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 001914460
007 cr mnu|||uuuuu
008 071024s2007 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001982
Modeling considerations for vadose zone soil moisture dynamics
h [electronic resource] /
by Jing Zhang.
[Tampa, Fla.] :
b University of South Florida,
ABSTRACT: Reproducing moisture retention behavior of the upper and lower vadose zone in shallow water table settings provides unique challenges for integrated (combined surface and groundwater) hydrological models. Field studies indicate that moisture retention in shallow water table settings is highly variably affected by antecedent state and air entrapment. The theory and vertical behavior of a recently developed integrated surface and groundwater model (IHM) is examined through comparisons to collected field data in West-Central Florida. The objectives of this study were to (1) Identify important considerations and behavior of the vadose zone for reproducing runoff, ET and recharge in shallow water table settings; (2) Develop a conceptual model that describes vertical soil moisture behavior while allowing for field scale variability; (3) Test the model against observations of the vertical processes; (4) Investigate the sensitivity of model parameters on model vs. observed vertical behavior, and (5) offer recommendations for improvements and parameterization for regional model application. Rigorous testing was made to better understand the robustness and/or limitations of the methodology of the IHM for upper and lower vadose zone. The results are also generally applicable and useful to the upper zone and lower zone conceptualization and parameterization of stand alone HSPF and perhaps other surface water models. Simulation results indicate IHM is capable of providing reasonable predictions of infiltration, depth to water table response, ET distributions from the upper soil, lower soil and water table, and recharge while incorporating field scale variability of soil and land cover properties.
Dissertation (Ph.D.)--University of South Florida, 2007.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 116 pages.
Advisor: Mark A. Ross, Ph.D.
Shallow water table.
Surface water and groundwater interaction.
x Civil Engineering
t USF Electronic Theses and Dissertations.
Modeling Considerations for Vadose Zone Soil Moisture Dynamics by Jing Zhang A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Civil and Environmental Engineering College of Engineering University of South Florida Major Professor: Mark A. Ross, Ph.D. Mahmood H. Nachabe, Ph.D. Ahmed Said, Ph.D. Mark T. Stewart, Ph.D. Jeffrey S. Geurink, Ph.D. Date of Approval: March 8, 2007 Keywords: Upper zone, Lower zone, Evapotrans piration, Shallow water table, Integrated model, Model calibration, Surface wa ter and groundwater interaction Copyright 2007, Jing Zhang
ACKNOWLEDGMENTS This is the moment to say something from deep in my heart. This endeavor would not be possible without the help of many indi viduals that have assisted me along this journey. Especially, I would like to thank my advisor Dr. Mark Ross for his continued support, encourage and shared insights in my professional growth. I would also like to sincerely thank all of my committee member s for their contributions. Dr. Geurink was especially helpful with the code and mode l. Also, without your guidance, calibration would have been a much tougher task for me. I would also like to extend many thanks to all my friends at the Center for Modeling Hydrologic and Aquatic Systems (CM HAS), professors and staffs in CEE, their suggestions in this res earch, their help in my life he re, and their care for me have always been a cheris hed source of energy. My parents, sister, my feixiong, all my family, without your love, everything is impossible for me.
i TABLE OF CONTENTS LIST OF TABLES iii LIST OF FIGURES iv ABSTRACT vi CHAPTER 1 INTRODUCTION 1 1.1 Background 1 1.2 Objective and Scope 5 CHAPTER 2 VADOSE ZONE CHARACTERI ZATION IN SELECTED INTEGRATED SURFACE AND GROUNDWATER MODELS 7 2.1 SHE and MIKE SHE 7 2.2 tRIBS 9 2.3 MODHMS 10 2.4 SWATMOD 12 2.5 FHM 13 2.6 IHM 14 CHAPTER 3 OVERVIEW OF VERTICAL BEHAVIOR OF THE VADOSE ZONE IN IHM 18 3.1 IHM 3-Layer Soil Moisture Model 18 3.2 Interception, Depression and Surface Detention Storage 21 3.3 Vadose Zone Storage 24 3.4 Infiltration 27 3.5 Recharge 32 3.6 Evapotranspiration 34 CHAPTER 4 METHODOLOGY FOR SIMULATING VADOSE ZONE 38 4.1 Soil Zonation 38 4.1.1 2-Layer Soil Discretization in IHM 39 4.1.2 Upper Zone as the A Horizon 41 4.1.3 Observations from Field Studies 43 4.1.4 Formulation of Relative Moisture 47 4.2 Air Entrapment/pressurization 53 4.2.1 Background 53
ii 4.2.2 Theoretical and Model Tes ting of Excess Pressure 55 188.8.131.52 Numerical Model 56 184.108.40.206 Calculation of Excess Pr essurization using Ideal Gas Law 58 4.3 IHM Testing 62 4.3.1 Site Description 64 4.3.2 Model Setup 64 4.3.3 Data Collection 65 220.127.116.11 Basin Landuse 66 18.104.22.168 Soil Moisture 67 22.214.171.124 Water Table 67 126.96.36.199 Rainfall 67 188.8.131.52 Streamflow 68 184.108.40.206 Potential Evapotranspiration 68 4.3.4 Model Calibration 69 CHAPTER 5 RESULTS AND DISCUSSION 72 5.1 Soil Zonation 72 5.1.1 Moisture Conditions for the High ET Period 72 5.1.2 Moisture Conditions for the Low ET Period 73 5.1.3 Statistical Summary and Discussion 74 5.2 Air Entrapment 79 5.3 IHM Testing 81 5.3.1 Sensitivity Analysis 81 5.3.2 Results and Statistics Analysis 82 5.3.3 Discussion 93 220.127.116.11 Calibration Parameters 95 18.104.22.168.1 UZSN and UZET 95 22.214.171.124.2 Capillary Fringe/Root Zone and GWET 95 126.96.36.199.3 INFILT and Distribution of Available Moisture 97 188.8.131.52.4 Plant Coefficient and LZET 98 184.108.40.206 Air Entrapment 100 CHAPTER 6 CONCLUSIONS AND RECOMMENDATIONS 103 REFERENCES 110 ABOUT THE AUTHOR End Page
iii LIST OF TABLE Table 2.1. Similarities and Diffe rences Between FHM and IHM ET Conceptualization 17 Table 4.1. Derived Calibration Parameters for Forested and Grassed Land Cover 70 Table 5.1. Statistical Results for Relative Moisture Condition, UZ and LZ and Total Soil Moisture, UZ and LZ for both Upper and Lower Unsaturated Zone by Landuse Gr oup in Selected a) High ET (4/1 /2003-8/31/2003) and b) Low ET (11/1/2003-12/31/2003) Periods 77 Table 5.2. Statistical Results for Relative Moisture Condition, UZ and LZ and Total Soil Moisture, UZ and LZ in Upper and Lower Unsaturated Zone by Landuse Groups for All Data Periods (1/1/2002-6/27/2004), (a) Forested Cover; (b) Grassed Cover 78 Table 5.3. Model Sensitivity Analysis fr om Calibration Parameters for Grassed Land Cover 83 Table 5.4. Model Sensitivity Analysis from Calibration Parameters for Forested Land Cover 84 Table 5.5. Model Daily Performance Statisti cs for (a) Grassed and (b) Forested Land Cover 85
iv LIST OF FIGURES Figure 3.1. Three-Layer Soil Moisture Model 19 Figure 3.2. Storages Pertaining to th e Vadose Zone Described in IHM 21 Figure 3.3. HSPF Linear Probability Density Function 29 Figure 3.4. Distribution of Lowe r Zone Infiltration and Percolation 34 Figure 3.5. Vertical Moisture Fluxes and Storage in IHM 35 Figure 4.1. Typical Soil Profiles in Sedi mentary Soils and Graphical Depiction of Upper and Lower Soil Moisture Zones 42 Figure 4.2. Equilibrium Moisture Retent ion Characteristics of the A Horizons in Florida Fine Sandy Soils 43 Figure 4.3. Location Map of Observation Wells and Soil Moisture Monitoring Sites 44 Figure 4.4. Representative FieldScale Variability Shown by Concurrent Vertical Soil Moisture Observations from Six Stations in (a) Deeper Water Table Period; (b) Shallow Water Table 46 Figure 4.5. Examples of Fitted Mean and Minimum Curves to Daily Observations of Total Soil Moisture vs. Depth to Water Table in (a) Upper Zone for PS42; (b) Lower Zone for PS41 50 Figure 4.6. Sub-Basins, Landsegments and Observ ation Wells at the Long Flat Creek Study Site 66 Figure 4.7. Calibration Values Used for (a) Plant Coefficient and b) Inte rception Storage for Grassed and Forested Land Cover 70 Figure 5.1. Upper and Lower Zone Total Soil Moisture vs. Relative Moisture for Representative High ET Period (Apr. Aug. 2003) for Forested (a-e) and Grassed (f-j) Cover 75
v Figure 5.2. Upper and Lower Zone Total Soil Moisture vs. Relative Moisture for Representative Low ET Period (Nov. Dec. 2003) for Forested (a-e) and Grassed (f-j) Cover 76 Figure 5.3. Excess Pressure as Calculated from a Spreadsheet Solution of the Ideal Gas Law and a HYDRUS Solution 80 Figure 5.4. Calibration Results for Cu mulative ET Fluxes for (a) Grassed and (b) Forested Land Cover 86 Figure 5.5. Calibration Results for Daily Total ET Flux for (a) & (c) Grassed and (b) & (d) Forested Land Cover 87 Figure 5.6. Calibration Results fo r Daily Lower Zone ET Flux for (a) & (c) Grassed and (b) & (d) Forested Land Cover 88 Figure 5.7. Calibration Results fo r Daily Upper Zone ET Flux for (a) & (c) Grassed and (b) & (d) Forested Land Cover 89 Figure 5.8. Calibration Results for Daily Infiltration for (a) & (c) Grassed and (b) & (d) Forested Land Cover 90 Figure 5.9. Calibration Results for Depth to Water Table for (a) Grassed and (b) Forested Land Cover 91 Figure 5.10. Time Scale Analysis on Daily TAET for Forested Land Cover (a) 3-Day Average (b) 5-Day Average 94 Figure 5.11. Sensitivity to Plant Coefficient Seasonal Variability for LZET (F1 with PC1, F2 with PC2, F3 with PC3) 99 Figure 5.12. Air Entrapment Periods for Grassed Section (a) 5/4/02~7/1/02 and (b) 4/ 1/03~6/1/03 and Corresponding HYD RUS Solution (c) and (d) 101 Figure 5.13. IHM Model Cumulative Infilt ration Compared to Observations During the Periods with or withou t Air Entrapment 102
vi MODELING CONSIDERATIONS FOR VADOSE ZONE SOIL MOISTURE DYNAMICS Jing Zhang ABSTRACT Reproducing moisture retention behavior of the upper and lower vadose zone in shallow water table settings provides unique challenges for integrated (combined surface and groundwater) hydrological models. Field studi es indicate that moisture retention in shallow water table settings is highly vari ably affected by antecedent state and air entrapment. The theory and vertical behavior of a recently developed integrated surface and groundwater model (IHM) is examined thr ough comparisons to collected field data in West-Central Florida. The objectives of th is study were to (1) Identify important considerations and behavior of the vadose z one for reproducing runoff, ET and recharge in shallow water table settings; (2) Develop a conceptual model that describes vertical soil moisture behavior while allowing for fi eld scale variability; (3) Test the model against observations of the vertical processe s; (4) Investigate the sensitivity of model parameters on model vs. observed vertical behavior, and (5) offer recommendations for improvements and parameterization for regional model application. Ri gorous testing was made to better understand the robustness and/or limitations of the methodology of the IHM for upper and lower vadose zone. The resu lts are also generally applicable and
vii useful to the upper zone and lower zone con ceptualization and para meterization of stand alone HSPF and perhaps other surface water models. Simulation results indicate IHM is capable of providing reasonable predictions of infiltration, de pth to water table response, ET distributions from the upper soil, lower soil and water table, and recharge while incorporating field scale variability of soil and land cover properties.
1 CHAPTER 1 INTRODUCTION 1.1 Background Recent investigations of field data ha ve shown that the hydrologic behavior of runoff, ET and groundwater rech arge is controlled by vadose zone moisture which is strongly non-uniform (Rahgozar et al ., 2005). For example, observations from field data indicate that 50-70% of the to tal ET comes from a clearly identifiable distinct soil zone very near land surface (Rahgozar et al ., 2005). ET is an important element of the hydrologic cycle and is the dominant component of the annual rainfall of most regions, as high as 70 or 80% in Florida, (Bidlake, et al ., 1993). Unfortunately, ET can be the most difficult hydrologic process to analyze. Ther efore, there is a strong need in hydrologic models seeking to predict cont inuous runoff and recharge beha vior in shallow water table to simulate ET processes correctly (Ross et al ., 2005a). As ET is dependent on both surface and groundwater condition, this has give n rise to the popularity of integrated surface and groundwater models. In the last couple of decades, severa l integrated surface and groundwater models have been developed. Examples include MIKE-SHE (Refsgaard & Storm, 1995; DHI, 1998), MOD-HMS (HydroGeoLogic Inc., 2003; Panday and Huyakorn, 2004), SWATMOD (Sophocleous et al ., 1999), tRIBS (Vivoni et al ., 2003), FHM (Ross et al .,
2 1997) and IHM (Ross et al ., 2004). These models provide ve ry different approaches in characterizing and describing the vertic al behavior of the vadose zone. One such model, the Integrated Hydrologic Model (IHM) integrates the significant surface and subsurface hydrologic pr ocesses of the hydrologic cycle into a single software package. Through the coupling of surface water and ground water process models, represented by HSPF (Bicknell, et al ., 2001) and MODFLOW (Harbaugh and McDonald, 1996) models respec tively, IHM provides an adva nced predictive capability of the complex interactions of surface water and groundwater features in shallow water table environments. The model is characterized by a deterministic, distributed-parameter, semi-implicit real-time simulation model, with variable time steps and spatial discretization. The model components explicit ly account for all significant hydrologic processes including precipita tion, interception, evaporation, runoff, recharge, stream flow, baseflow, and all component storages of surface, vadose and deep groundwater zones (Ross et al ., 2005a and b). One particular problem for integrated hydr ologic models applie d to shallow water table environments is the effects from air tr apped in the voids of the shallow vadose zone. It has been widely recognized th at air entrapment plays a sign ificant role in shallow water table environments at limiting infiltrati on and increasing water table response to infiltration. Another phenomenon fo und in these environments is a rapid rise in the water level of observation wells screened below th e water table during high intensity rainfall events. The process, known as the Lisse E ffect (Weeks, 2002), occurs when an intense rainfall event effectively seals the surface so il layer thus trapping the soil air below the advancing wetting front. As th e wetting front advances, pres surization of the soil air
3 occurs. As a result of this increased ai r pressure, observation wells which are only screened below the water table show a rapid rise in water-level, despite the fact that the actual elevation of sa turation is essentially unchanged. As mentioned in Weeks (2002) the effect was noted as early as 1932 by Thal Larsen in the village of Lisse, Holland and was given its name by Hooghoudt (1947). Heliotis and DeWitt (1987), and Meyboom (1967) have reported observations of Lisse effect in water table hydr ographs; however, their explana tion is more from the point of view of identifying anomalies in water tabl e observations rather than a way to quantify air pressurization. Weeks (2002) attempted to mathematically link ai r pressurization to the anomalous water level rise in observation wells, but his analysis was overly simplistic and proved useful only for calculating the maximum possible water-level rise for a specific soil type Nonetheless the effort provides a background relating air-entrapment and water table fluctuations. Some previous work has been done to improve the conceptual basis of the IHM. Ross et al ., (2005a) advanced a new model to provide a smooth transition to satisfy ET demand between the vadose zone and deeper saturated ground water. The resultant IHM (v.1) approach provided a more sound representation of the actual soil profile than the predecessor model, the FIPR hydrologic model (FHM) (Ross et al ., 1977). Shah and Ross (2006) explored vadose zone storage conceptu alized in IHM (v.1) as well as the physics and mechanics of this moisture variability in shallow water table environments. Zhang and Ross (2006) showed the importance of diffe rentiating upper and lower regions of the unsaturated zone (vadose zone). Field soil mo isture observations and soil characterization data were used to formulate a recommended basis for the upper zone and lower zone in
4 IHM (v.1). Also, they developed a new methodology to describe relative moisture condition in both zones from field measurem ents to further test a model for soil hydrologic response. In shallow water table, low gradient e nvironments, the vertical behavior of the vadose zone, including the proximity of th e water table has been shown in field observations to dictate the r unoff and recharge rates for bot h seasonal and storm response (Rahgozar et al ., 2005). Therefore, for a model to ade quately predict thes e processes, the water table fluctuation, soil moisture condi tions, and ET fluxes from the vadose zone must be reproduced. With a very shallow groundwater table, the interaction between unsaturated and saturated groundwater b ecomes very strong. The groundwater table strongly influences the water co ntent in the unsaturated part of the root zone and the groundwater table represents a moving bounda ry between saturated and unsaturated conditions. In conceptualizing a model such as IHM, questions arise as to how well is the ET from interception, upper zone, lower zone and groundwater predicted? Also, does the water table described in IHM reasonably repr esent a real observati on? Aly (2005) applied a preliminary version of the IHM (v 0.9) to a small basin in West-central Florida. Results indicated that the model could reproduce gross ET and water table behavior but no distributed ET behavior was i nvestigated. Therefore, extensive investigation of the theoretical basis of vertical processes and demonstrati on of the performance of ET distribution in longer duration applic ations are still strongly warranted.
