infiltration and Curve Numbers: So me Pragmatic and Theoretic Relationships. Proceedings of the Symposium on Watershed Management ASCE, 925 937. Hawkins, R.H. (2001). Discussion of Another Look at SCS CN Method, by S.K. Mishra and V.P. Singh. Journal of Hydrologic Engineering ASC E, 6(5): 451 452.

PAGE 117

104 Hawkins, R.H., Jiang, R., Woodward, D.E., Hjelmfelt, A.T., and Van Mullem, J.A. (2002). Runoff Curve Number Method: Examination of the Initial Abstraction Ratio. Proceedings of the Second Federal Interagency Hydrologic Modeling C onference, Las Vegas, Nevada Heatwole, C.D. (1986). Field and Basin Scale Water Quality Models for Evaluating Agricultural Nonpoint Pollution Abatement Programs in a South Florida Flatwoods Watershed. Ph.D. Dissertation University of Florida, Gaine sville, Florida. Hernandez, T. (2000). Rainfall Runoff Modeling in Humid, Shallow Water Table Environments. Masters Degree Thesis University of South Florida, Tampa, Florida. Jarrett, A.R., Hoover, J.R., and Paulson, C.D. (1980). Subsurface Draina ge, Air Entrapment, and Infiltration in Sand. Transactions of the American Society of Agricultural Engineers 23(6): 1424 1428. Jordan, J.P. (1994). Spatial and Temporal Variability of Stormflow Generation Processes on a Swiss Catchment. Journal of Hydrology 153: 357 382. Lamb, R., Beven, K., and Myrab, S. (1997). Discharge and Water Table Predictions Using a Generalized TOPMODEL Formulation. Hydrological Processes 11(9): 1145 1167. Lewelling, B.R. (1997). Hydrologic and Water Quality Conditions in the Horse Creek Basin, West Central Florida. U.S. Geological Survey Water Resources Investigation Report 97 4077 Tallahassee, Florida. Loague, K., and Abrams, R.H. (2001). Stochastic Conceptual Analysis of Near Surface Hydrological Res ponse. Hydrological Processes 15(14): 2715 2728. McCuen, R.H. (1998). Hydrologic Analysis and Design 2 nd ed., Prentice Hall, Upper Saddle River, New Jersey. Mishra, S.K., and Singh, V.P. (1999). Another Look at SCS CN Method. Journal of Hydrolog ic Engineering ASCE, 4(3): 257 264. Mockus, V. (1949). Estimation of Total (and Peak Rates of) Surface Runoff for Individual Storms. Interim Survey Report Grand (Neosho) River Watershed, Exhibit A of Appendix B, USDA, Lincoln, Nebraska. Moore, R.D ., and Thompson, J.C. (1996). Are Water Table Variations in a Shallow Forest Soil Consistent with the TOPMODEL Concept? Water Resources Research 32(3): 663 669.

PAGE 118

105 Morel Seytoux, H. J. and Billica, J. A. (1985). A Two Phase Numerical Model for Predict ion of Infiltration: Case of an Impervious Bottom. Water Resources Research 21(9): 1389 1396. Morgan, K.T., Parsons, L.R., Wheaton, T.A., Pitts, D.J., and Obreza, T. A. (1999). Field Calibration of a Capacitance Water Content Probe in Fine Sand Soil s. Soil Science Society of America Journal 63: 987 989. Myers, R.D. (1999). Hydraulic Properties of South Florida Wetland Peats. Masters Degree Thesis University of Florida, Gainesville, Florida. Pearce, A.J., Stewart, M.K., and Sklash, M.G. (1 986). Storm Runoff Generation in Humid Headwater Catchments: 1. Where Does the Water Come From? Water Resources Research 22(8): 1263 1272. Perrone, J. and Madramootoo, C.A. (1998). Improved Curve Number for Runoff Prediction. Canadian Journal of Civil Engineering 25:728 734. Perry, R.G. (1995). Regional Assessment of Land Use Nitrogen Loading of Unconfined Aquifers. Ph.D. Dissertation University of South Florida, Tampa, Florida. Ponce, V.M., and Hawkins, R.H. (1996). Runoff Curve Numbe r: Has It Reached Maturity? Journal of Hydrologic Engineering ASCE, 1(1): 11 18. Rallison, R.E. (1980). Origin and Evolution of the SCS Runoff Equation. Proceedings of the Symposium on Watershed Management ASCE, 912 924. Robbins, J.M., Jr., Ford, R.D., Werner, J.T., and Cowherd, W.D. (1984). Soil Survey of Hardee County, Florida. Soil Conservation Service, USDA, Washington D.C. Seibert, J., Bishop, K.H., and Nyberg, L. (1997). A Test of TOPMODELs Ability to Predict Spatially Distributed G roundwater Levels. Hydrological Processes 11: 1131 1144. Seymour, R.M. (2000). Air Entrapment and Consolidation Occurring with Saturated Hydraulic Conductivity Changes with Intermittent Wetting. Irrigation Science, 20: 9 14. Sivapalan, M., Bev en, K., and Wood, E.F. (1987). On Hydrologic Similarity: 2. A Scaled Model of Storm Runoff Production. Water Resources Research 23(12): 2266 2278.