5 1.2 Objective and Scope In this study, considerations for mo deling shallow vadose zone moisture dynamics with integrated models were investig ated. There are four important issues that have been recently identified through field stud ies: 1) The hydrologic behavior of runoff, ET and groundwater recharge are controlled by vadose zone moisture which is strongly non-uniform. Observations indica te that the soil zone needs to be differentiated into a minimum of two separate distinguishable zone s. 2) For a given depth to water, within both soil zones variable moisture retention al so strongly effects hydr ologic response. 3) Also, field-scale variability (s trong differences in retention behavior) exists in shallow water table environments over very small spa tial scales (<100 m) requiring a different moisture retention model. 4) And, finally, ai r entrapment plays a significant role in controlling infiltration and observed depth to water table in shallow water table environments. Also, this study seeks to rigorously test one integrated hydrologic model against a detailed field-scale data set in West-Central Florida which includes soil moisture dynamics and ET distribution. The Integrated Hydrologic Model (IHM) uses a unique relative soil moisture approach for land segment integration and is intended to simulate the complex interaction between surface wa ter and groundwater systems. No prior rigorous investigation or validation of this model for the upper and lower for performance and predictive capability of vadose zone response has been made. In this study, continuous field soil moisture observations a nd soil characterization data were used to formulate a new basis for the upper zone and lower zone for possible testing and use in
6 the IHM. Several tests were performed to illustrate how the new conceptual model reduces field-scale variability in soil moistu re behavior and enhan ces representation of antecedent conditions. Evidence is presente d to document the existence of prolonged (many days) air entrapment and excess pore pr essures, which effect soil water storage and observed water table levels. The current study employed field data and numerical modeling to quantify the varia tion of air pressurization valu es. A simple analysis based on ideal gas law was also done to help understand air pressurization effects. Finally, intensive sensitivity tests were applied to the IHM model to investigate dependent hydrologic process response incl uding distribution of ET flux, depth to water table and recharge. It was desired that by repr oducing field data in a calibrated model and performing sensitivity testing further insight would be gained con cerning the reliability and calibratability of the model for regi onal scale investigations. Also, reproducing moisture retention behavior of the upper a nd lower vadose zone in shallow water table settings provides unique challenges for the integrated (combined surface and groundwater) hydrological models. The theory and vertical behavior of the IHM is examined through comparisons to collected fi eld data in West-Central Florida. These objectives were to (1) test a model of the vertical processes co ntrolling water table behavior, ET and recharge, (2) investigate the sensitivity of model parameters on model vs. observed vertical behavi or, and (3) offer recommendations for improvements and parameterization for regional model applica tion. Rigorous testing was done to better understand the robustness and/or limitations of the methodology of the IHM for upper and lower vadose zones.
7 CHAPTER 2 VADOSE ZONE CHARACTERIZATION IN SELECTED INTEGRATED SURFACE AND GROUNDWATER MODELS Six models selected for further invest igated included MIKE-SHE (Refsgaard & Storm, 1995; DHI, 1998), MOD-HMS (Hydr oGeoLogic Inc., 2003; Panday and Huyakorn, 2004), SWATMOD (Sophocleous et al ., 1999), tRIBS (Vivoni et al ., 2003), FHM (Ross et al ., 1997) and IHM (Ross et al ., 2004). Only in the cases of MIKE SHE was the linkage of groundwater and surface wate r components created as part of a unified model development process (i.e., specific focus of the code development). This fact illustrates the relative difficulty in designi ng integrated surface water/groundwater models. All the above models are relativ ely recent products. MODHM, SWATMOD and IHM were created by linking previously developed surface water and groundwater models. Only in the case of IHM are the component models (HSPF and MODFLOW) widely used in the industry. 2.1 SHE and MIKE SHE Freeze and Harlan (1979) proposed a blueprint fo r distributed hydrological modeling using a physics-based representation of the underl ying catchment processes. This blueprint was the basis for the devel opment of the European Hydrological System SHE (Abbott et al ., 1986a, b) and MIKE SHE (Refsgaar d & Storm, 1995). The original
8 MIKE SHE (DHI, 1998) model was devel oped and became operational in 1982, under the name Systme Hydrologique Europen (SHE). The model was sponsored and developed by three European organizations: the Danish Hydraulic Institute (DHI), the British Institute of Hydrology, and the Fren ch consulting company SOGREAH. MIKE SHE is a physically based, distributed, in tegrated hydrological and water quality modeling system for regional scale investig ation. It simulates the hydrological cycle including ET, overland flow, channel flow soil water and ground water movement. MIKE SHE is proprietary but the execu table code is widely marketed and available for a substantial licensing fee. The source code is generally unavailable. Two of the available unsaturated zone methods in MIKE SHE are 1) the full RichardÂ’s equation and 2) a simplified Richar dÂ’s equation that negl ects capillary tension. The full and simplified RichardÂ’s equation me thods use real soil properties and soil moisture-relationships that can be develope d using Brooks and Corey or van Genuchten relationships. A simplified wetland module that uses a linear relationship between depths to the water table and average soil moisture content and a linear in filtration equation can be used in place of the full and simplified RichardÂ’s equation modules. MIKE SHE includes a simplified ET model th at is used in the Two-Layer UZ/ET model in addition to the Kris tensen and Jensen model. Th e Two-Layer unsaturated model divides the unsaturated zone in to a root zone, from which ET can occur and a zone below the root zone, where ET doesnÂ’t occur. The Two-Layer UZ/ET module is based on a formulation presented in Yan and Smith ( 1994). The upper layer extends from the ground surface to the higher of the water table or the ET extinction depth (DHI, 2003). Several characteristics about the layer stru cture are summarized as follows:
9 1) It uses a conditional two layer soil struct ure, root zone and below root zone. 2) The upper layer varies and extends from the land surface to the higher of the water table or the ET extinction depth dur ing the run; the lower layer extends from the bottom of the upper layer to the wa ter table; If the wa ter table is above the ET extinction depth, the thickness of the lower layer is zero. 3) There are three options in MIKE SHE for calculating vertical flow in the vadose zone: (1) the full Richards equation (2) a simplified gravity flow procedure (3) a simple two-layer water balance method for shallow water tables. 2.2 tRIBS The TIN-based Real-tim e Integrated Basin Simulator (tRIBS) (Vivoni et al ., 2003) is a collection of C++ codes designed fo r distributed hydrologi c modeling at small to mid-size catchment scales (Vivoni et al ., 2003). The object-oriented software design offers several advantages over traditiona l procedural programmi ng. In particular, by grouping data and functions operating on these va riables into distinct classes, it becomes possible to separate the various hydrologic processes operating on th e TIN mesh from the procedures for creating the mesh itself (Tucker et al ., 2001). The object-oriented approach also allows for code modularit y, facilitating model development for other applications through code integration or s ubstitution of new process modules. Such a strategy permitted the development of th e tRIBS model from the CHILD modeling framework (Tucker et. al, 2001) within a reasonable amount of time and effort. Hydrologic modules from the RIBS model (G arrote and Bras, 1995) and new hydrologic
10 process models were incorporated into the CHILD framework as separate classes. In addition, the modularity allowed for the integr ation of additional process modules and the potential for finite element modeling (FEM) w ithin the existing mesh architecture (tRIBS User Manual, 2002). The tRIBS Distributed Hydrologic Model simulates the coupled surface and subsurface response to rainfall ove r complex topographies represented using multiple resolutions of triangular irregul ar networks (TINS). The public domain availability of tRIBS for both executable a nd source code is uncertain. It has some characteristics about laye r structure as follows: 1) The model reports to account for a partia lly saturated vadose zone and predicts the land surface response to continuous storm and inter storm stresses. 2) In vadose zone, one-dimensional infiltra tion, modified Green-Ampt infiltration scheme (Cabral et al 1992 and Ivanov, 2002) in the surface normal direction is redistributed by both the lateral fluxes in th e vadose zone and in the phreatic aquifer during storm a nd interstorm periods. 3) No soil layer structure is apparent. 2.3 MODHMS MODHMS (MODFLOW Hydrologic Modeli ng System) (HydroGeoLogic Inc., 2003; Panday and Huyakorn, 2004) is base d on MODFLOW and includes additional modules to simulate overland flow, channe l flow, and solute transport. MODHMS was developed by HydroGeoLogic Inc. Proprietar y, and is not freely distributed. Prior to development of MODHMS, HydroGeoLogic deve loped a number of codes to deal with
11 variable saturated, variable density, and multi-phase flow and transport primarily driven by an interest in transport processes. MODHMS is reportedly a physically based, spatially distributed, finite difference, inte grated surface water and groundwater model. It is actually a collection of codes used to interface with using the MODFLOW regular discretization. Datasets fo r an earlier version of MODHMS, MODFLOW-SURFACT, could be used to generate a basic fr amework for a MODHMS simulation. MODHMS is currently being test by St. Johns River Wate r Management District and the Southwest Florida Water Management Dist rict on basins exhibiting sh allow water table conditions. MODHMS is reportedly capable of modeling open channel flow and closed pipe flow (Priesmann slot) using the diffusi ve wave approximations. MODHMS also reportedly simulate structures (dams, weirs, culverts, and gates) with levels that vary between stress periods. Dynami c structure operations are not currently available in MODHMS. Overland flow is simulated usi ng the diffusive wave approximation and special provisions are available for flow be tween the overland flow plane and channels that depend on channel bank geometry. The surface water components have not been extensively applied to watershed scale and design problems. Water-quality capabilities are currently not availa ble for the surface water components in MODHMS. Characterization of the verified vado se zone in MODHMS is as follows: 1) The effects of depressions which include s rills, furrows and other detention features as well as of storage exclusion ha ve be taken into ac count in the modelÂ’s storage term as well as in the horizontal flow conductance term. 2) The storage effects of de pression storage and obstruc tion storage exclusion are modeled by assuming that the geometry of depressions and exclusions combined
12 has a maximum elevation and that the horizontal area covered by surface-water varies between zero and full area as th e water level rises from land surface (defined here as the bottom of the depr essions) up to this maximum elevation (land surface + height of depression storage + height of storage within obstructions. 2.4 SWATMOD SWATMOD (Sophocleous et al ., 1999) links the USDA model SWAT with the USGS model MODFLOW (McDonald and Harb augh, 1988). SWAT is a watershed-scale model used to predict water, chemical, a nd sediment movement in large basins. The model is used for extended time periods and not for single event flood modeling. The linked models are used to simulate long-term surface water and groundwater interactions, and do not simulate individual storm events. Ti me steps are typically daily or better. The SWATMOD model has been used to predic t conditions during simulation of water shortage periods (Sophocleous et al ., 1999). SWATMOD is reported to be public domain open source code but agency dist ribution support does not exist. 1) A limitation of SWAT is its inability to model the unsaturated zone beyond the root zone. Therefore, percolation (recha rge) is applied directly to the ground water table. 2) SWATMOD is really just a series of subroutines that links the two models: MODFLOW and SWAT. One subroutine, HYDBAL passes data between SWAT and MODFLOW and tracks the water balance of SWAT. The other MODSWB
13 links SWATÂ’s hydrologic basins with MODFLOWÂ’s grid and converts SWATÂ’s fluxes into flow rates for MODFLOW (Sophocleous et al ., 1999). 3) SWATMOD can be run in one of two modes. The first mode is where MODFLOW is treated as a subroutine of SWAT and is called at the end of each aquifer time step. The second mode involves SWAT and MODFLOW being performed successively and linked through a separate hydrologic balance data file (Sophocleous et al ., 1999). 4) Intended application watershed-sc ale model for long-term periods. 5) No soil layer structure. 2.5 FHM In 1988, the Florida Institute of Phospha te Research (FIPR) funded a research project to develop an advanced hydrologic model used for phosphate mine reclamation in west-central Florida. The intended pr oduct was to include a dynamically coupled comprehensive surface water and groundwater model. Each model component was to represent state-of-the-art cap abilities in hydrologic simulati on, including codes which are in the public domain, widely accepted and va lidated and compatible for integration. A geographic information system (GIS) databa se, as well as other available digital hydrologic and meteorological data sources (Powers, et al ., 1989) provided extensive data needs for this model. The model was to possess sufficiently simple user interfaces to provide for rapid applications and assessment of model results (Fielland and Ross, 1991;
14 Ross and Tara, 1993). The result of this effo rt was the FIPR Hydrologic Model (FHM), an integrated model which coupled HSPF and MODFLOW. 1) The basis for the lower zone storage in FH M was assumed to be that part of the vadose zone above the capillary fringe of the water table limited by the upper limit value of HSPF (254 cm or 100 inches). Nominal storage was assumed to be equilibrium moisture content at field capaci ty, and lower zone storage ratios could vary fro near zero to greater than 2.5, corresponding to near saturation of the lower zone. The FHM described spatia lly averaged ET behavior, but the parameterization (and calibrati on) was not explicitly tied to land use. This was considered a limitation of the FHM (and ISGW, a derivative model promoted by SDI, Inc.) and was one reason the model was later rewritten (Ross et al ., 2005a). 2) The FHM ET method was based mainly on coupling the ET methods of HSPF and MODFLOW. 3) The simplistic two-layer model. 2.6 IHM In the mid-nineties an early version of the FHM was adapted and modified by SDI under the name Integrated Surface and Groundwater model (ISGW). SDI and Tampa Bay Water used ISGW in the west-central Flor ida region for well-field pumping and surfacewater withdrawal investigati ons (SDI, 1999). Considerable review of that model and applications occurred through a series of projects (Ross, et al ., 1998; Waterstone, 2001; West Consultants, et al ., 2001). Recommendations resulted from those reviews to
15 reformulate the ISGW and apply, calibrate, a nd test the new model on a 4000 square mile region of west-central Florida. The outcome of a research effort to reformulate the theoretical and conceptual basi s of the model resulted in th e Integrated Hy drologic Model (IHM v. 1) (Ross et al 2004). All further reference to IHM v.1 will just be abbreviated IHM. IHM was designed to reportedly provide an advanced predictive capability of the complex interactions of surf ace water and groundwater features in shallow water-table environments. The model can be characteri zed as deterministic, semi-distributedparameter, semi-implicit, real-time formula tion, with variable time steps and spatial discretization. Reportedly, the model component s explicitly account for all significant hydrologic processes including precipitation, inter ception, evapotranspiration, runoff, recharge, irrigation flux applied to land, streamflow, wetland hydroperiod, baseflow, groundwater flow, and all the component storag es of surface, vadose and saturated zones. Input requirements include pr ecipitation and potential eva potranspiration time series, surface topologic features (i.e. land use, soil s, topography and derived slopes), irrigation fluxes, hydrography characteristics, rating c onditions, hydrogeologic parameters of the groundwater system and information about we ll pumping and surface-water diversions. Output includes detailed water balance info rmation on all major hydrologic processes, including surface water and groundwater fl ows to wetlands, streams and lakes, evapotranspiration losses from all storages, reach stage, soil moisture, recharge to the groundwater system and storage, heads and fluxes in the groundwater system. 1) Fundamental to the IHM is the definition of the lower zone storage, which is the moisture variability available to the root zone for an given water table elevation
16 that is above the wilting poi nt, or driest profile, for a given water table depth. For a deep water table the lower zone st orage can exhibit the largest values incorporating the rang e of variable soil moisture retention to an effective depth below the root zone (Ross et al 2005a). 2) It is said there are two zones in the va dose zone conceptualiz ation in IHM, the upper zone and lower zone. The upper zone is just the few inches upper soil layer and lower zone is the zone from upper zone down to the water table. However, clear definitions are still needed. 3) The IHM reportedly uses a theoretically sound, three-layer step -wise linear soil moisture retention model as opposed to a van Genuchten or other analytical retention model. 4) There were other improvements reporte d in IHM ET concept compared to the predecessor FHM (see, e.g., Ta ble 2.1 adapted from Ross et al ., 2005a).
17 Table 2.1. Similarities and Differences Between FHM and IHM ET Conceptualization (adapted from Ross et al ., 2005a) Component FHM IHM Interception ET Considered as first source Considered as first source Upper Zone ET Depression storage and shallow soil storage ET. Considered nest if available, supply rate based on relative storage Depression storage and shallow soil storage ET. Considered nest if available, supply rate based on relative storage Lower Zone ET (1) Deeper root zone storage based on all moisture above the capillary zone. (2) Uses seasonally variable plant coefficient. (3) ET rate based on relative moisture and remaining potential. (1) Deeper root zone storage based on all moisture above the capillary fringe in excess of the dry moisture profile. (2) Uses theoretically sound 3-layer soil moisture retention. (3) Base plant coefficients are seasonally variable, however, are dynamically adjusted by depth of water table (4) ET rate then based on relative moisture and adjusted remaining potential Groundwater ET (1) Uses MODFLOW linear extinction package. (2) Uses remaining potential after considering hierarchal storage contributions (1) Uses MODFLOW linear extinction package. (2) Uses consistent plant coefficients with lower zone after considering relative depth of water table. (3) Partitions potential af ter hierarchal storage contributions are met, considering relative depth of water table. (4) Provides smooth transition to free surface evaporation as capillary zone transitions land surface.
18 CHAPTER 3 OVERVIEW OF VERTICAL BEHAVI OR OF VADOSE ZONE IN IHM Among processes modeled in the vados e zone in IHM are the water table fluctuation, soil moisture conditions, and ET fluxes and distributions Details about the moisture flux and retention distributi on concepts in IHM can be found in Ross et al ., (2004) and Ross et al ., (2007). Only a brief summary is provided herein for completeness. 3.1 IHM 3-Layer Soil Moisture Model A 3-layer soil moisture model which is used as the basis for IHM landsegment integration, based on physical so il characteristics and represen tative of that soil type, describes the vadose zone storage behavior for any water table relative to the root zone (Figure 3.1).