PAGE 119

106 Sivapalan, M., and Wood, E.F. (1990). On Hydrologic Similarity: 3. A Dimensionless Flood Freque ncy Model Using a Generalized Geomorphologic Unit Hydrograph and Partial Area Runoff Generation. Water Resources Research 26(1): 43 58. Sivapalan, M., Woods, R.A., and Kalma, J.D. (1997). Variable Bucket Representation of TOPMODEL and Investigation of the Effects of Rainfall Heterogeneity. Hydrological Processes 11(9): 1307 1330. SFWMD. (2002). Environmental Resource Permit Information Manual, Volume IV South Florida Water Management District, West Palm Beach, Florida. Smith, R.E. (1997). D iscussion of Runoff Curve Number: Has it Reached Maturity? by V.M. Ponce and R.H. Hawkins. Journal of Hydrologic Engineering ASCE, 2(3): 145 148. Spier, W.H., Mills, W.C., and Stephens, J.C. (1969). Hydrology of Three Experimental Watersheds in Southern Florida: A Progress Report. ARS Publication No. 41 152 USDA Agricultural Research Service, Washington, D.C. Springer, E.P., McGurk, B.J., Hawkins, R.H., and Coltharp, G.B. (1980). Curve Numbers from Watershed Data. Proceedings of the Symp osium on Watershed Management ASCE, 938 950. Steenhuis, T.S., Winchell, M., Rossing, J., Zollweg, J.A., and Walter, M.F. (1995). SCS Runoff Equation Revisited for Variable Source Runoff Areas. Journal of Irrigation and Drainage Engineering ASCE, 121 (3): 234 238. Taboada, M.A., Lavado, R.S., Rubio, G., and Cosentino, D.J. (2001). Soil Volumetric Changes in Natric Soils Caused by Air Entrapment Following Seasonal Ponding and Water Table Rises. Geoderma 101: 49 64. Trommer, J.T., Loper, J.E., and Hammett, K.M. (1996). Evaluation and Modification of Five Techniques for Estimating Stormwater Runoff for Watersheds in West Central Florida. U.S. Geological Survey Water Resources Investigation Report 96 4158 Tallahassee, Florida. USDA. (1985) SCS National Engineering Handbook. Section 4: Hydrology, Chapter 4 Soil Conservation Service, U.S. Department of Agriculture, Washington D.C. Valeo, C., and Moin, S.M.A. (2001). Hortonian and Variable Source Area Modeling in Urbanizing Basins. Journ al of Hydrologic Engineering ASCE, 6(4): 328 335.

PAGE 120

107 Wang, Z., Feyen, J., Nielsen, D.R., and van Genuchten, M.T. (1997). Two Phase Flow Infiltration Equations Accounting for Air Entrapment Effects. Water Resources Research 33(12): 2759 2767. Wang, Z ., Feyen, J., van Genuchten, M.T., and Nielsen, D.R. (1998). Air Entrapment Effects on Infiltration Rate and Flow Instability. Water Resources Research 34(2): 213 222. Wangemann, S.G., Kohl, R.A., and Molumeli, P.A. (2000). Infiltration and Percolat ion Influenced by Antecedent Soil Water Content and Air Entrapment. Transactions of the American Society of Agricultural Engineers 43(6): 1517 1523. White, D. (1988). Grid Based Application of Runoff Curve Numbers. Journal of Water Resources Plann ing and Management ASCE, 114(6): 601 612. Willeke, G.E. (1997). Discussion of Runoff Curve Number: Has it Reached Maturity? by V.M. Ponce and R.H. Hawkins. Journal of Hydrologic Engineering ASCE, 2(3): 145 148. Willgoose, G. and Perera, H. (2 001). A Simple Model of Saturation Excess Generation Based on Geomorphology, Steady State Soil Moisture. Water Resources Research 37(1): 147 155. Wilson, B.N., Slack, D.C. and Young, R.A. (1982). A Comparison of Three Infiltration Models. Transact ions of the American Society of Agricultural Engineers 25(2): 349 356. Yu, Bofu. (1998). Theoretical Justification of SCS Method for Runoff Estimation. Journal of Irrigation and Drainage Engineering ASCE, 124(6): 306 309. Yu, Bofu. (2001). Discu ssion of Another Look at SCS CN Method, by S.K. Mishra and V.P. Singh. Journal of Hydrologic Engineering ASCE, 6(5): 451 452.