19 Figure 3.1. Three-Layer Soil Moisture Model (from Ross et al ., 2005a) The first layer represents the near-s aturation capillary fringe, followed by the intermediate capillary rise both assumed fixe d by the soil type. For deeper water table conditions, the upper layer represents the near ly uniform soil moisture region above the capillary rise when the depth-to-water table is large enough for this layer to exist. Three profiles are shown corresponding to dry, equilibrium and wet soil moisture conditions of a mildly sorptive soil (e.g., fine sand or loamy sand). The thick lines on the figure represent the actual profiles in a uniform soil and the thin lines represent a stepwise, linear approximate profile developed for computational efficiency. A conceptual Dry Profile zWTb3= CF zRZ = zLSRZ Zone 3: Lower Capillary Zone (Capillary Fringe) (z) Zone 1: Upper Gravity Zone Zone 2: Intermediate Capillary Zone Wet Profile Equilibrium Profile zCZzCFb1b2= ICZ Land Surface z Root Zone Elevation a) Deep root zone ( RZ CZ) b ) Shallow root zone ( RZ < CZ ) Lower Zone Storage Upper Zone zLS Land Surface Root Zone Elevation Lower Zone Storage Upper Zone
20 representation of soil moisture is shown for a deep root zone in Figure 3.1(a) and for a shallow root zone in Figure 3.1(b). Variability of the moisture profile is dependent on the antecedent moisture condition and the water table proximity to or within the root zone Fundamental to the IHM is the definition of the lower zone storag e which is the moisture variability available to the root zone for any give n water table elevation that is above the wilting point, or driest profile, for a given water table depth. For a deep water table the lower zone storage can exhibit the largest values incorporating the range of variable soil moisture retention to an effective depth below the root zone (a ssumed to be the soil intermediate capillary zone thickness). This follows the physical behavi or that within the root zone plants can reduce the moisture content to near wilting and therefore reduce the moisture retained (over a limited region) below the root zone due to capillary suction gradients. This is a formal definition for the lower zone soil storag e, which, interestingly, is still true to the imprecise Â“hydrologically activeÂ” soil moisture definition used by HSPF and the original model, the Stanford Watershed Model (Bicknell, et al ., 2001). Following satisfaction of PET from interception and depression storages, remaining potential ET ( PET ) is applied (and partitioned) to the vadose zone storag e (lower zone) and directly to groundwater (water table), depending on th e proximity of the water ta ble within the root zone described below.
21 3.2 Interception, Depression and Surface Detention Storage Conceptually in an IHM and consistent with typical applic ation of HSPF alone, interception is assumed to be the first extraction for a storm event and can be a significant loss if the land segment possesses substantial vegetative cover. As the interception storage capacity is filled, pr ecipitation begins filling the surface depressions and, for pervious surfaces, infiltration commences. As rainfall proceeds, soil infiltration capacity diminishes with increasing soil moisture a nd Hortonian rainfall ex cess (with or without air entrapment) and/or saturation excess r unoff can contribute to overland flow. The surface storages and unsaturated zone in IHM are depicted in Figure 3.2. X Land Surface (z) ET Extinction DepthRZ CZ Water Table Interception Surface Detention Water Table Zone (DLZS)Saturated Zone Upper Unsaturated Zone and Surface Depression Storage(DSURS)Unsaturated Storage(DCEPS) Stream or Wetland Storage Lower ) (DUZS) X Land Surface (z) ET Extinction DepthRZ CZ Water Table X Land Surface (z) ET Extinction DepthRZ CZ Water TableX Land Surface (z) ET Extinction DepthRZ CZ Water Table Interception Surface Detention Water Table Zone (DLZS)Saturated Zone Upper Unsaturated Zone and Surface Depression Storage(DSURS)Unsaturated Storage(DCEPS) Stream or Wetland Storage Lower ) (DUZS) Interception Surface Detention Water Table Zone (DLZS)Saturated Zone Upper Unsaturated Zone and Surface Depression Storage(DSURS)Unsaturated Storage(DCEPS) Stream or Wetland Storage Lower ) (DUZS) Figure 3.2. Storages Pertaining to the Vadose Zone Described in IHM
22 Interception storage (water stored on the canopy of vegeta tive cover, building roofs and other surfaces) is primarily a function of land use. In IHM, depression storage, also referred to as upper zone storage ( DUZS), includes micro-depre ssion features such as cracks, potholes, small yard depressions, a nd water required to wet ground litter. Depression storage is primarily a function of surface conditions such as land use, topography, and time of year. Su rface detention storage is water contained in rainfall excess that is available for runo ff: rainfall that is not infiltra tion, interflow, or captured in interception or depressions. The amount of water in surface detention storage is a function of rainfall intensity, infiltrati on capacity, hydraulic slope, hydraulic length, Manning's roughness coefficient and degree of saturation of the lower zone. IHM uses these conceptual definitions in a more physic ally-based parameterizating of the empirical equations found in the component model HSPF (Bicknell, et al ., 2001; Ross et al ., 2004) The following sections describe the co mponents and equations used in this interpretation. Assumptions and Equations Water that is not intercepted and is in excess of infiltration is termed potential direct runoff ( PPR). Consistent with HSPF, surface de pressions can capture much of the runoff in some landscapes. In HSPF (and IHM) surface depressions are referred to as upper zone storage. Then, the fraction of potential direct runoff which becomes upper zone storage, FUZFRAC is a function of the relative mo isture condition of the upper zone determined by the ratio, RUZRAT of the upper zone storage, DUZS to the upper zone nominal (depression) storage, DUZSN. Upper zone nominal storage is difficult to determine from land cover/soil conditions apriori so is normally adjusted during calibration. The
23 two equations below represent the method of subroutine UZINF2 in HSPF. The HSPF derived equation for FUZFRAC (adapted from Bicknell et al ., 2001) when RUZRAT 2 (drier conditions) is, UZRATR UZRAT UZRAT UZFRACR R F 34 1 2 1 (3.1) when RUZRAT > 2 (wetter conditions), the equation is, 3 21 2 1 UZRATR UZRAT UZFRACR F (3.2) The inflow to upper zone storage, IUZI, is determined by, PDRO UZFRAC UZID F I (3.3) where DPDRO is the volume of potential direct runoff, which is determined based on lower vadose zone conditions describe d in later sections. Upper zone storage can then be calculated as, t PERC t t UZS t UZI t UZSI D I DP (3.4) where, ( t ) and ( ttP) = superscripts refer to current and prior model time interval respectively IPERC = vertical percolation from uppe r zone to lower zone per model time interval [L]; define d in the following sections tP = the HSPF PERLND user defined computational time step (e.g., 15-mins.) [T]
24 The upper zone flux equations represent th e only completely impirical equations included in IHM still used directly from HSPF. But, to date the treatment of depression storage as a hydrologic process is only by empirical equations. Surface detention storage, DSURS, is the volume of water stored on pervious or impervious land as temporary rainfall excess (instantaneous mean depth of the kinematic overland flow wave). DSURS is a temporary land-surface storage that can become surface overland flow (runoff), infiltration, upper z one storage, or interflow. In HSPF, DSURS is determined by subtracting infiltration, upper zone, interflow, and overland runoff fluxes from the moisture supply, DMSUPY. For each model time interval DSURS is determined by, SURO PSUR SURSQ D D (3.5) where, DPSUR = potential surface detention st orage volume [L] QSURO = overland runoff flow [L] DMSUPY = moisture supply to the surface detention storage process for the current model time interval [L] 3.3 Vadose Zone Storage The vadose or lower zone in the IHM is defined as the remainder of the unsaturated zone between the upper zone and th e saturated zone below the water table. It accounts for the much of the sustained tran spiration burden of the vegetative cover Unique to the IHM, the lower zone storage is that part of the vadose zone moisture that does not affect the water table. Excess moistu re in the vadose zone that becomes recharge
25 to the water table is not include d in lower zone storage. In HSPF, soil moisture content is not explicitly calculated. HSPF considers the vado se zone soil moisture to be contained in the lower zone storage volume, DLZS representing the hydrologically active moisture. The infiltration, percolation from upper z one storage and evapotranspiration involving the lower zone are each a func tion of the relative moisture condition of the lower zone given by the ratio, RLZRAT defined as, RLZRAT = DLZS/DLZSN (3.6) where DLZSN is a nominal storage volume equal to the moisture that can be stored in the vadose zone between the equilibrium moisture profile for dWT > CZ and the dry moisture profile. The term equilibrium vadose zone storage is used to refer to DLZSN which is computed in IHM for a lower zone thickness, bLZ as, dz z dz z DLZ LZb b LZSN dry m equilibriu (3.7) The dry moisture profile dryz ) ( depends on depth-to-water ta ble and the position of the top of the capillary zone. The lower limit of the dry moisture profile, the wilting point moisture content, can be reached only if the top of the capillary zone is below land surface. For the IHM, considerable review a nd testing of the functional form and conceptual basis for DLZS and DLZSN were made (e.g., see Zhang and Ross, 2006). Nevertheless, the lower zone storage ratio, RLZRAT can be considered as a functional and now specifically defined expression for the re lative vadose zone soil moisture condition.
26 For the IHM, the following limits for RLZRAT result: (1) RLZRAT = 0 corresponds to the dry moisture profile, (2) RLZRAT =1 corresponds to the equilibrium moisture profile and (3) RLZRAT >1 corresponds to a persiste nt moisture profile wetter th an equilibrium. Note that short-term (< 1 day) transient moisture flux (i.e., from a wetting fr ont) is not included in the lower zone storage but is tracked as groundwater recharge. When dWT is near or less than CF additional functional limits on DLZSN and specific yield must be used. To avoid nu merical errors with division by zero, HSPF imposes the limit DLZSN 0.01. With the form proposed above, DLZSN < 0.01 can occur as dWT is near or less than CF. Therefore, when DLZSN is calculated to be less than 0.01, IHM sets DLZSN =0.01 until such time as DLZSN is calculated to be greater than 0.01 by integration algorithms of IHM. While DLZSN = 0.01, specific yield is set to a functional lower limit by IHM which is discussed in a subsequent section. Also note that no mass balance concern is raised by this functional limit as the actual volume is DLZS is allowed to vary to zero (wilt point) moisture condition. HSPF also imposes a maximum limit DLZSN 100 for deep water table and root zone conditions. Recall, IHM defi nes the lower zone thickness, bLZ to be the thickness of the root zone above the water table. Given the definition for DLZSN in equation 3.7, it is believed that the limit DLZSN 100 will not be violated for reasonable root zone thicknesses. Nevertheless, IHM applies this limit whenever computed DLZSN exceeds 100. Again, no mass balance concern arises as DLZS is functionally not limited.
27 3.4 Infiltration Infiltration is the movement of water from the soil surface into the unsaturated lower zone with some high amounts becoming recharge in a simplified form of Phlips equation used in HSPF (Fielland and Ross, 1991). However, infiltration/recharge is derived from the unique interp retation for unsaturat ed soil (lower) zone storage in IHM. Percolation is defined as th e vertical movement of wate r from upper zone storage to lower zone storage or saturated groundwat er storage thus becoming an important component of groundwater recharge. IHM redistributes infilt ration and percolation between vadose zone storage and recharge to the water table by considering the proximity of the water table and relative moisture condition of the vadose zone. The HSPF calculation, (lFLZFRAC ), determines the fraction of the volume of infiltration and percolation that is redistribut ed to recharge. Disposition of infiltration into vadose zone storage and/or groundwater recharge is uniquely interpreted in IHM. In the context of IHM, infiltration is simulated using the simplified Philip equation with concern for physical soil hydr aulic conductivity, sorp tion behavior and time step sensitivity shown in previous study (Fielland and Ross, 1991; Geurink and Ross, 2006). Infiltration is a function of ma ny factors including soil type, moisture content, air entrapment conditions, vegetativ e cover and the depth to the water table. Infiltration also depends upon redistributi on (vertical downward movement of water within the soil) from previous events. When infiltration exceeds the capacity of the unsaturated zone to vertically redistribute water for extended periods, surf ace saturation occurs. This can occur from
28 infiltration excess (precipitati on exceeding infiltration capac ity), with or without air entrapment excess void pressure, and/or fully saturation excess (saturated soil conditions) mechanisms. In IHM, fully saturation ex cess condition is only allowed over the distributed discretization provided by regul ar grid cells forming the MODFLOW ground water domain (Ross et al ., 2004). Where variable saturated cells exist, the infiltration rates for pervious land cover are adjusted downward in an area-weighted manner (Ross et al ., 2004). Code improvements to provide for explicit variable saturation are being considered and/or tested but are not yet implemented. Infiltration in HSPF and IHM is a functi on of infiltration capacity (the maximum rate at which soil will accept in filtration), lower zone storage, DLZS, and lower zone nominal storage DLZSN Infiltration capacity is a function of soil and environmental conditions which can vary spat ially. Infiltration cap acity also varies with time as a function of the antecedent moisture condition ( RLZRAT). When rainfall supply exceeds the infiltration capacity, water is allocated to ot her storages and fluxes (e.g., surface detention storage). Therefore, infiltration capacity is a function of both fixed and variable characteristics of the watershed. Generally, fixed characteristics include such parameters as soil type and land-surface cover; vari able characteristics include soil surface conditions, soil moisture content and dept h-to-water table. Fixed and variable characteristics vary spatially over the land segment. Traditionally, HSPF uses a linear probability density function (Figure 3.3) to ac count for spatial variation of infiltration over the land segment. This function allows for simple characterization of field-scale variably which has been shown to exist at th e length-scales of 10Â’s of meters in WestCentral Florida (Zhang and Ross, 2006).
29 The linear probability density function (PDF ) that describes the spatial variability of infiltration in IHM relates maximum to mean infiltration capacity. This linear PDF changes in time using algorithms to repres ent the dynamic nature of the infiltration LINE I LINE II IIBAR IIMAX IMAX 100 50 0 % OF AREA MSUPY IIMIN IMIN IBAR POTENTIAL SURFACE DETENTION / RUNOFF / UPPER ZONE INFLOW POTENTIAL INTERFLOW INFLOW/UPPER ZONE INFLOW INFILTRATION INCHES OF WATER / INTERVAL Figure 3.3. HSPF Linear Probability Density Function (after Bicknell, et al ., 2001) capacity as a function of the changing so il moisture in the unsaturated zone. The governing equations represent the dependen ce of infiltration ra te on soil moisture conditions and are based loosel y on the work of Phillip ( 1957) as adapted into the Stanford Watershed Model (Crawford and Lins ley, 1966), explored in Fielland and Ross, 1991. IHM implements the infiltration PDF over pervious land segments in the following manner. The spatial mean infiltration capacity of the land segment IBARI is
30 determined from relative soil moisture condition, represented by the ratio of DLZS to DLZSN. Unique to the particular water table depth derived from MODFLOW component, IBARI is then multiplied by the landuse/soil base d ratio of maximum to mean infiltration capacity, PINFILD to determine the maximum infiltration capacity of the land segment IIMAX. The value of PINFILD can vary from 1 to 2, with a value of 1 corresponding to spatially uniform infiltration and a value of 2 corresponding to the maximum spatial variability in infiltration. A value of 2 for PINFILD yields a minimum infiltration capacity for the land segment, IIMIN of zero and a maximum infiltration rate IMAX that is twice the mean. INFEXP PLZSN LZS INFILT IBARD D P I (3.8) IBAR INFILD IMAXI P I (3.9) IMAX IBAR IMINI I I 2 (3.10) where, PINFILT = infiltration index [L/T], equal to the mean infiltration rate of the soil at equilibrium moisture c ondition and average water table depth; IBARI = soil moisture dependent spatial mean infiltration volume over the pervious segment per model time interval [L]; DLZS = HSPF derived time and depth to water table dependent lower zone storage volume [L];
31 DLZSN = available soil moisture retenti on at equilibrium, depth to water table dependent lower zone nom inal storage volume [L]; PINFEXP = soil retention based exponent (gre ater than 1) expressing the soil sensitivity to variable soil moisture higher for clays and lower for sandy soils (see Geurink and Ross, 2006); IIMIN = HSPF derived minimum infiltra tion rate expressed as a volume per model time interval [L]; IIMAX = HSPF derived maximum infiltration volume per model time interval [L]; PINFILD = landuse based ratio of maximum to mean infiltration capacity over the subbasin (expressing variability in infiltration conditions over the land segment) The IIMIN IBARI, and IIMAX points represent the infiltration lin e (Line I, see Figure 3.2). All moisture supply (DMSUPY) below the infiltration line is considered as infiltration inflow, IINFIL. Above the infiltration line, DMSUPY is considered to be potential direct runoff, DPDRO DMSUPY consists of the precipitation water remaining after in terception and upper zone storage are removed, plus lateral overla nd inflow from an adjacent land segment for the current model time interval, plus the su rface detention storage remaining from the previous model time interval. If the moisture supply is less than th e minimum infiltration capacity, then all moisture is assigned to infiltr ation. If the moisture supply is greater than the maximum
32 infiltration capacity, then infiltration is the m ean infiltration capacity, and potential direct runoff is the remaining moisture supply. When the moisture supply is greater than the minimum infiltration capacity but less than or equal to the maximum infiltration capacity, infiltration occurs variably over the whole domain. Also potential direct runoff only occurs over part of the domain. In all cases, the infiltration potential line is established prior to calculation of the infiltration and the potential direct runoff each time step based on relative moisture condition of the soil. Wate r that is infiltrating combines with water that is percolating from the upper zone storage to the lower zone storage as lower zone inflow plus groundwater (w ater table) recharge. 3.5 Recharge In the HSPF application w ithin IHM, active groundwater storage is turned off because it is explicitly accounted for by the MODFLOW code and IHM integration components. In IHM, recharge to MODFLO W is the groundwater inflow volume, IGWI, from HSPF. Because of variable discretizati on, recharge from multiple land segments comprising a rectangular groundwat er grid element must be ar eaweighted. Details about discretization can be found in Ross et al. (2004 and 2005b). However, due to the unique interpretation of the lower zone (LZ) in IHM, the percolation di stribution function in HSPF warranted modification. Consistent with the formulatio n of HSPF, the fraction of infiltration that becomes recharge to the water table is a function of the relative moistu re condition of the LZ (RLZRAT). Also, already noted, RLZRAT = 1 corresponds to equili brium moisture retention.