PAGE 121

108 APPENDICIES

PAGE 122

109 Appendix A. Soils Data The data for Myakka fine sand, Immokalee fine sand, and Smyrna s and as given by the IFAS (Carlisle et al. 1989) are presented below. Tables 12 to14 provide physical properties of the soils while Tables 15 to17 present water retention data. The water retention data were used in the manual calculation of soil moisture storage values. Table 12. Soil Properties of Myakka Fine Sand Horizon Depth (cm) Saturated Hydraulic Conductivity (cm/hr) Bulk Density (g/cm 3 ) 1 0 18 38.8 1.44 2 18 64 28.0 1.53 3 64 76 12.8 1.38 4 76 91 9.0 1.52 5 91 150 11.2 1.58 6 150 203 9.5 1.60 Table 13. Soil Properties of Immokalee Fine Sand Horizon Depth (cm) Saturated Hydraulic Conductivity (cm/hr) Bulk Density (g/cm 3 ) 1 0 13 15.8 1.18 2 13 109 20.4 1.44 3 109 119 0.2 1.37 4 119 140 5.0 1.36 5 140 165 1.9 1.4 9

PAGE 123

110 Appendix A (Continued) Table 14. Soil Properties of Smyrna Sand Horizon Depth (cm) Saturated Hydraulic Conductivity (cm/hr) Bulk Density (g/cm 3 ) 1 0 13 18.4 1.23 2 13 38 14.8 1.51 3 38 46 11.2 1.44 4 46 56 34.2 1.45 5 56 89 18.4 1.61 6 89 114 10.6 1.71 7 114 142 7.6 1.68 8 142 203 1.3 1.81 Table 15. Water Content for Selected Capillary Pressures for Myakka Fine Sand Horizon 3.5 cm 20 cm 30 cm 45 cm 60 cm 80 cm 150 cm 200 cm 1 0.380 0.322 0.283 0.214 0.169 0.143 0.110 0.099 2 0.370 0.342 0.274 0.160 0.107 0.077 0.048 0.045 3 0.428 0.376 0.359 0.326 0.300 0.276 0.242 0.231 4 0.361 0.333 0.316 0.254 0.210 0.183 0.151 0.144 5 0.376 0.358 0.303 0.231 0.183 0.149 0.115 0.107 6 0.347 0.332 0.315 0.226 0.180 0.143 0.106 0.098

PAGE 124

1 11 Appendix A (Continued) Table 16. Water Content for Selected Capillary Pressures for Immokalee Fine Sand Horizon 3.5 cm 20 cm 30 cm 45 cm 60 cm 80 cm 150 cm 200 cm 1 0.506 0.479 0.474 0.360 0.246 0.188 0.134 0.121 2 0.378 0.353 0.352 0.242 0.156 0.08 6 0.051 0.049 3 0.399 0.377 0.377 0.375 0.372 0.368 0.351 0.340 4 0.432 0.393 0.367 0.328 0.297 0.271 0.240 0.234 5 0.381 0.376 0.354 0.313 0.281 0.250 0.210 0.200 Table 17. Water Content for Selected Capillary Pressures for Smyrna Sand Horizon 3.5 c m 20 cm 30 cm 45 cm 60 cm 80 cm 150 cm 200 cm 1 0.480 0.473 0.471 0.444 0.406 0.346 0.281 0.259 2 0.337 0.320 0.317 0.292 0.177 0.122 0.059 0.048 3 0.413 0.401 0.388 0.366 0.345 0.322 0.286 0.274 4 0.384 0.362 0.339 0.300 0.263 0.223 0.164 0.153 5 0.3 45 0.341 0.333 0.285 0.244 0.184 0.094 0.082 6 0.343 0.334 0.333 0.324 0.257 0.160 0.073 0.060 7 0.332 0.319 0.317 0.305 0.281 0.248 0.177 0.158 8 0.314 0.303 0.302 0.298 0.241 0.158 0.104 0.092 The parameters for the soil moisture capacity equation ( 7) as calculated from the regression constants A B C and D and as determined by fitting data to the Brooks and Corey model are in Table 18. The latter values represent the average over all horizons of the soil. The constants A B C and D for units o f inches are presented in Table 19. They

PAGE 125

112 Appendix A (Continued) were found by optimizing the storage equation with the soil moisture storages as calculated from water retention data in the tables above. Table 18. Brooks and Corey Model Parameters From Regression Constants From Water Retention Data Soil N ame l h A (cm) q s q g l h A (cm) q s q g Myakka Fine Sand 1.111 18.09 0.292 1.144 19.37 0.341 Immokalee Fine Sand 0.878 23.89 0.458 0.605 15.66 0.427 Smyrna Sand 0.922 28.27 0.254 0.888 28.69 0.377 Table 19. Soil Moisture Capacity Equation Regression Const ants for Units in Inches Soil A B C D Myakka Fine Sand 23.75 0.1107 0.2924 20.91 Immokalee Fine Sand 26.72 0.1223 0.4583 30.93 Smyrna Sand 29.73 0.0780 0.2540 33.43 The rate constant, f from Eq. (11) was calculated for a number of soils using sat urated hydraulic conductivity data from the Hardee County soil survey (Table 20).

PAGE 126

113 Appendix A (Continued) Table 20. Saturated Hydraulic Conductivity Data (Robbins et al. 1984) Cassia fine sand Myakka fine sand Pomello fine sand Ks (cm/hr) z (cm) Ks (cm/ hr) z (cm) Ks (cm/hr) z (cm) 79.5 0 27.5 0 17.4 0 79.5 15 27.5 15 17.4 13 28.6 16 14.5 16 17 a 14 28.6 68 14.5 53 17 a 38 5.3 69 5.6 54 19.3 39 5.3 86 5.6 64 19.3 117 9.3 87 6.4 65 1.3 118 9.3 107 6.4 76 1.3 147 10.3 108 9.6 77 16.5 148 10.3 145 9. 6 102 16.5 168 2.9 146 0.2 103 3.3 169 2.9 165 0.2 117 3.3 203 1.4 166 3.5 118 1.4 203 3.5 137 4.1 138 4.1 203 Smyrna sand Ona fine sand Immokalee fine sand b Ks (cm/hr) z (cm) Ks (cm/hr) z (cm) Ks (cm/hr) z (cm) 14.8 0 0. 4 0 15.8 0 14.8 13 0.4 10 15.8 13 16.3 14 5.5 11 20.4 14 16.3 41 5.5 23 20.4 109 6.2 42 12 24 0.2 110 6.2 51 12 41 0.2 119 10.7 52 9.4 42 5 120 10.7 61 9.4 61 5 140 10.8 62 8.5 62 1.9 141 10.8 74 8.5 107 1.9 165 4.9 108 4.9 152 a Estimated value used because no data was provided. b Data from Carlisle et al. (1989) because not listed in soil survey.

PAGE 127

114 Appendix B. Long Flat Creek Data The initial soil moisture profile data presented in Table 21 were used in determining soil moisture storage for the rainfall event occurring on June 25, 2002. The data were collected from instrumentation installed by the University of South Florida. The names of the soil moisture probes are the same as those of the nearest well. Table 21. Soil Moistu re Probe Water Content Before Storm on June 25, 2002 (%) Depth (cm) USF 1 USF 3 PS 40 PS 41 PS 42 PS 43 10 27.0 32.7 14.3 12.6 20.4 22.2 20 24.9 34.1 11.0 16.8 16.8 18.4 30 27.9 10.3 13.7 15.2 17.6 50 14.6 18.5 21.7 27.8 70 22.8 35.0 90 29.0 Porosity values used for maximum saturated water content, as determined from soil sample analyses, are in Table 22. The porosity for an untested depth was assumed to be equal to the known porosity of the nearest range of depth or was li nearly interpolated if the depth was relatively far from the nearest known range of depth. Soil texture by depth is also provided in Table 22. Despite most soil layers containing fine to medium sand, different horizons were established based on color, or ganics, and other aspects in sample appearance.