33 The fraction of infiltration percolation that stays in the LZ (recharges the vadose zone) is FLZFRAC. Allowing for field-scale variability a nd potential significant macro-pores (bypassing through the vadose zone and all uncer tainty in soil retention and percolation processes), the FLZFRAC form can be take on the characteristic of the solid line in Figure 3.4 which is the default HSPF formulation. However, in field-scale observations (Rahgozar et al., 2005) and theoretical i nvestigations (Shah and Ross, 2006) distribution is shown to be more consistent with the form ulation of the behavior depicted by lines 1, 2 and 3 in Figure 3.4. For this modified (optio nal) formulation, infiltration is completely vadose zone recharge until equilibrium retention is observed and water table recharge only commences for wetter conditions with this fraction rapidly approaching zero for sandy soils. The fraction FLZFRAC, controls the distribution of in filtration plus percolation from upper zone storage to vadose zone storage and recharge to the water table. The default form in HSPF allows continuous variation of FLZFRAC for the complete range of RLZRAT (i.e., some finite recharge occurs to the water table even when the soil moisture content is near wilting point) arguably un likely in most cases. Therefor e, the alternate conceptual basis for FLZFRAC was sought that provided for vadose zone recharge to the equilibrium
34 Figure 3.4. Distribution of Lower Zone Infiltration and Percolation moisture content before there is water table recharge. This is simply, 1 for 1 0 for 1LZRAT R k LZRAT LZFRACR e R FLZRAT FLZ (3.11) Where, kFLZ is a decay rate accelerator to ac count for the tendency for the wet soil moisture profile to reside near the equilibrium profile (RLZRAT = 1) characteristic of more sandy soils. Refer to Figure 3.3 for a comparative plot of FLZFRAC for the HSPF default formulation and for different parameters values for kFLZ (e.g., kFLZ = 3, 5, and 7). IHM allows either use of the HSPF defa ult or the altern ate concept for FLZFRAC.
35 3.6 Evapotranspiration IHM partitions ET between surface storag es, vadose zone storage and saturated ground water storage by consider ing evaporative flux from su rface sources, proximity of the water table to land surf ace, relative moisture condition of the unsaturated zone, thickness of the capillary zone, thickness of the root zone and relative plant cover density in the manner of Ross et al. (2005a), briefly summarized below. While both HSPF and MODFLOW have ET subroutines, which are often used separately, IHM actually employs both in a unique interpretati on and hierarchical approach (see in Figure 3.5). IHM accounts for ET following user specification of a potential atmospheric evaporation-rate (PET) time series determined apriori based on estimates from open pan data, Penman refere nce ET calculations or other meteorologic data. IHM considers ET distribution in a unique hierarchal a pproach considering satisfaction of PET by surfacewater storages first, starting with interception (QCEPE), then depression storage (QUZET) then proceeding to distribute reduced PET to the vadose zone (lower zone ET, QLZET and/or water table, QGWET). Both vadose zone and saturated groundwater ET are dictated by vegetative cove r characteristics, in cluding monthly plant coefficient (PPC), root-zone (rhizosphere) depth (RZ ), soil characteristics and depth to the saturated groundwater (WTd). The extinction depth (x ),and maximum ET surface (ZMAXET) for the EVT package of MODFLOW are di stinctively defined in the manner of Ross et al., (2005a), following the physical behavi or of extinction elucidated in Shah et al., (2006).
36 Figure 3.5. Vertical Moisture Fluxes and Storage in IHM The ET concept of IHM considers water stored in the vadose zone and groundwater as one unit from wh ich transpiration and direct evaporation from the soil occurs. Groundwater is available for ET th rough upward capillary flux from the water table into the root zone and from direct cont act between the water table and the root zone. IHM provides a smooth transition from all uns aturated zones suppor ting ET, to shallow water table free surface evaporation. The same plant coefficients regulate water uptake for both unsaturated and saturated zones. Al so, there is smooth transition from plantbased uptake rate to free surface (open water) direct evaporation as the water table nears land surface. The definition for the lower zone presen ted in the previous section explicitly tracks the antecedent moisture variability (w et or dry conditions) for any given water table depth. All component part itioning (i.e., upper zone, lowe r zone and water table) are Infiltration Upper Zone Stream Channel Impervious Lens Saturated Zone Water Table Unsaturated Zone Percolation Lower Zone Interflow Interception ET (CEPE) Upper Zone ET (UZET) Lower Zone ET (LZET) Groundwater ET (GWET) Precipitation Infiltration Upper Zone Stream Channel Impervious Lens Saturated Zone Water Table Unsaturated Zone Percolation Lower Zone Interflow Interception ET (CEPE) Upper Zone ET (UZET) Lower Zone ET (LZET) Groundwater ET (GWET) Precipitation Interception ET (CEPE) Upper Zone ET (UZET) Lower Zone ET (LZET) Groundwater ET (GWET) Precipitation Unsaturated Lower Zone
37 true to the above moisture retention beha vior. Groundwater is available for ET through upward capillary flux from the water table into the root zone and from direct contact between the water table a nd the root zone (Ross, et al., 2005a). In IHM, ET demand from saturated groundw ater zone occurs only when the water table is above a well-defined groundwater ET extinction depth,x The extinction depth has been theoretically show n to be the sum of the pl ant root zone thickness (RZ ) and, roughly, the soil capillary zone thickness (CZ ) (Ross, et al., 2005a; Shah et al., 2006) and this is the approach used in IHM. Cons istent with field study observations in shallow water table conditions of Florida, IHM transf ers more of the ET burde n of plants to the water table than previous versions of th e model (i.e., FHM) and other widely-used models (Ross, et al., 2005a; Ross, et al., 2004) to be more consistent with field observations in shallow wa ter table hydrology (Rahgozar, et al., 2005). Another important variable in IHM is the monthly variable plant coefficient, PPC Ross et al., (2005a). Vegetative ET (plant uptake) time series can be derived from field data, e.g., Rahgozar, et al., (2005). Typical applications of IHM then require the following input variables controlling ET: (1) User specifi cation of an atmospheric open-water potential ET rate, PET(t), which varies continuously in time, (2 ) Derived plant/soil extinction depth x (based on GIS overlays of vegetative root zone thickness plus soil capillary zone thickness), and (3) A vegetation based, monthly variably, plant coefficient, PPC..
38 CHAPTER 4 METHODOLOGY FOR SIMULA TING VADOSE ZONE There are four important model consid erations of recognized importance and investigated herein as combined factors fo r simulating vadose zone moisture. 1) Soils have predictable but highly vari able moisture retention propert ies. 2) Recent investigation from field data have shown that hydrologic behavior including runo ff, recharge and ET are controlled by vadose zone moisture retent ion which is strongly non-uniform. 3) Fieldscale variability is pronounced, even for sim ilar soils and land cover. 4) For shallow water table settings, air entrapment strongly effects infiltration and observations of water table. This chapter proposes the methodology to investigate these considerations for simulating vadose zone moisture retention behavior in IHM and follows up with a parameter sensitivity analysis for vadose zone prediction. 4.1 Soil Zonation Recent investigations of field data have shown that the hydrologic behavior of runoff, ET and groundwater rech arge is controlled by vadose zone moisture which is strongly non-uniform (Rahgozar et al., 2005). Vertical observations indicate that the soil zone needs to be differentiated into a minimu m of two separate dist inguishable zones. For example, observations from field data indica te that 50-70% of the total ET comes from a
39 clearly identifiable distinct soil zone very near land surface (Rahgozar et al., 2005). This top 10-20 cm of soil, effectively comprisi ng the A horizon, has been proposed as effectively defining the upper zone (Zhang a nd Ross, 2006). Most soil moisture available to the root zone, especially in sustained dry periods, however, is stored in the lower part of the vadose zone, defined herein as the lo wer zone. Root moisture uptake from this layer contributes to the sust ained soil ET burden. Field data is presented showing that these two zones can be and frequently are in different antecedent (i.e., relative wet or dry) states (following the work by Zhang and Ross, 2006). Data also indicates that stations in close-proximity exhibit significant field-scale variability. Most integrated models (noted in Chapter 2) do not differentiate the vados e zone (especially th e hydrologically active vadose zone). For IHM, clear definitions are still needed. 4.1.1 2-Layer Soil Discretization in IHM The surface storages and unsaturated zone in IHM are partitioned into two layers, the upper (UZ) and lower soil z one (LZ). The upper soil zone plays an important role in surface hydrologic response, which has a direct effect on the ET, including direct soil evaporation, infiltration, and runoff. Figure 4.1 is a simple conceptualization of the UZ and LZ as proposed by the IHM. Note the insert in Figure 3.2 shows the root zone, which in shallow water table settings, can extend down below the upper and lower vadose zone. The figure also depicts other perv ious upland storages in the IHM In the IHM, it is proposed th at the upper zone includes shallow soil storage and surface depression storage from micro t opography (horizontal scale < 1 m), mesodepression storage (1-100 m), and any larger storage features not e xplicitly included as
40 hydrography in the model. Micro-depressi on storage includes cracks, potholes, depressions, etc., on the land surface that can hold water and remove water from runoff supporting delayed infiltration and direct eva poration after each rainfall event. Water captured in these depressions and shallow so il moisture then consequently furnishes much of the post-storm event evaporati on demand, thereby co mprising a significant fraction of the annual hydrologic budget (Ross et al., 2005a). For larger basin applications (with more coarse discretizati on), macro-scale depression storage features, including small isolated wetlands, ponds and sinkholes, can become significant components of the pervious land segment depression storage. The upper vadose zone is aff ected by the initial abstra ctions, interception and depression storages in the IHM. Interception is assumed to o ccur first during storm events and can be a significant loss if the land segm ent possesses substantial vegetative cover. As the interception storage capacity is filled, precipitation begins filling depressions and contributes to infiltration. Depending on antecedent conditions, depression storage (included in upper unsat urated zone storageUZSD) is a significant rainfall capture mechanism with regards to generation of runoff and recharge and may reach capacity relatively quickly (Ross et al., 2005a). Following rainfall, water moves out of the upper unsaturated zone storage and pe rcolates to lower unsaturated zone, where it is available for sustained plant transpiration. Vadose zone water flux that becomes water table recharge (i.e., results in wate r table movement) is not part of the lower zone storage. A more thorough presentation of ET in the IHM can be found in Ross et al. (2004, 2005a). The lower zone in the IHM is defined as the remainder of the unsaturated zone between the upper zone and the saturated soil above the water table. It accounts for the
41 sustained transpiration burd en of the vegetative cover Unique to the IHM, the lower zone storage is that part of the vadose zone moisture that does not affect the water table. Excess moisture in the vadose zone that b ecomes recharge to the water table is not included in lower zone storage. Surface and soil hydrologic pro cesses in the IHM are further discretized using irregularly shaped but hydrologically si milar, hydrologic re sponse units (HRUs). However, very small HRUs with like propertie s (e.g., grass land with similar soils) within a meteorological region are gr ouped for computation. Even within these HRUs, fieldscale variability exists. For large regional appl ications, sufficient data does not exist nor is it practical to solve RichardÂ’s equation for each unique soil moisture distribution. Subsequently, runoff and recharge are distributed over irregular hydrographic discretization for surface water and regular (grid cell) discre tization for below water table ground water flow computations. More detail about the discretization of IHM can be found in Ross et al. (2005b). 4.1.2 Upper Zone as the A Horizon In West-Central Florida (i.e., coastal plai n type) soils, there are several distinct layers or horizons of hydrological importanc e. When one examines a hole dug at the study site, what is observed is fairly typical of any of the upland soils in the Gulf coastal plain. These soils are made up of distinct so il layers consisting of O, A, B, C, E and R classifications. However, most of the co astal plain soils have, at most, three hydrologically distinct horizons, that is, the surface horizon A, the subsoil horizon B, and
42 the substratum horizon C which can be identified from soil classification data (Carlisle et al., 1989). Some soils have an organic horizon O on the surface, or buried at some depth. In this study, particular attention is paid to the A horizon which comprises the topsoil, rich in organic matter and typically darker in color than the deeper soils. The A horizon is the zone of major biological activity. Here, pl ants and animals and their residues interact with an e normously diverse and dynamic mu ltitude of microorganisms. There is considerable moisture retention capability in pore spaces (including macropores) and readily available air. Macro-pores and extensive root matter are readily apparent in this layer. Figure 4.1 shows t ypical soil profiles in sedimentary soils and graphical depiction of upper and lower soil moisture zones. Figure 4.1. Typical Soil Profiles in Sedimentary Soils and Graphical Depiction of Upper and Lower Soil Moisture Zones (modified from NRCS web source) Reviewing soil classification data (Carlisle et al. 1989), one finds that fine sand, fine sandy loams and sandy loams comprise th e bulk of soils in Fl orida. The A horizon Upper Zone Lower Zone O B C A Horizon
43 averages 15 cm ( 5 cm) throughout the domain with very little variabilit y in thickness (Figure 4.2). Therefore, the upper zone can be conveni ently described as the A horizon with distinct hydrologic propertie s characteristic of the top 10-20 cm soil layer for these coastal plain soils. This layer consists of extensive organic material and micro1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean Water Content (%)Elevation above Water Table (log) 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean Water Content (%) 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean Water Content (%) 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean Water Content (%) 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean Water Content (%)Elevation above Water Table (log) 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean Water Content (%) 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean Water Content (%) 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean 1 10 100 1000 01020304050 % Water ContentElevation above WT (log) Levyvillefine s and A Kendrick fine sand Ap Sparrfine sand Ap Bradenton fine sand Ap Feldafine sand Ap Malabar fine sand Ap Valkariafine sand Ap Non-designated fine sand A Ft.GreenAp Mabel fine sand A Sumtervillefine s and Ap Adamsville fine s and Ap Lake fine sand Ap EauGalliefine sand A Oldsmar fine sand A1 Vero fine sand A Apopka find sand Ap Arredondo sand Ap Sparrfine sand A Tavares fine sand Ap Myakka fine sand Ap Vero Variant fine sand Ap Millhopperfind sand Ap median mean Water Content (%) Figure 4.2. Equilibrium Moisture Retention Characteristic of the A Horizon in Florida Fine Sandy Soils topography depressions that are indistinguishable as a storage unit. Also, the storage characteristics of this unit are governed by the proximity of the water table which is further explored below. 4.1.3 Observations from Field Studies Nested transect shallow water table (5 m) wells were installed in an intensive study area located in a typical shallow water table, coasta l Flatwoods and pasture land
44 setting in West-Central Florid a (see Figure 4.3). The wells ut ilized for this investigation were designated as PS43, PS42, PS41, PS40, USF1 and USF3 (Figure 4.3). Figure 4.3. Location Map of Observation Wells and Soil Moisture Monitoring Sites The vegetation in the upland area wa s primarily ungrazed Bahia grass. The vegetative communities adjacent to the stream were dominated by alluvial mixed Slash Pine/hardwood (water oak) forested wetland fa irly typical of undeveloped West-Central Florida. Green foliage density at this site is nondeciduous but follows a typical seasonal pattern, reaching maximum coverage during summer wet periods (June to August) and minimum coverage during winter dr y periods (December to February). Vertical profiling soil moisture probes were installed adjacent to monitoring wells PS43, PS42, PS41, PS40, USF1 and USF3. Ver tical resolution was achieved with sensor placement at 10, 20, 30, 50, 70, 90, 110 and 150 cm below land surface. The moisture
45 probes from PS43 to PS40 were along a downhill transect from the upland grassed area at PS43 to the riparian forest near the stream at PS40. Ten minute data where converted to daily average values for th e period (1/1/2002-6/27/2004) for this analysis. All soils classified for the site are hydrologically similar to Myakka Fine Sand (Carlisle et al., 1989). For the upper zone behavior, only the top sensor (10 cm) was used while the remaining seven sensors below were used to describe the lower soil zone moisture content. An important observation from soil mois ture measurements was the pronounced occurrence of field-scale variab ility. Figure 4.4 illustrates the variability in soil moisture from six synoptic observation wells at the st udy during a deep (a) a nd shallow (b) water table period. The spacing between these observation wells was relatively small horizontal distances (~100 m) in near identical hydrological settings The distance between USF1 and USF3 was less than 30 m. These six obser vation wells were approximately 5 m deep, cased for the first meter and screened below that. The soil classification for all stations was Myakka Fine Sand. The moisture observatio ns shown in Figure 4.4 are typical of the strong variability of moisture retention within field-scale ho rizontal dimensions of 10-100 meters. This scale of variability is clearly smaller than (or comparable to) the discretization scale of most re gional models (>100-1000 m). Imp lications for this relative scale of high variability may be obvious but more is discu ssed about this to later in this study.
46 Figure 4.4. Representative Field-Scale Variability Shown by Concurrent Vertical Soil Moisture Observations from Six Stations in (a) Deeper Wa ter Table Period; (b) Shallow Water Table Period When plotting the observed total soil moisture, derived by depth integration over the respective vadose zones, ag ainst depth to water table for the upper and lower zones, respectively, several interesting observation can be made. First, strong vertical variability in moisture retention is observed at all sta tions. Second, it is obse rved that field-scale variability occurs in both zones for the six observation stations even though all six sites have the same soil classification (i.e., Myakka Fine Sand) and nearly identical texture classes. From daily observations (exhibiting daily characteristics very similar to this one
47 example) it was observed the total soil moisture is highly varied most of the time. What is offered as a hypothesis is that the different stations have slightly different moisture retention characteristics (from very small di fferences in % clays and/or organics) and slightly different depth to water table ev en though they are in similar antecedent condition. Consequently, what is pr oposed is a unique formulation of relative soil moisture and the abandonment of representation of actual soil moisture retention in the model through a proposed transformation utilizi ng the vertical and temporal mean moisture content at any particul ar depth to water table. For a better understanding of the upper and lower soil zone behavior, one needs to examine a plot of the total soil moisture vs depth to water table for each station (example in Figure 4.5). The obvious tendency bands give us some inspirations: the mean and minimum curve may be used to present the tendency behavior of total soil moisture. Actual moisture content can be compared to the mean and minimum values to describe a quantitative relative moisture condition. 4.1.4 Formulation of Relative Moisture For the hypothesis proposed above, the appro ach is to fit the mean and minimum total soil moisture curves to the band of observations. From observation data plotted (previously shown in Figure 4.4) it can be seen that these six stations all have much variability in the actual soil moisture di stribution. A more effective expression for available soil moisture must be proposed. A van Genuchten type (van Genuchten, 1978) mathematical model is proposed for this purpose.
48 For the van Genuchten type m odel, the dimensionless water retention functioneS, is given by M N WT r s r ed S ) ) ( 1 1 ( ) /( ) ( (4.1) And, solving for ] ) ( ) ) ( 1 1 [(r r s M N WTd (4.2) Where is volumetric moisture content [L/L]; r is residual volumetric moisture content [L/L]; s is volumetric moisture content at saturation [L/L] ;WTd is the depth to water table [L]; is a dimensional parameter [1/L]; N and M are dimensionless curve fitting parameters and M = 1 Â– 1/ N ( N >1).This equation contai ns four independent parameters (s ,r and N ), which have to be estimated to represent the observed soilmoisture retention behavior. From the field site, soil moistu re profiles were monitore d continuously, averaged for the day for each level (i.e., 10, 20, 30 cm etc.) and integrated over the soil column to yield the total moisture cont ent in the top 150 cm (limit of the probe depth). For any given day and depth to water table value, ther e is total soil moisture value which varies over a limited range. All the data points are from field observations. Plotting the daily average, vertically integrated soil moisture vs depth to water table, the considerable variability displayed by each station was explor ed. First, the mean value for every range of depth to water table (i.e ., ranges like 0-0.5 cm, 0.5-1 cm, 1-1.5 cm Â…etc.) was calculated. The corresponding mean of total soil moisture fo r water table range is shown in the plotted mean_PS42 points in Figure 4.5. A similar approach was used to get the
49 minimum points (shown as min_PS42 in Figure 4.5). Next van Genuchten type mathematical functions were fitted to the mean and minimum values (shown as Fit_mean and Fit_min in Figure 4.5). The best fit equation was used to fi nd the relative moisture condition described below. In this approach, van Genuchten parameters were fitted to match the mean soil moisture behavior and anothe r set to the minimum soil moisture behavior for every station using Equation (4.2). Two examples are shown in Figure 4.5 for the fitted mean and minimum van Genuchten curves. An example of fitted mean and minimum curve to daily observation of upper zone total soil moisture for PS42 is shown in Figure 4.5a. Figure 4.5b shows another example of fitted mean and minimum curves to daily observations of lower zone total soil moisture for PS41. It should be noted by the reader that this is not a standard application of th e van-Genuchten moisture retention curves for Equation (4.2). Instead, the use was a matter of convenience (i.e., a reasonable mathematical relationship) to fit the observed behaviors to explore th e concept of relative moisture. Other mathematical models could have been used to fit the data as well but were not explored in this study. It is proposed that describing relative moisture condition based on the actual condition relative to the minimum and mean so il moisture behavior will better represent antecedent moisture condition, available Â“fr ee moistureÂ” for any water table depth, and reduce field-scale variability in moisture observations at multiple stations. Â“Free moistureÂ”, or free vadose zone storage, is here in defined as the variable moisture that can exist for any water table depth. Thus, the moisture condition could change from the maximum to the minimum content with negligible wate r table fluctuation. For more
50 details of the physics and mechanic s of this moisture variability the reader is directed to Shah and Ross (2006). The foundation for th e relative moisture condition is the Figure 4.5. Examples of Fitted Mean and Minimum Curves to Daily Observations of Total Soil Moisture vs. Depth to Water Table in (a) Upper Zone for PS42; (b) Lower Zone for PS41 following. At any stable water table depth th ere is limited minimum, mean (normal) and maximum total soil moisture. Thus, the availabl e vadose zone moisture at that water table elevation (for negligible water table change ) is the excess moisture above the minimum. Moisture conditions declining beyond this th reshold results in si gnificant water table decline (i.e., ET stress to the water table). Total soil moisture below the mean represents
51 relative dry antecedent condition and above th e mean, relative wet condition (for that corresponding water table depth). The quant itative relative moisture condition is therefore the actual content minus the mini mum divided by the mean minus the minimum for both zones, upper and lower (both also specific to that water table depth). The upper zone relative moisture condition, Equation (4.8) can be described as the ratio of the difference between the current total soil moisture, Equation (4.3), and the minimum total soil moisture, Equation (4.4 ), the definition of minimum total soil moisture, to the difference between the mean total soil moisture, Equation (4.5), the definition of mean total soil moisture, and th e minimum total soil moisture at the same depth to water table. In this way, each station, st rongly exhibiting field-scale variability, can be normalized to a more consistent quantita tive relative moisture condition. The model formulation can then be based on relative mo isture condition instead of actual moisture content. For example, during shallow water ta ble conditions with high moisture content, the soil moisture condition can vary from relati vely wet to dry very quickly and with very little volume change. The representative de finition equations for the upper zone are: UZdz d z dWT WT UZ0) ( ) ( (4.3) dz d z dUZWT WT UZ 0 min min ,, (4.4) dz d z dUZWT mean WT mean UZ 0 ,, (4.5) min UZ UZ UZSD (4.6) min , UZ mean UZ UZSND (4.7)
52 ) ( ) ( ) ( ) (min , min WT UZ WT mean UZ WT UZ WT UZ UZSN UZS UZd d d d D D (4.8) The denominator of Equation (4 .8) could be considered the Â“nominalÂ” upper zone storage,UZSND in the IHM documentation consistent with terminology used for the HSPF model (Bicknell et al ., 2001). The numerator of Equati on (4.8) then represents the available storage in the upper zone,UZSD Where UZSD = upper zone storage. [L]; UZSND = upper zone nominal storage. [L]; UZ = upper zone relative moistu re condition. [Dimensionless]; LZ= lower zone relative moisture condition. [Dimensionless]; WTd z = the actual volumetric moisture content of the soil, [L/L]; WT UZd = upper zone total soil moisture influenced by depth to water table,WTd [L]; WT mean UZd, = upper zone total soil moisture fr om the corresponding fitted mean curve for the given depth to water table,WTd [L]; WT UZdmin = upper zone total soil moisture corresponding to the fitted minimum curve for the given depth to water table, WTd [L]; UZ = the fixed thickness of the upper z one layer (e.g. soil A horizon). [L] Similar equations are applied to the lower zone as well, exception being that the lower zone comprises the soil zone below the A horizon down to the minimum of either the water table or the gr oundwater extinction depth, X The ground water extinction
53 depth is defined herein as the depth below which the vegetation can no longer effectively derive ET from the water table. This depth has been mathematically shown to be the sum of the soil capillary zone thickness,CZ plus the plant root zone thickness, RZ (Ross et al ., 2005a) as X =CZ +RZ (4.9) 4.2 Air Entrapment/pressurization 4.2.1 Background The role of air entrapment in inhibiting infiltration ha s long been recognized (e.g., Adrian and Franzini, 1966; Morel-S eytoux and Khanji, 1974; Vachaud et al ., 1974; Parlange and Hill, 1979). Several theoretical and experimental studies, e.g., Youngs and Peck (1964) and McWhorter (1971), have qua ntitatively defined the impact of air compression on infiltration. These studies found that, air compression ahead of a wetting front, in some water table conditio ns, brings about a sharp decrea se in the infiltration rate. However, as pointed out by Parlange and Hill (1979) and observed by Wang et al (1998), air compressibility has been generally found negligible, when the air is free to move ahead of the wetting front. Hence, the importance of air compression in an unconfined aquifer with deep water table is ge nerally considered negligible. However, for shallow water table environments (WTd <2 m) air compression plays a significant role in determining infiltration in many soils (Touma et al ., 1984). Because air entrapment in shallow water table environments reduces infiltration and causes artificial rise in the water table, it has significant implications for estimating and modeling ground water recharge. Heal y and Cook (2002) presented a thorough
54 review of methodologies to estimate rechar ge using groundwater levels, but commented that one of the major limitati ons of any method for shallo w unconfined aquifer was the Lisse effect. As the artificial rise in the water table is difficult to identify it can easily be mistaken for recharge (Healy and Cook, 2002). Accurate estimation of soil air pressure is thus of great importance for modeling runoff and water table recharge. Mathematical solutions derived fr om laboratory studies e.g., Wang et al ., (1997, 1998) provide very useful insight into the process of air entrapment, however the use of the labor atory derived equatio ns have not been adequately tested against field conditions. Latifi et al ., (1994) concluded that air pressure buildup was more pronounced in soil columns of two layers than in a monolithic soil. Zhang and Ross (2006) discuss the importance and prevalence of so il layering in most coastal plane soils. Na tural soil layering intr oduces uncertainty in the applicability of laboratory results, derived under homogenous soil conditions. Another important aspect to note is that most of the theoretical, experimental work or field observations have been limite d to an event based approach wherein the effects of single rainfall event on air pressurization/ water ta ble fluctuation are noted and analyzed any for only short dur ation. For the purpose of l ong term modeling of stream flow and aquifer recharge a continuous m onitoring and analysis is needed. For field conditions subjected to multiple events and varying antecedent conditions, air effects may become compounded and/or prolonged. Recently, Crosbie et al ., (2005) proposed a time series approach to infer gr ound water recharge using a wate r table fluctuation method. The approach tried to overcome the limita tions mentioned in H ealy and Cook (2002) and was reported to be applicable to long-ter m records of precipitation and water table
55 elevation. Even though the proposed model by Crosbie et al ., (2005) was innovative in its accounting for air pressurization, the model eliminated all wate r level rise, if the assumed criteria for Lisse effect (see Crosbie et al ., 2005, Equation 2) was satisfied. This may, during long continual rainfall events, neglect the actual water table rise due to wetting fronts reaching the water table. The above discussion clearly illustrates the need for a more physicallybased analysis of air entrapment over longterm (multi-event) records. The current study attempts to address this need by using shallow water table elevation records in conjunction with observed soil water content pr ofiles that were measured during a field study. The specific objective of the investigation was to: (1) detect the presence of Lisse effect, (2) quantify the air pre ssurization values in field da ta, and (3) use quantified air pressurization values to determine the locati on of true elevation of the water table. The approach used in the study was to ca librate a RichardsÂ’ equation model to the observed water content profile and derive dept h to water table from resultant pore water tension pressure, as it is unaff ected by air pressurization. The soil moisture behavior can then be used to determine the true depth to water table. The difference between the observed and the true depth to water table wi ll hence give the value of air pressurization (Shah et al ., 2006). Also a simple analysis based on ideal gas law was also done to help understand air pressurization effects. 4.2.2 Theoretical and Model Testing of Excess Pressure Due to air entrapment, traditional rainfall infiltration models such as Green and Ampt (1911), tend to over predict infiltrati on with physical soil parameters in shallow
56 water table environments. For the current study, infiltration can be derived directly from integrated volume changes since soil water content was explicitly measured. Assuming a one dimensional soil column, integration of the soil water content values gives the total water content (TWC) per unit area of soil column at any instant in time. Subtraction of two consecutive values will, hence, give an estimate of net infiltration or net ET (depending on the algebraic sign of the difference) in the units of length. For the purposes of this study, net infiltration or net ET refe rs to all inflow and outflow respectively (including lateral flows) for details of the approach one is directed to Rahgozar et al ., (2005). Nachabe et al ., (2005) and Rahgozar (2006) used a similar approach to determine ET and found the methodology to give a very good match with calcu lated values from other methods. For this particular study, given the spatial distribution of the soil moisture sensors, a simple numerical integration was done to calculate TWC fo r the soil column of length 1.5 m. The mathematical equation used is 8 1 i iz TWC (4.10) where zi [L] is the depth associated with each sensors, and i [L3L-3]is the water content values observed at th e corresponding sensor. 220.127.116.11 Numerical Model Soil water content profiles were mode led using a single-phase, one-dimensional RichardsÂ’ equation simulation model know n as HYDRUS -1D (version 3) (Simunek et
57 al ., 2005). Calibrated versions of this model have been used and verified in a number of studies (e.g., Hernandez et al ., 2003; Simunek and van Ge nuchten, 1999). Also, an independent team of hydrologists scrutini zed HYDRUS and found the model reliable and highly capable (Software Spotlig ht, 2000). The model uses the Galerkin type linear finite element method for space discretization and a finite difference method for temporal discretization of the Richar ds (1931) equation. This equation for a one dimensional vertical column can be written as: S z h K z t t z 1 ) ( ) ( ) ( (4.11) where h [L] is the water pressure head, [L3L-3] is the volumetric water content, t [T] is time, z [L] is the spatial coordinate (positive upwards), K [LT-1] is the unsaturated hydraulic conductivity, and S [L3L-3T-1] represents the sink term. HYDRUS was previously used by Hammecker et al (2003) to try and quantify the effect of air compression. The approach they used was to apply Dirichlet conditions, namely the upper boundary given by the ponding water le vel in the plot and the lower boundary given by the depth of the water table, as the two boundary conditions. The lack of match with the observed data was attributed to the air compression, as all the other processes were assumed to be accounted for in HYDRUS. No further analysis was done to quantify the air entrapment from the numerical solution. And limited comparisons to observed soil moisture profiles were made.
58 As described in Hillel (1998), due to ai r entrapment, the so il-water content does not attain total saturation but some maximal value lower than saturation, which he called satiation. Satiation can be taken into acc ount by considering that the maximum water content in a soil only reaches to a value smalle r than porosity, more commonly referred to as natural saturation or e ffective porosity (Charbeneau, 2000). Hence, laboratory determination of soil saturation water conten t normally overestimates the values found in the field. This phenomenon was considered in the calibration of soil parameters. For the current investigation, data fo r two months (May and June) in 2002 and another two months (April and May) in 2003 were analyzed, and modeled numerically using HYDRUS. This period of record wa s selected because it represented the transitional months when condi tions changed from very dry to very wet. Hence, a good contrast between the conditions with and with out air pressurization can be expected. Due to hysterisis, the effective porosity shows a long term s easonal behavior. Hence, for calibration purposes, saturated water content values that are used correspond to the maximum water content values observed duri ng the period of record. As expected, the values were found to be less than the labor atory determined porosity, by as much as 78%. Additional details and findings from this numerical model can be found in Shah et al ., (2006). 18.104.22.168 Calculation of Excess Pressurization using Ideal Gas Law The difference between the WTd obtained from theoretical solution (HYDRUS1D) and field observations, gives a quantitati ve estimate of air pressurization. If the pressure of the entrapped air is atmos pheric then the observed and the actual WTd will be
59 at the same location, void pressures above atmospheric pressure will cause well water levels to increase because the well is only scre ened below the water table. The pressure of the compressed air in excess of atmospheric, herein denoted as Â“excess pressureÂ”, is defined as the difference between the observed WTd and the HYDRUS-1D generatedWTd (elevation of zero tension). It is expre ssed in terms of depth of water column. In an attempt to quantify the amount of excess pressure and, potential thresholds for air eruption, a simple spreadsheet-ai r-excess-pressure-analy sis was set up. The maximum saturated water content for every se nsor from the entire period of data collection was found. To th is value 7.5 % (Nachabe et al ., 2004) was added to account for the residual air, crudely representing th e actual total soil porosity at each sensor. Multiplication of porosity by the depth associat ed with each sensor gives the available pore space in the soil column (per unit cross sectional area). Subtracting total soil water content obtained by in tegrating water content values along the soil prof ile (like in Equation 4.9) from the porosity gives the am ount of pores filled with air in the soil column. It is important to know the inherent assumptions involved in the spreadsheet calculation of excess pressure. The first and possibly most important assumption is that all the entrapped air present between the wetting front and the water table has the same pressure. This significant limitation will be discussed later. The second assumption is that continuous counter flow of air during an event is neglected prior to eruption. Therefore, the only way the soil air can leave the soil column is via air eruption. Finally, the temperature is assumed to be constant a nd the ideal gas behavior is assumed under adiabatic conditions.