PAGE 128

115 Appendix B (Continued) Table 22. Laboratory Soil Texture and Porosity Data for Long Flat Creek Soil Probe Range of Depth (cm) Soil Texture Porosity (cm 3 /cm 3 ) USF 1 61 91 Fine to Medium Sand 0.36 USF 1 91 122 Clay 0.41 USF 3 61 91 Fine to Medium Sand 0.37 USF 3 91 122 Clay 0.38 PS 40 0 15 Fine to Medium Sand, Silty Organics 0.29 PS 40 15 46 Fine to Medium Sand, Silty Organics 0.30 PS 40 274 305 Clay 0.36 PS 41 46 61 Fine to Medium San d 0.41 PS 41 107 122 Fine to Medium Sand 0.39 PS 42 30 61 Fine to Medium Sand 0.41 PS 42 91 107 Fine to Medium Sand 0.40 PS 42 107 122 Fine to Medium Sand 0.26 PS 43 30 61 Fine to Medium Sand 0.41 PS 43 122 152 Fine to Medium Sand, Trace Clay 0.36 Air encapsulation data for the runoff test bed are in Table 23. Additional data for the rest of the western part of the LFC subbasin are in Table 24. These data were used to find the actual storage available for stormwater to occupy in the s oil matrix. Table 25 provides the water table depths at LFC before June 25, 2002.

PAGE 129

116 Appendix B (Continued) Table 23. Air Encapsulation Data for the Runoff Test Bed Soil Sensor Depth (cm) Final Depth to Water (cm) Pressure Head Above Probe (cm) Final Wa ter Content (%) Porosity (%) Air Encapsulation (%) USF 1 4/13/2002 40 27.6 12.4 25.3 36.2 10.9 11:25 70 27.6 42.4 26.8 36.2 9.4 6/26/2002 20 0.0 20 33.9 36.2 2.3 17:45 30 0.0 30 32.0 36.2 4.2 50 0.0 50 27.1 36.2 9.1 70 0.0 70 26.0 36.2 10.2 6/20/2002 90 70.9 19.1 26.5 36.2 9.7 9:20 110 70.9 39.1 41.5 41.5 0.0 USF 3 4/14/2002 40 36.0 4.0 26.1 37.1 11.1 16:00 70 36.0 34.0 27.2 37.1 9.9 100 36.0 64.0 29.8 38.5 8.7 6/26/2002 20 12.2 7.8 34.2 37.1 2.9 17:45 30 12.2 17.8 28.3 37.1 8. 9 50 12.2 37.8 27.8 37.1 9.3 70 12.2 57.8 28.0 37.1 9.1 6/20/2002 90 86.6 3.4 27.2 37.1 9.9 9:20 110 86.6 23.4 31.8 38.5 6.7

PAGE 130

117 Appendix B (Continued) Table 24. Air Encapsulation Data for Long Flat Creek Soil Sensor Depth (cm) Final Depth to Water (cm) Pressure Head Above Probe (cm) Final Water Content (%) Porosity (%) Air Encapsulation (%) PS 43 6/26/2002 30 21.6 8.4 30.9 41.0 10.1 17:40 50 21.6 28.4 29.0 41.0 12.0 PS 42 6/26/2002 50 31.3 18.7 29.0 41.0 12.0 17:40 PS 41 6/26/2002 50 37.8 12.2 27.1 41.0 13.9 17:40 70 37.8 32.2 38.7 41.0 2.3 PS 40 6/26/2002 70 68.2 1.8 26.1 33.0 6.9 17:40 90 68.2 21.8 29.2 34.0 4.8 Table 25. Pre Storm Water Table Depths in the Long Flat Creek Subbasin Well 6/25/02 ( cm) USF 1 28.7 USF 3 42.7 PS 1 140.2 PS 2 56.7 PS 3 99.1 PS 4 81.7 PS 5 44.2 PS 40 95.9 PS 41 82.1 PS 42 70.9 PS 43 57.9

PAGE 131

118 Appendix C. Brooks and Corey Model of Soil Water Retention The Brooks and Corey model of soil water retention provides a re lationship between soil water content and soil water suction (matric potential). The normalized water content, S e is defined as (20) where: q equals volumetric water content q r equals residual water content f equals porosity h A equals air entry pressure h C equals capillary pressure, and l equals the pore size distribution index Equation (20) is applied only when h C > h A For h C = h A S e is e qual to one. Porosity was calculated from bulk density (BD) by the relationship [1 (BD / 2.65)] while the residual water content was considered equivalent to the wilting point or field capacity ( q g ), two values that are small in sand and are often publi shed with other water content data. Equation (20) may be rearranged into the form of the equation of a line by log S e = l log h C + l log h A (21)