60 Morel-Seytoux and Khanji (1975) pro posed a model for quantifying air compression using BoyleÂ’s law. As BoyleÂ’ s law assumes the mass of the gas to be constant, this methodology becomes invalid in case of air eruption. It is for this reason the ideal gas law is used for the spreadsheet analysis, with the unde rlying assumption that void air behaves like an ideal gas. Consis tent with the HYDRUS solution, hourly time steps were used for pressure calculations. T hus, hourly values of to tal soil water content were used to determine the changes in the volume from which the void air pressure is derived. Mathematically the ideal gas law can be defined as PV = nRT (4.12) where P is the absolute pressure (N/cm2), T is absolute temperature (K) assumed constant at 298K, V is volume of the void air (cm3), n is the number of moles, and R is the gas constant [= 831.41 N-cm / (mol/ K)]. As mentioned earlier, both the simu lation periods were preceded by dry conditions. Therefore, the ini tial pressure of the entrap ped air is assumed to be atmospheric, P0, i.e. 10.13 N/cm2. The initial volume V0 of entrapped air was determined by subtraction of observed total soil water cont ent (initial value) from the total pore space (constant =68.92 cm3) of the soil column. At the ne xt hour the new volume of air ( V1) is similarly calculated, using the corresponding observed total soil wate r content. Assuming a constant temperature T Equation 4.12, is used to determ ine the initial number of moles ( n0). Using n0 and the volume at the next hour V1 the pressure P1 was found again using Equation 4.12.
61 From this approach, excess pressure (expr essed as centimeters of an equivalent water column) is determined as follows: g P P P0 1 (4.13) Where P is the excess pressure (cm), is the density of water, and g is the acceleration due to gravity, and g is assumed as 0.00981 N/cm3. Between consecutive time steps two processes are possible. First, due to net ET, the new volume of air is greater than the previous volume or secondly, due to net infiltration, voids are reduced and excess pressure ensues. It is important to note that at an hourly time step sufficient infiltration can occur to cause the excess pressure to beco me quite large. Therefore, excess pressure may reach an upper limit where by rapid air eruption occurs. This breaking value, as defined in Wang et al (1997), results in eruption and a lowering of air pressure values. Consider the ET case where the volume of air increases. In this case the new value of air pressure will decrease, except th at there is no wetting front to preclude air uptake by the soil from the atmospheric bounda ry. As a result the pressure cannot significantly decrease below atmospheric. T hus, during the spreadsh eet analysis the new pressure value is made atmospheric if the solution of the Equation 4.12 results in sub atmospheric pressure during dr ying conditions. However, no ad justment is made if the new pressure comes out to be greater than at mospheric. One problem that remains is that the ideal gas law cannot be used to determ ine the air eruption thresholds. Also, as a consequence of air eruption, an undeterminable number of moles of ai r is lost. Hence, for the infiltration case, to incorporate air brea king value thresholds, pressures must be set through observation of the data to constrain the maximum pressure.
62 In the absence of any other indicator s, excess pressure determined from comparison of the HYDRUS solution with the field observation, was used to limit the excess pressure values calculated in the spre adsheet Air eruption was evident in several events in both periods, requiri ng constraining the maximum pr essure. Thus, if the excess pressure calculated from Equation 4.13 exceed ed the thresholds for air breaking derived by HYDRUS, the excess pressure was set at th e threshold and the numbers of moles lost were calculated using the ideal gas law As will be seen later in the results sec tion, the excess pressures calculated using HYDRUS show large variations depending on the infiltration magnitude and the antecedent conditions. However, critical thresholds were more consistent. This implies that, in order to determine air eruption for each event, different thresholds have to be set. To avoid this cumbersome approach, the an alysis was done only on the events occurring in the month of May of 2002 and 2003. 4.3 IHM Testing These last two sections investigated the methodology for soil zonation and air entrapment affects which are important fact ors for modeling vadose zone behavior. This section focuses on testing and model application. Limited previous work has been done to improve the concept basis of the IHM. Ross et al ., (2005a) made improvements to provide a smooth transition to satisfy ET de mand between the vadose zone and deeper saturated ground water. While the IHM appr oach provides a more sound representation of the actual soil profile th an original FIPR hydrologic model (FHM). Shah and Ross
63 (2006) explored the criteria a nd behavior of free vadose zo ne storage used in IHM and the physics and mechanics of this mois ture variability. Zhang and Ross (2006) differentiated upper and lower re gions of the unsaturated zone (vadose zone). Field soil moisture observations and soil characterizati on data were used to formulate a new basis for the upper zone and lower zone in IHM. And they developed a new methodology to describe relative moisture condition in both zones for modeling soil hydrologic response. Within the IHM, consistent with th e physical processes, the water table fluctuation, infiltration, and ET fluxes control vadose zone moisture response. With a very shallow groundwater table, the interaction between unsat urated and saturated zone becomes very strong. The groundwater table st rongly influences the water content in the unsaturated part of the root zone and the groundwater tabl e represents a moving boundary between saturated and unsaturated conditions. In testing the model behavior, questions are offered such as is the distribution of ET from interc eption, upper zone, lower zone and groundwater distribution reasonably simu lated; are the water table fluctuations comparable, and is the infiltration to recharge behavior adequate. Aly (2005) applied a preliminary version of IHM to a small basin in West-central Florida. But, more extensive investigation of theoretical basis of the simulated vertical processes and longer-term application were still needed. The behavior of the IHM is examined th rough comparisons with collected data at a study site in West-Central Flor ida. The objectives of this exploration we re to (1) test the model of the vertical processes controlli ng water table behavior, ET distribution and infiltration, (2) investigate the sensitivity of model parameters, and (3) offer recommendations for improvements and parame terization for regional model application.
64 Rigorous testing was done to better understa nd the robustness and/or limitations of the methodology of the IHM for upper and lower vadose zones. 4.3.1 Site Description The application site is shown in Figure 4.6. More information about the field study can be found in Trout and Ross (2005) and Ross et al ., (2005b). This watershed encompasses an area of 450 acres and is loca ted in a shallow surficial aquifer setting within a small catchment of Long Flat Creek, a tributary of the Al afia River in WestCentral Florida. The site is charact erized by a shallow water table (0
65 4.3.2 Model Setup For the surface water system, the waters hed was subdivided into six sub-basins. Each sub-basin was further divided into land segments based on land use categories (Figure 4.6). The surface hydrology of each land segment is simulated separately as HSPF PERLND (pervious land) units. Sub-basi ns 1-4 include one PERLND. Sub basin 5 includes a grass (landuse ID 2 in Figure 4.6) and a citrus land segment (landuse ID 3 in Figure 4.6) and sub-basin 6 in cludes a grass and a mined la nd segment (landuse ID 4 in Figure 4.6), that is, the surface hydrology of s ub-basins 5 and 6 are simulated with two PERLNDs for each sub-basin. The groundwater system was conceptual ized as a single-layer, unconfined surficial system with no-flow boundaries coincident with the topographic divides that form the boundaries of the surface basin (Aly, 2005). Greater than 12 m (40 ft) of head difference exists between the water table and the underlying confined aquifer. This head different has been sustained over the availa ble groundwater record (multiple decades). Therefore, the lower boundary of the single-la yer system is reasonably assumed to be a no-flow boundary. Then, recharge to the water ta ble is discharged laterally to the stream or is transported vertically to support ET demand. 4.3.3 Data Collection Field data were collected at five-minute intervals from September 2001 until June 2004 including: soil moisture at 10 cm depth in tervals, stream flow into and out of the basin, precipitation, potential ET, runoff rate s from a controlled plot and complete
66 meteorological conditions. Also included were daily water table heads for all observation wells. 1 6 5 2 3 4 PS43 PS42 PS41 PS40 P-S-1 P-S-2 USF-3 USF-4 P-S-9 P-S-5 P-S-4 P-S-69 P-S-68 P-S-65 P-S-64 P-S-45 P-S-44 P-S-23 P-DI-63 P-SI-63 Well Stations Landsegments (Basin,Landuse) 1, 2 2, 2 3, 2 4, 2 5, 2 5, 3 6, 2 6, 4 0110220330440 55 Meters Figure 4.6. Sub-Basins, Landsegments and Observation Wells at the Long Flat Creek Study Site 22.214.171.124 Basin Landuse The 1999 land-use map was obtained from the SWFWMD online GIS database (SWFWMD online resources). The land use a nd land cover features were categorized according to the Florida Land Use and Cover Forms Classification System (FLUCCS). Sub Basins Reach
67 Each FLUCCS code was assigned to one of fi ve general land-use categories (PERLNDs): grass or pasture, forested, irrigated agri culture, urban, and mined or disturbed. 126.96.36.199 Soil Moisture Capacitance shift type (Sentek Model Enviro SMART) soil moisture probes were installed adjacent to monitoring wells PS43, PS42, PS41, PS40, USF1 and USF3 with manufacturer reported accurac ies to 0.1%, and laborator y verified to 0.05% to gravimetric moisture content using specific calibration curves. Vertical resolution was achieved with sensor placement at 10, 20, 30, 50, 70, 90, 110 and 150 cm below land surface. The moisture probes from PS43 to PS40 were along a downhill transect from the upland grassed area at PS43 to the riparian forest near the stream at PS40. 188.8.131.52 Water Table The study location was instrumented with several water table observation wells. Vented water table observation wells hous ed submersible pressure transducers (Northwest Inc.). The transducers were calib rated to measure pressure from 0-34 KPa (5 psi) with an accuracy of 0.034 KPa (0.005 psi). 184.108.40.206 Rainfall Rainfall data were collected from January 2002 to July, 2004 with tipping bucket rain gauges, first laboratory cal ibrated and continually verified with manual (NWS type) rain gages. Five minute precipitation records were collected through the period with only minimal data gaps.
68 220.127.116.11 Stream Flow Stream gages were installed near upstr eam, mid-stream and downstream of the basin along Long Flat Creek us ing installed multi-section calibrated weirs. Five minute flow records were obtained from upstream a nd downstream gages as inflow and outflow of the watershed with considerable data gaps due to frequent weir failures. 18.104.22.168 Potential Evapotranspiration Potential ET ( PET ) was estimated from several sources including onsite using open pan evaporation measurements multiplied by a constant pan coefficient (0.7) and/or onsite or offsite meteorological data. More re liable PET estimates (based on comparison to open water evaporation rates) were cal culated based on the empirical equation of Jensen and Haise, J & H (1963) using te mperature and solar radiation records. ) 08 0 ) 025 0 (( 2450 &ave S H JT R ETP (4.14) The input parameters for the J & H equati on were: instantaneous solar radiation, R [Wh/m2 per hour] and daily average temperature, aveT [ C]. Solar radiation and temperature data were obtained from the onsite measurements and were supplemented with Florida Automated Weather Network (FAWN) ONA (URC) data. The FAWN ONA site was selected due to the close proximity to the research site. Details of data collection can be found in Rahgozar et al (2005), Trout et al (2005) and Ross et al ., (2006).
69 4.3.4 Model Calibration Model calibration was used to establish th e most suitable values for several key model parameters. The objectiv e of calibration was to comp are model performance to observations and to test the robustness of the model conceptual framework. Calibration was carried out for two distinct land cover types, grassed an d forested land cover in the study area for the years January 1, 2002 through June 30, 2004. Parameters found by calibration included soil infiltration index ( INFILT ) and upper zone (depression) storage capacity ( UZSN ). Other parameters were found though soil analyses or published characterization data (Carlisle et al 1989) pertinent calibratio n data are summarized in Table 4.1. The two land cover types differe d greatly in ET, resistance to surface runoff and interception (i.e., vegetative parameters differed greatly), however soil properties were similar. Thus, very different plant coe fficient, root zones, upper zone storage and interception parameters were used for these two land cover types (Fi gure 4.7a and b). In Figure 4.8a, the modified plant coefficient for grassed and forested land cover was only adjusted slightly from values derived by independent measurement of Rahgozar et al ., (2005). This analysis focused on the comparison of flux rates and cumulative fluxes of ET components: interception ET (ICET), uppe r zone ET (UZET), lower zone ET (LZET), groundwater ET (GWET) and total ET (TAET) Also, infiltration and depth to water table (DTWT) were compared to observations.
70 Table 4.1. Derived Calibration Parameters for Forested and Grassed Land Cover Calibration Value Parameter Forest Grass INFILT (cm/hr) 4.32 2.29 UZSN(cm) 0.05 4.06 Saturation 0.37 0.34 Field Capacity (cm/cm) 0.13 0.16 Root Zone Thickness (cm) 100 50 Capillary Fringe Thic kness (cm) 30.48 30.48 Capillary Zone Thickness (cm) 100 100 Wilting Point 0.05 0.05 Hydraulic Conductivit y (cm/day) 30.48 30.48 0.0 0.2 0.4 0.6 0.8 1.0 1.2 123456789101112 MonthPlant Coefifien t FOREST Modified Forest GRASS Modified Grass 0.00 0.10 0.20 0.30 0.40 123456789101112 MonthInterception Storage (cm) FOREST GRASS Figure 4.7. Calibration Values Used for (a) Plant Coefficient and (b) Interception Storage for Grassed and Forested Land Cover a) b)
71 So that appropriate processes were co mpared, the observed upper zone ET and lower zone ET from Rahgozar et al ., (2005), were plotted agai nst lower zone ET adjusted to LZET plus GWET when the depth to water table was deeper than 30 cm (1 ft). Similarly, for the upper zone ET, UZET was added to GWET when the depth to water table was less than 30cm (1 ft) (corres ponding to capillary fringe at land surface).
72 CHAPTER 5 RESULTS AND DISCUSSION This chapter presents the results and discussion of soil zonation analysis, air entrapment analysis and IHM testing. 5.1 Soil Zonation In West-Central Florida, a cool, dry a nd low ET winter season and a warm, rainy summer season constitute a typical annual clima tic cycle. The seven to eight months of the dry season (usually from October to Ma rch) show rapidly changing but generally milder average temperatures and lower monthly rainfall. Rainfall in the winter is usually associated with frontal passage and with the in frequent low pressure systems that form in the Gulf of Mexico close to the Florida coast (Myers and Ewel, 1991). Winter in Florida is typically dry when there is not an El Ni o effect in the Pacific. The mean monthly rainfall varies from 70 to 110 mm in North Florida, 50 to 90 mm in Central Florida, and 40 to 50 mm in South Florida (MacVicar, 1981) during this period, contrasting values two or three times that in the p eak of the summer rainy season. In general, the period March to April, in West-Central Florida, represents the driest soil moisture conditions with low rainfall and high springtime ET stress. This
73 combination results in the deepest water tabl e conditions. While from June to September, West-Central Florida experiences high ra infall, high ET, and shallow but rapidly fluctuating water table. Re ferring to Figure 5.1, the broad summer rainfall peak from June through September was slightly wetter th an normal for the period. Typically, rainfall accounts for 50-60 percent of the annual total during the 3-month period. 5.1.1 Moisture Conditions for the High ET Period For the high ET period analys is, the four months that make up the wet season (June through September, 2003) were chosen this period usually s hows relatively uniform high temperatures, high solar ra diation, and high monthly ra infall. In 2003, the period from April to Aug (4/1/2003-8/31/2003), an unusually high rainfall accumulation of 96.5 cm was observed at the site. Expected fo r high ET and frequent rainfall during this period, total soil moisture showed much tem poral variability in both the upper and lower zone (Figure 5.1b, g, d and i) for both fore sted and grassed cover; while the resultant relative moisture, UZ and LZ (Figure 5.1c, h, e and j) fo r forested and grassed cover clearly showed more uniform response (little variability between similar stations) and thus reducing field-scale va riability. Relative moisture also showed strong dynamic fluctuation early in this peri od. Interestingly and somewhat expected, relative moisture fluctuation was greatly dampened when water table was at land surface. 5.1.2 Moisture Conditions for the Low ET Period Examining a low ET period (1 1/1/2003-12/31/2003), with an associated observed rainfall sum of 70 mm, the total soil moisture and corresponding upper zone
74 and lower zone relative moisture behavior we re quite different. During the November to December period, a characteristic low ET rate and low rainfall accumulation followed a prolonged period of above average rainfall 30.48 cm (12 in) in the proceeding 15 months; yet, the water table became deeper during this period. Referring to Figure 5.2, comparing the distinct vegetative cover groups, the relative moisture in the upper and lower zone exhibited similar behavior within a group, but distinctly different behavior between groups (Figure 5.2c, h, e and j). It was also observed that the water table (Figure 5.2a, f) and the total soil moisture (Figure 5.2b, g, d and i) are on decline while relative moisture both in the upper and lowe r zone, showed variability and periodic increases. Again, similar to what was observed in the high ET period, relative moisture shows differences in upper and lower zone. 5.1.3 Statistical Summary and Discussion Summaries of statistical anal ysis are provided in Table 5.1 and Table 5.2 to compare the relative moisture condition and actual total soil moisture mean and standard deviation for the different vegetative cover groups. In Table 5.1, both high ET (a) and low ET (b) periods are compared. Relative moisture in both soil zones, UZ and LZ show consistency with values around unity (1) for all wells (this is also consistent with obs erved rainfall being close to seasonal average most periods). Given the high wa ter table fluctuation characteristic of this period, the mean of the total soil moisture shows much more variability (coefficient of variation, CVÂ’s 6-28% shown in Table 5.2). But for both upper and lower zone, the variability of
75 actual soil moisture exhibited from one site to another is very large while relative moisture condition, UZ and LZ is more consistent. It was also observed that depth to Figure 5.1. Upper and Lower Zone Total Soil Moisture vs. Relative Moisture for Representative High ET Period (Apr. Aug. 2003) for Forested (a-e) and Grassed (f-j) Cover
76 Figure 5.2. Upper and Lower Zone Total Soil Moistu re vs. Relative Moisture for Representative Low ET Period (Nov. Dec. 2003) for Forested (a-e) and Grassed (f-j) Cover
77 Table 5.1. Statistical Results for Relative Moisture Condition, UZ and LZ and Total Soil Moisture, UZ and LZ for both Upper and Lower Unsaturated Zone by Landuse Group in Selected a) High ET (4/1/2003-8/31/2003) and b) Low ET (11/1/2003-12/31/2003) Periods Grassed Land Cover Forested Land Cover High ET Period USF3 USF1 PS43 PS40 PS41 PS42 UZ 0.990 0.983 1.016 1.003 1.017 1.005 LZ 0.999 0.999 0.999 1.009 1.016 1.009 UZ (cm) 3.08 2.95 3.18 1.85 2.48 2.59 LZ (cm) 41.22 43.47 40.61 36.09 39.96 40.83 Depth to Water Table (m) 0.360 0.311 0.376 0.994 0.494 0.550 Grassed Land Cover Forested Land Cover Low ET Period USF3 USF1 PS43 PS40 PS41 PS42 UZ 0.994 1.018 0.995 1.007 0.995 1.005 LZ 1.004 0.994 1.008 0.984 0.967 1.002 UZ (cm) 2.18 2.23 1.90 0.84 1.02 1.52 LZ (cm) 39.44 40.71 36.97 27.53 32.48 33.83 Depth to Water Table (m) 0.744 0.629 0.950 1.340 0.993 1.056 water table in grassed cover averaged sha llower for both periods than forested cover considering both the observed behavior (Figs. 5.1 and 5.2) and the summary statistics (Table 5.1 and 5.2), even though the forested cover is considered a Â“wetlandÂ” and the grassed cover was further up the hillslope and considered and Â“uplandÂ” community. Relative moisture was also much more dyna mic than total moisture or moisture concentrations reflecting variability in ant ecedent condition. But, relative moisture was much more consistent between stations indica ting that the stations were all in similar antecedent condition at any time. a) b)
78 Table 5.2. Statistical Results for Relative Moisture Condition, UZ and LZ and Total Soil Moisture, UZ and LZ in Upper and Lower Unsaturated Zone by La nduse Groups for All Data Periods (1/1/20026/27/2004), (a) Forested Cover; (b) Grassed Cover Comparison Forested Land Cover (cm) (cm) Depth to Water Table (m) PS40 0.999 0.999 1.53 32.16 1.141 PS41 1.001 0.998 2.39 37.45 0.735 Individual Station Means PS42 1.001 0.998 2.36 37.46 0.727 Mean 1.008 1.015 1.99 35.72 Standard Deviation 0.055 0.057 0.51 3.41 Mean Daily Comparison* Coefficient of Variation (%) 5.7 4.9 28.2 9.7 Comparison Grassed Land Cover (cm) (cm) Depth to Water Table (m) USF3 1.004 1.022 2.8 39.59 0.548 USF1 0.995 1.02 2.79 42.76 0.51 Individual Station Means PS43 1.000 1.009 2.76 39.02 0.625 Mean 1.026 1.022 2.84 40.43 Standard Deviation 0.0598 0.044 0.226 2.598 Mean Daily Comparison* Coefficient of Variation (%) 6.4 4.2 11.3 6.6 statistically summaries are means of the daily comp arisons with respect to average daily moisture conditions for stations with similar vegetative cover. One final point is that the met hod for expressing relative moisture presented herein does not dampen the overall description of an tecedent moisture condition. In fact, it accentuates the quantification of relative moisture state: wet or dry. An analysis of the relative moisture, overall coefficient of variability (CVÂ’=overall standard deviation/overall mean), show ed that for the upper zone CV was 50% and the lower zone CV was 40% compared to CV for the total moisture content values which was 30% and 20%, respectively. Further discussion of th is significance can be found in Zhang and Ross (2005). UZ LZ UZ LZ UZ LZ UZ LZ a) b )
79 5.2 Air Entrapment Figure 5.3 a and b shows the variation of excess pressure calculated from spreadsheet analysis of void air pressure s using the ideal gas law along with the HYDRUS solution, and the observed WTd .The number and variation of air moles are also included in the figure to dem onstrate air eruption. A review of Figure 5.3 (a) and (b) shows that rate of pressure decline calcu lated from the spreadsheet was significantly more than the decline calculated from HYDRUS. The results from the spreadsheet analysis raise a big questi on: what is going on with air pressure in shallow WTd environment and why are the air excess pr essure periods so prolonged? Another observation might be that RichardsÂ’ equation solution ma y not represent WTd and infiltration behavior well enough in shallow water table settings to reasonably quantify runoff (Hortonian or saturation excess) and rech arge processes. In an attempt to answer this question and investigat e the profound observation, basic processes in porous gas behavior need further exploration. RichardsÂ’ equation as solved by HYDRUS ignores void air pressurization. Hence for all boundary conditions and soil moisture variation it solves for WTd, as the elevation of atmospheric moisture pressure (zero su ction). The spreadsheet solution discussed on the other hand is highly dependent on the so il air volume changes from which (uniform) excess pressure is calculated. The spreadshee t solution did not take any soil property, or variability in air pressurizati on into account. The onl y driving variable in the spreadsheet solution was the change in void air volume, whic h is inherently assumed to be occurring
80 Figure 5.3. Excess Pressure as Calculated from a Spreadsh eet Solution of the Ideal Gas Law and a HYDRUS Solution. Figures Show Pressure Variation for a) May 2003 b) May 2002 Excess Pressure ( cm ) Observed dW T ( cm ) 30 40 50 60 70 80 90 100 110 1205/ 1 8/ 0 3 5/19/03 5 / 20 / 03 5/ 2 1/ 0 3 5 / 22 / 03 5 / 2 3/ 0 3 5/ 2 4/03 5/25/03 5 / 26 / 0 3 5/ 2 7/03 5/28/03 5 / 29 / 030 10 20 30 40 50 60 70 80 90 Observed dWTExcess Pressure Derived From HYDRUS Spreadsheet Solution A 30 40 50 60 70 80 90 100 110 1205/ 1 8/ 0 3 5/19/03 5 / 20 / 03 5/ 2 1/ 0 3 5 / 22 / 03 5 / 2 3/ 0 3 5/ 2 4/03 5/25/03 5 / 26 / 0 3 5/ 2 7/03 5/28/03 5 / 29 / 030 10 20 30 40 50 60 70 80 90 Observed dWTExcess Pressure Derived From HYDRUS Spreadsheet Solution AExcess Pressure ( cm ) 90 100 110 120 130 140 1505/18/02 5/19/0 2 5 /2 0/02 5/21/02 5 /2 2 /0 2 5 /2 3/02 5/2 4 /0 2 5 /2 5 /0 2 5/26/02 5/2 7 /0 2 5 /2 8/02 5/29/02 5 /3 0 /0 20 10 20 30 40 50 60 Observed dWTExcess Pressure Derived From HYDRUS Spreadsheet Solution BObserved dW T (cm) a) b)
81 between the wetting front and the water ta ble and was always assumed uniformly distributed. Further analysis of the physical process responsi ble for this curious behavior is strongly warranted. It would strongly warrant pore pressure and soil tension measurements in addition to water c ontent and water elevation measurement. 5.3 IHM Testing 5.3.1 Sensitivity Analysis Parameter sensitivity analysis yields an indication of the importance of a parameter on the model result. Small changes in the values of highly sensitive parameters produce large changes in model predictions and, conversely, large changes in insensitive parameters have little eff ect on the model results (Said et al 2006; Doherty, 2001a). The following parameters were tested in the IHM field-sc ale application: saturation (SA), field capacity (FC) upper zone nominal storage ( UZSN ), root zone thickness (RZ), capillary fringe thickness (C F), index of mean soil infiltration rate ( INFILT ) and plant coefficient (PC). The result s of sensitivity analysis for process responses of: total ET, upper zone ET, lowe r zone ET, groundwater ET, recharge, runoff, infiltration and depth to water table, are list ed for each of the estimated parameters in Table 5.3. Parameters were varied within reasonable physical ranges, resultant model response was compared to the original calibra tion response and results presented in Table 5.3 for grassed land cover and Table 5.4 for fo rested land cover. Table 5.3 a) lists the respective parameter setting and process relativ e change (%) and b) averaged sensitivity (process change / % parameter change for th e range) for grassed land cover. Similar results are shown in Table 5.4a) and b) for forested land cover.
82 Of special interest for ET was the sensitivity results for seasonal plant coefficient variability carried out using three shape f unctions designated as: PC1, PC2 and PC3 for forested land cover. Forested land cover wa s used because it has a deeper root zone, therefore, it was believed that the plant coefficient would be considered to have more effect on lower zone and water table ET. Di stribution PC2 is constant representing the averaged 12 month value from PC1 (f ound through measurements by Rahgozar et al ., (2005) which is used in Figure 4.8 (a) for fo rested land cover. PC3 is adjusted to the values and distributi on reported by Aly (2005). 5.3.2 Results and Statistics Analysis The cumulative ET flux (total ET, lower zone ET, upper zone ET and interception ET) comparisons to observations are shown in Figure 5.4 (a) for grassed and (b) for forested land cover. Model calibration statisti cs are presented in Table 5.5 for daily time scale and two different land covers: (a) gra ssed and (b) forested. Widely used error statistics are reported including: mean error (ME), root mean square error (RMSE) and mean absolute error (MAE). Scrutiny of e rror results indicates that model simulated results for ET fluxes compared reasonably we ll for all processes. For example, daily UZET for grassed land cover was 0.002 cm, 0.102 cm and 0.064 cm for ME, RMSE and MAE respectively and 0.002 cm, 0.065 cm and 0.028 cm for forested land cover for the calibration period. In addition to comparing cumulative ET flux, the daily (temporally variable) ET flux performance is shown in Figure 5.5, 5. 6 and 5.7 for total ET, lower zone ET and
83 Table 5.3. Model Sensitivity Analysis from Calibration Parameters for Grassed Land Cover Parameters Settings Parameter Relative Process Relative Change (%) Tests Saturation Field Capacity UZSN (cm) CF (cm) Root Zone (cm) Change (%) TAET UZET LZET GWET Re charge Runoff Infiltration DTWT 1 0.35 0.15 0.38 30.48 50 Basis run This is the basis run for comparison 2 0.3 0.15 0.38 30.48 50 -14 -0.9 -0.4 -1.3 -1.1 -6.3 1.9 -0.5 1.6 3 0.47 0.15 0.38 30.48 50 34 2.2 -4.2 5.2 3.2 12 -4.5 2.4 -3.6 4 0.35 0.1 0.38 30.48 50 -33 -0.8 2.5 -2.3 5 19.7 2.9 -3.7 -1.4 5 0.35 0.2 0.38 30.48 50 33 0.1 -0.9 0.5 -3.5 -23.4 -1.6 3.2 2.5 6 0.35 0.15 0.18 30.48 50 -53 -2.5 -35.3 10.3 12.1 11.7 6.4 7.5 2.9 7 0.35 0.15 0.56 30.48 50 47 1.6 19 -5 -6.7 -8.3 -4.3 -3.7 -2.1 8 0.35 0.15 0.38 15.24 50 -50 0.9 21.8 -7.4 23.8 35.3 -10.5 23.7 -25.8 9 0.35 0.15 0.38 45.72 50 50 -0.3 -3.7 1 3.4 6 2.9 -2.2 12.7 10 0.35 0.15 0.38 30.48 25 -50 -4.6 7 -10.4 -6 -5.4 9.9 -8.5 -11.7 11 0.35 0.15 0.38 30.48 100 100 -1.2 1.6 -2.6 -1.8 -1.8 2.4 -2 -2.7 Average Sensitivity Parameter TAET UZET LZET GWET Rec harge Runoff Infiltration DTWT Saturation 0.06 -0.05 0.12 0.08 0.39 -0.13 0.05 -0.11 Field Capacity 0.01 -0.05 0.04 -0.13 -0.65 -0.07 0.10 0.06 UZSN 0.04 0.53 -0.15 -0.19 -0.20 -0.11 -0.11 -0.05 Capillary Fringe -0.01 -0.23 0.08 -0.23 -0.33 0.11 -0.24 0.30 Root Zone 0.01 -0.02 0.03 0.01 0.01 -0.03 0.02 0.03 a ) b ) 83
84 Table 5.4. Model Sensitivity Analysis from Calibration Parameters for Forested Land Cover Parameters Settings Process Relative Change (%) Tests Plant Coefficient INFILT (cm/hr) Root Zone (cm) Parameter Relative Change (%) TAET UZET LZET GWET Rech arge Runoff Infiltration DTWT F1 PC1 2.54 100 Basis run This is the basis run for comparison F2 PC2 2.54 100 See Fig 5.11 -0.14 0.32 -0.28 -4.09 -2.82 0.25 -0.5 -0.23 F3 PC3 2.54 100 See Fig 5.11 -0.39 68.76 16.31 27.77 1.8 1.48 15.57 -0.86 4 PC1 1.27 100 -50 -0.27 5.05 -1.63 -6.32 -9.22 5.57 -8.08 7 5 PC1 3.81 100 50 0.06 -2.87 0.79 4.64 3.83 -2.28 3.93 -3.32 6 PC1 2.54 50 -50 -4.21 6.89 -7.82 -5.09 -4.4 9.73 -6.03 -10.82 7 PC1 2.54 200 100 2.28 -3 4.05 4.45 3.87 -4.91 3.12 5.89 Average Sensitivity Parameter TAET UZET LZET GWET Rec harge Runoff Infiltration DTWT PC1 PC2 PC3 See Figure 5.11 INFILT 0.00 -0.08 0.02 0.11 0.13 -0.08 0.12 -0.10 Root Zone 0.06 -0.10 0.12 0.10 0.08 -0.15 0.09 0.17 a ) b) 84
85 upper zone ET respectively. In Figure 5.5, the simulated TAET pattern (Figure 5.5 a, b) is generally similar to observed values with a good-to fair 2 R =0.59 (Figure 5.5c) for grassed and 0.57 for forested (Figure 5.5d). Pa tterns also matched for lower zone ET for both land cover (Figure 5.6) with fair 2 R values of 0.55 and 0.52, respectively. From Figure 5.7, however, it seems the model al ways over-predicted the upper zone ET, especially during the months of Ju ly and August which is indicated by 2 R values lower than 0.4. One possible reason may be the upper zone in the model con ceptually represents only depression storage (surface based ET) and the observed UZET represented an estimated for depression storage ET based on PET measurement only when water table is very close to land surface, pr imarily in the wet period (Jul y and August). As pointed out in Rahgozar et al ., (2005), the measurement method of Rahgozar (2006) does not provide direct measurement of PET for water table elevations at or near land surface. Table 5.5. Model Daily Performance Statistics for (a) Grassed and (b) Forested Land Cover Statistics on Daily Values for All Periods Model Performance (Grassed Land Cover) Mean Error (cm) Root Square Mean Error (cm) Mean Absolute Error (cm) UZET 0.002 0.102 0.064 LZET -0.002 0.085 0.064 TAET 0.107 0.161 0.122 ICET -0.001 0.053 0.026 INFILTRATION -0.022 0.350 0.083 DTWT -1.194 1.949 1.537 Statistics on Daily Values for All Periods Model Performance (Forested Land Cover) Mean Error (cm) Root Square Mean Error (cm) Mean Absolute Error (cm) UZET 0.002 0.065 0.028 LZET 0.002 0.108 0.081 TAET 0.004 0.146 0.089 ICET 0.000 0.062 0.030 INFILTRATION -0.039 0.360 0.104 DTWT 0.003 1.817 1.348 a) b)
86 ` 0 100 200 300 1/1/027/20/022/5/038/24/033/11/04ET Flux (cm ) Observed TAET Model TAET Observed LZET Model LZET Observed UZET Model UZET Observed ICET Model ICET PET 0 100 200 300 1/1/027/20/022/5/038/24/033/11/04 ET Flux (cm ) Observed TAET Model TAET Observed LZET Model LZET Observed UZET Model UZET Observed ICET Model ICET PET Figure 5.4. Calibration Results for Cumulative ET Fluxes for (a) Grassed and (b) Forested Land Cover a) b)
87 0.0 0.2 0.4 0.6 0.8 1.0 1/1/027/20/022/5/038/24/033/11/04Daily TAET (cm ) Observed Modeled R2 = 0.59050.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 Observed (cm)Modeled (cm) 0.0 0.2 0.4 0.6 0.8 1.0 1/1/027/20/022/5/038/24/033/11/04Daily TAET (cm ) Observed Modeled R2 = 0.570.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 Observed (cm)Modeled (cm) Figure 5.5. Calibration Results for Daily Total ET Flux fo r (a) & (c) Grassed and (b) & (d) Forested Land Cover a) c) d) b) 87
88 0.0 0.2 0.4 0.6 0.8 1.0 1/1/027/20/022/5/038/24/033/11/04Daily LZET (cm ) Observed Modeled R2 = 0.55470.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 Observed (cm)Modeled (cm) 0.0 0.2 0.4 0.6 0.8 1.0 1/1/027/20/022/5/038/24/033/11/04Daily LZET (cm ) Observed Modeled R2 = 0.52380.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 Observed (cm)Modeled (cm) Figure 5.6. Calibration Results for Daily Lower Zone ET Flux for (a) & (c) Grassed and (b) & (d) Forested Land Cover a) c) d) b) 88
89 0.0 0.2 0.4 0.6 0.8 1.0 1/1/027/20/022/5/038/24/033/11/04Daily UZET (cm ) Observed Modeled R2 = 0.3340.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 Observed (cm)Modeled (cm) 0.0 0.2 0.4 0.6 0.8 1.0 1/1/027/20/022/5/038/24/033/11/04Daily UZET (cm ) Observed Modeled R2 = 0.25420.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 Observed (cm)Modeled (cm) Figure 5.7. Calibration Results for Daily Upper Zone ET Flux for (a) & (c) Grassed and (b) & (d) Forested Land Cover a) c) d) b) 89
90 0.0 2.0 4.0 6.0 8.0 10.0 1/1/027/20/022/5/038/24/033/11/04Daily INFILTRATION (c m Observed Modeled R2 = 0.73680.0 2.0 4.0 6.0 8.0 0.02.04.06.08.0 Observed (cm)Modeled (cm) 0.0 2.0 4.0 6.0 8.0 10.0 1/1/027/20/022/5/038/24/033/11/04Daily INFILTRATION (c m Observed Modeled R2 = 0.77060.0 2.0 4.0 6.0 8.0 0.02.04.06.08.0 Observed (cm)Modeled (cm) Figure 5.8. Calibration Results for Daily Infiltration fo r (a) & (c) Grassed and (b) & (d) Forested Land Covera) c) d) b) 90
91 0 6 1/1/027/20/022/5/038/24/033/11/04DTWT(cm)0 3Rainfall (cm ) Observed DTWT Model DTWT Rainfall 60 120 180 7.5 0 2 4 6 1/1/027/20/022/5/038/24/033/11/04DTWT(cm)0 3Rainfall (cm ) Observed DTWT Model DTWT Rainfall 60 120 180 7.5 Figure 5.9. Calibration Results for Depth to Water Table for (a) Grassed and (b) Forested Land Cover a) b)
92 Comparing Figure 5.4 to Figures 5.5-5. 7, the cumulative ET (total and all ET components in Figure 5.4) compares much bette r than the daily totals (Figures 5.5-5.7). The possible reason why the 2 R for daily rates are not very high even though the cumulative plots compare well may be from small differences in time scale between the observed data, derived by integrated soil mois ture observations and the conceptualization of ET response in the model. A time scale exam ple is shown in Figure 5.10, compared to Figure 5.5(d) with 2 R =0.57, it is dramatically increased to 0.73 by plotting modeled and observed values using 3-day central averagi ng ( Figure 5.10 a). Inte resting only modest improvement is achieved with 5-day m oving averages (Figure 5.10 b). For both observations and model behavior, daily comp arisons were derive d from midnight to midnight totals. It should be noted that, Rahgozar (2006) reported that soil moisture observations exhibit an inherent delay in mois ture flux observations of several hours, also shown in Rahgozar et al ., (2005). Infiltration values were also compared against observations as a de-facto test on model rainfall excess (runoff) prediction. Figure 5.8 shows that the model results compared to observations of daily inf iltration volumes during the period with 2 R = 0.74 for grassed and 0.78 for forested land cover. One interesting note from the daily scatter plot in Figure 5.8a and b is that, there were periodically signifi cantly Â“over-predictedÂ” infiltration volumes during April to July, 2002 that occurred for both of these land covers. Further investigation of this observation will be discussed later. Another observation was that the forested land cover with deep rooted vegetation and higher ET stress, exhibited profound differe nce in other fluxes (e.g., infiltration and runoff) when compared to grassed land cover even though the soils where similar. This is
93 attributed to higher ET dema nd whereby water table elevations average lower. The effects on the water table are illustrated in Fi gure 5.9 (a) for grassed and (b) for forested land cover. It can be noted that, contrary to typical hillslope mode ls, the lower forested area exhibited frequently deeper water table than the upper grassed domain owing principally, it is believed, to the high ET demand of the fore sted land cover. It is noted that the model reasonably simulates fluctua tion of depth to water table during deeper periods and more poorly during near-surface water table conditions. It was found that during July to September, 2002 at the site the water table was above the land surface about 75% of the time (Aly, 2005). From Figur e 5.9, the simulated depth to water table for this period was consistently below the obs erved values. Aly (2005) also observed this behavior during earlier testing of the IHM. Reasons for the poorer performance for nearsurface water table are unclear. 5.3.3 Discussion The results described above for the da ily observed and simulated component ET fluxes, depth to water table and infiltrati on indicated reasonably good calibration for the period. Most of the ET fluxes were reproduced by the IHM model eith er temporal (daily scale) or cumulative with small mean and absolute errors. The calibration sensitivity analysis suggests several issu es warrant further discussion.