PAGE 132

119 Appendix C (Continued) so that l is the slope and l log h A is the y intercept. Table 26 show s how the model was prepared for fitting the top two horizons of Immokalee fine sand. Refer to Tables 13 and 16 for additional data on this soil. Figure 46 shows a linear regression performed on the data. The model should be fit only to capillary pressu res larger than the bubbling pressure. When selecting values to use in the model, lower capillary pressures should be removed from the regression. The equations of the line for both horizons 1 and 2 were used to derive the parameters l and h A There are a total of 5 horizons in the soil, and the two parameters were found for each one and then averaged to obtain the values previously presented in Table 18. Table 26. Brooks and Corey Model for Immokalee Fine Sand Horizon 1 q h C (cm) S e log h C log S e 0.506 3.5 0.904894 0.544068 0.0434 0.479 20 0.852184 1.30103 0.06947 0.474 30 0.842423 1.477121 0.07447 0.360 45 0.619869 1.653213 0.2077 0.246 60 0.397316 1.778151 0.40086 0.188 80 0.284087 1.90309 0.54655 0.134 150 0.178667 2 .176091 0.74795 0.121 200 0.153288 2.30103 0.81449 Horizon 2 q h C (cm) S e log h C log S e 0.378 3.5 0.821665 0.544068 0.08531 0.353 20 0.764945 1.30103 0.11637 0.352 30 0.762676 1.477121 0.11766 0.242 45 0.513109 1.653213 0.28979 0.156 60 0.317993 1.778151 0.49758 0.086 80 0.159178 1.90309 0.79812 0.051 1 50 0.079771 2.176091 1.09816 0.049 200 0.075233 2.30103 1.12359

PAGE 133

120 Appendix C (Continued) y = -0.9294x + 1.2833 R 2 = 0.9797 y = -1.3219x + 1.8329 R 2 = 0.9684 -1.2 -0.8 -0.4 0 1.2 1.4 1.6 1.8 2 2.2 2.4 log h C log S e H1 H2 Linear (H1) Linear (H2) Figure 46. Brooks and Corey Model Fitting for Immokalee Fine Sand


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 Ka
controlfield tag 001 001413343
003 fts
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 030703s2002 flu sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000040
035
(OCoLC)51164763
9
AJJ0759
b SE
SFE0000040
040
FHM
c FHM
049
FHME
090
TA145
1 100
Masek, Caroline Humphrey.
0 245
Adapting the SCS method for estimating runoff in shallow water table environments
h [electronic resource] /
by Caroline Humphrey Masek .
260
[Tampa, Fla.]:
University of South Florida,
2002.
502
Thesis (M.S.)--University of South Florida, 2002.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 120 pages.
520
ABSTRACT: Rainfall-runoff modeling in the United States has made extensive use of the Soil Conservation Service (SCS) curve number method for computing infiltration losses from rainfall. Even though the method is well established and may be applied to a wide range of environments, it often results in highly erroneous runoff estimates for shallow water table environments. Flat topography, wetlands, and fine sands are characteristics that make places like Florida very different from the environments where the SCS method was originally developed. The SCS method arose from experiments with soils that are dominated by infiltration excess (Hortonian mechanism), where runoff occurs after rainfall intensity exceeds the infiltration capacity of the soil. In contrast, Florida is likely dominated by saturation excess runoff (Dunne mechanism), where the soil storage capacity between a shallow water table and the ground surface is filled, and all remaining rainfall becomes runoff. The sandy^soils of Florida have very high infiltration capacities, and thus infiltration excess is less likely than saturation excess. As a consequence of the saturation-excess mechanism, wetlands expand in the wet season as the soil moisture storage around the perimeter is filled.A modified form of the SCS method is proposed with the objective that it is more suitable than the current method in flatly sloped, humid environments. Initial conditions, such as the pre-storm soil moisture profile and depth to water table, are critical when predicting runoff in these areas. Air encapsulation is addressed because its presence causes the soil storage capacity to be filled significantly faster than in its absence. Equations are presented that provide an estimate of the average depth to water table and average soil storage capacity in a catchment.Two Florida catchments and one runoff test bed were selected for testing the new methodology. The runoff test bed demonstrated the saturation-excess mechanism while the catchments provided larger-scale testing of the method. Though more data is needed to fully assess the performance of the method, the approach offers a more plausible mechanism for runoff estimation in shallow water table environments with sandy soils.
590
Adviser: Nachabe, Mahmood
650
Runoff
x Measurement.
Water table.
690
Dissertations, Academic
z USF
Civil Engineering
Masters.
653
air encapsulation.
variable source area.
soil storage.
saturation excess.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.40