94 y = 0.956x R2= 0.73490.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 3-Day Average Observed (cm)3-Day Average Modeled (cm) y = 0.956x R2= 0.73490.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 3-Day Average Observed (cm)3-Day Average Modeled (cm) y = 0.9626x R2= 0.79220.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 5-Day Average Observed (cm)5-Day AverageModeled (cm) y = 0.9626x R2= 0.79220.0 0.2 0.4 0.6 0.8 0.00.20.40.60.8 5-Day Average Observed (cm)5-Day AverageModeled (cm) Figure 5.10. Time Scale Analysis on Daily TAET for Forested Land Cover (a) 3-Day Average (b) 5-Day Average a) b)
95 22.214.171.124 Calibration Parameters 126.96.36.199.1 UZSN and UZET From the sensitivity analysis in Tabl e 5.3 b, it was found that UZET is, not surprisingly, very sensitive to UZSN (sensitivity = 0.53). Increasing UZSN value increases the amount of water retained in the upper zone and available for ET, and thereby decreases the dynamic behavior of the surface and reduces direct overland flow. Zhang and Ross (2006) suggested that the upp er zone thickness should be approximated as the soil A horizon (e.g., top 15 cm of upper so il), but during calibration it was realized that the present algori thms for upper zone in IHM (based purely on the c onceptualization of HSPF) can only be considered as the depres sion storage. Therefore, in the present IHM model, there is no differentiation of the uppe r and lower vadose zone practically. The typical values of UZSN commonly used in HSPF range from 0.13 cm (0.05 in) to 12.7 cm (2.0 in) (EPA Technical Notes) which is broad, and at this po int, poorly defined from site physical conditions. LaRoche et al ., (1996) reported values ra nging from 0.04 cm (0.016 in) to 1.9 cm (0.75 in). The UZSN values found through calibra tion in this application were 4 cm (0.16 in) for grassed and 0.05 cm (0.02 in) for forested land cover and are within these published values but seem somewh at inconsistent. (i.e., lower for forested land cover). 188.8.131.52.2 Capillary Fringe/Root Zone and GWET Ross et al ., (2005a) discussed that there are four threshold conditions in the transition of vadose zone to water table ET in shallow water table settings and considered in conceptualizing the IHM. The satisfaction of land cove r ET demand from the vadose
96 zone or ground water (water table) can be (1) entirely by the va dose zone (deep water table condition) dictated by th e plant potential, (2 ) partially from both vadose zone and ground water (roots in contact with capillary zone) also limited by th e plant potential, (3) combined direct evaporation (augmentation) fr om the soil surface (capillary zone at land surface) and plant uptake, and (4) entirely by ground water at open water evaporation rates through direct evaporati on from the soil (water table or capillary fringe at land surface). The moving boundary of the capillary fringe or the capillary zone above the water table transitioning particul ar thresholds seems to be reasonable conceptually but, the model sensitivity to these thresholds remains uncertain. Decreasing capillary fringe thickness fr om 30 cm to 15 cm (-50% parameter change) significantly effects ET flux, in creasing UZET by 22%, GWET 24%, recharge 7% and infiltration 24%, respectively. Al so, LZET is decreased 7%, runoff 11% and DTWT 26%, respectively (in Table 5.3 a). Fo r the forested land cover (Table 5.4 a), increasing root zone thickness from 100 cm ( 40 in) to 200 cm (79 in) causes negligible change in UZET (3%), slightly decrease r unoff (5%), and only very slightly increases LZET (4%), GWET (4%) and infiltration (3 %), respectively, indicating relative insensitivity to root zone depth for this set of calibr ation conditions or this hydrological setting. One reason may be that, for this case, the deeper root zone extends well into the water table thereby only slightly influencing uptake of GWET. Capillary fringe and root zone thicknesses are physically-based but conceptual parameters derived from soil moisture retentio n data and/or expensive on-site analysis. In both observations and model behavior, under pa rticular conditions, either can strongly influence ET processes in shallow water ta ble settings indicated by the relatively high
97 sensitivity. For regional models, adequate char acterization data are not always available. Therefore uncertainty in these parameters s hould be considered in models for calibration and predictions. 184.108.40.206.3 INFILT and Distribution of Available Moisture From Table 5.4 b, it is shown that the total ET (TAET) is not very sensitive to soil infiltration rate (dictated by the INFILT parameter). While it is clear that the INFILT parameter effectively controls the overall division of the available moisture from precipitation (after interc eption) into surface and s ubsurface flow and storage components. Varying INFILT over a large range does not strongly influence TAET. However, the distribution of ET is somewhat more strongly sensitive to INFILT High values of INFILT will produce more water in the lowe r zone and groundwater (leading to more LZET and GWET); low values of INFILT will produce more upper zone water (more UZET), also resulting in greater direct overland flow. The INFILT parameter is primarily a function of soil characteristics and value ranges have been related to SCS hydrologic soil groups (Donigian and Davis, 1978) or soil characterization da ta (Fielland and Ross, 1991). Aly (2005) used exceptionally high INFILT values (15 cm/hr) for the same study site arguing that the We st-Central Florida soils are mostly sandy. However, Aly (2005) di d not examine the model for infiltration or ET distribution comparisons to observations. The INFILT values used in this study were 2 cm/hr for grassed and 4 cm/hr for fore sted land cover found through calibration comparisons to observed infiltration rates. For the soil type of the study site, Myakka
98 Fine Sand, the values used are reasonably close to hydraulic conductivities reported from soil tests (Carlisle et al ., 1989). Also the infiltration volumes and ET distributions observed through continuous soil moisture mon itoring compared quite favorable to model results for the 3-year record examined. 220.127.116.11.4 Plant Coefficient and LZET During the calibration and subsequent sensi tivity testing of the plant coefficient parameter, it was found that the distribution of plant coefficient play s a critical role in describing LZET and GWET. Figure 5.11 show s the comparison of monthly averaged LZET between sensitivity test F1 with distri bution PC1, F2 with the averaged monthly plant coefficient PC2 (constant values), and F3 with bell-shape pl ant coefficient PC3. PC1 was obtained from the analysis of da ta collected at the study site (Rahgozar et al ., 2005) and PC3 was an adjusted plant coeffi cient used by Aly (2005) (all the PCs are shown in Figure 5.11a). Figure 4.10b illust rated that the averaged monthly LZET with plant coefficient PC1 (used for calibration) has the best results when compared to the observed LZET. Poorer results are shown for PC 2 or PC3. It also appears that seasonal variability in plant coefficient is strongly warra nted as the constant value, PC2, exhibited both poor LZET and DTWT behavior. At the field study site in West-Central Florida, it was observed that plant communities develop new growth peaking in April with maximum leaf area in August, and perhaps this is the underlying basis for the best performance exhibited by PC1. the interesting dip is plant coefficient in July appears to be an artifact of high rainfall, shallow water table and ET dominated by inte rception. The model wa s later found to be
99 relatively insensitive to the particular value of PC for this period as dominate water table at land surface resulted in so il evaporation at or near PET for the period. Plant 0.4 0.6 0.8 1 123456789101112 MonthPlant Coefficient PC1 PC2 PC3 0.4 0.6 0.8 1 123456789101112 MonthPlant Coefficient PC1 PC2 PC3 0 3 6 9 12 15 123456789101112 MonthAveraged Monthly LZET (cm ) Observed F1 F2 F3 0 3 6 9 12 15 123456789101112 MonthAveraged Monthly LZET (cm ) Observed F1 F2 F3 Figure 5.11. Sensitivity to Plant Coe fficient Seasonal Variability for LZET (F1 with PC1, F2 with PC2, F3 with PC3) a) b)
100 coefficients obtained from field observations using th e method of Rahgozar (2006) appear to be reasonable and directly applicable to simulation modeling. 18.104.22.168 Air Entrapment Zhang and Ross (2006) and Shah, et al ., (2006) found significant Lisse effect evidence from field observations in the study site. Here, the IHM model application also illustrated these effects on infiltration and depth to water table as shown in Figure 4.10. Figure 4.10a and b are plots for the air entrap ment periods (May to July, 2002 and April to May, 2003). From IHM model simulati on compared to the observed DTWT for grassed land cover periodic differences are observed. The corresponding Figure 5.12c and d are the HYDRUS 1D (Simunek et al ., 2005) simulated solution during the same periods (Shah, et al ., 2006). Both the IHM and HYDRUS 1D model consistently showed the variations of the observed DTWT and the modeled DTWT with time. This was not surprising as both numerical models do not consid er air entrapment effect and thus results of DTWT from these models neglect to obser vations for these partic ular periods. It is believed that periodic departures from the actual DTWT during large rainfall events are indicative of air entrapment and pressurizati on warranting future i nvestigation. Figure 5.13 presents calibration results compared to the observed infiltration during periods (1) with or (2) and (3) without air entrapment s conditions. During air entrapment periods, observed infiltration was decreased compared to simulated values. In contrast, the simulation results without air entrapment ar e reasonably good compared to observed data for infiltration.
101 0 6 5/4/025/20/026/5/026/21/02DTWT(cm)0 3Rainfall (cm ) Observed DTWT Model DTWT Rainfall 180 7.5 0 6 5/4/025/20/026/5/026/21/02DTWT (cm)0 3Infilration (cm ) Equipment Failure Observed HYDRUS dW T Infiltration 180 7.5 0 6 4/1/034/21/035/11/035/31/03DTWT(cm)0 3Rainfall (cm ) Observed DTWT Model DTWT Rainfall 180 7.5 0 6 4/1/034/21/035/11/035/31/03DTWT(cm)0 3Infiltration (cm ) HYDRUS dWT Observed Infiltration 180 7.5 Figure 5.12. Air Entrapment Periods for Grassed Section (a) 5/4/02~7/1/02 and (b) 4/1/03~6/1/03 and Corresponding HYDRUS Soluti on (c) and (d) a) c) d) b) 101
102 0 50 100 150 200 1/1/027/20/022/5/038/24/033/11/04CumulativeInfiltration (cm) Observed Modeled Observed Modeled Air Entrapment Periods (1) (2) (3) 0 50 100 150 200 1/1/027/20/022/5/038/24/033/11/04CumulativeInfiltration (cm) Observed Modeled Observed Modeled Air Entrapment Periods (1) (2) (3) Figure 5.13. IHM Model Cumulative Infiltration Compared to Observations during Periods with or without Air Entrapment
103 CHAPTER 6 CONCLUSIONS AND RECOMMENDATIONS In this study, model consid erations for predicting vados e zone moisture dynamics, especially with respect to integrated surf ace and groundwater models, were investigated. The theoretical basis of vertical processes in IHM was discussed t horoughly and rigorous model sensitively and component performance testing was carried ou t to evaluate the robustness and limitation of the methodology of IHM. The conclusions and summary of the findings are organized as follows: 1) Several important modeling consider ations were identified warranting additional investigation and model development: a. Strong differential vadose zone rete ntion and frequent variability in antecedent conditions exists between the upper and lower vadose zone in field observations. b. Field-scale variability in moisture re tention and depth to water table is prevalent throughout, at relatively small spatial sc ale of < 100 m. c. Relative soil moisture appears to be a better indication of antecedent condition and tendency to support vadose zone vs water table ET. d. Air entrapment/pressurization affects the water table observation and thus infiltration, runoff and recharge. Also, air entrapment/pressurization appears to be relatively prolonged (sev eral days) in some instances.
104 2) A new and useful relative storage cond ition and further disc retization of the vadose zone was defined and advocated but was not fully tested within IHM. In order to explore the new mathematic al conceptualization of upper zone and lower zone behavior, analysis was made of daily average soil moisture obtained from field data. A formal definition for the upper soil zone was offere d as the A horizon (top 10~20 cm) which, possibly, could also be in cluded with surface depression storage. Relative soil moisture condition was defined as the difference between the actual and the minimum, divided by the mean avai lable (mean minus minimum) moisture for that water-table depth. A van Genuchten-type mathematical model was used to calculate the corresponding mean and minimum total mois ture values at all water table depths. Further, because of the unique behavior of the two distinct soil zones, exhibited in shallow coastal plain soils, and the discovery that the two zones can be and frequently are in different relative states (wet or dry), further supports the contention that the vadose zone relative storage should be spli t into an upper and lower region. For both the upper and lower zones, relative moisture is a function of water-table depth and stress history (wetti ng or drying). The lower zone extends from the upper zone (soil A horizon) down to the water table or the groundwater extinction depth, whichever is shallower. Moisture flux below this zone is only available to th e saturated groundwater domain and, thus, only is available as recharge. Because the upper and lower zone comm only exhibit different relative soil moisture conditions, the effects on infiltra tion, percolation and recharge will be significant. For example, a dry upper zone w ill yield higher infiltration when coupled
105 with a wet lower zone and may result in elevated recharge to the water table. Also, it is very apparent that the relative state of th e lower zone will dictate the distribution of plant ET from the vadose zone vs the water table in both the model and the physical domain. 3) The new relative soil moisture definition was tested considering field-scale variability. Several conclusions are offered fro this testing. Different land uses (vegetative covers) ex hibit different relative soil moisture behavior, even in close proximity and even with the same hydrologic soil classifications and subjected to the same mete orological stresses. Thus inf iltration, runoff and recharge behavior will be different. Therefore prediction by a model is strongly sensitive to and dependent on good ET performance, especially with regards to the distribution of ET from the various component storages. Comparison of ET performance by the IHM was made against field observations with gene rally good behavior shown, but further conclusions and observations are offered in item 5). Comparisons of UZ and LZ against total soil moisture for both upper and lower zones for different vegetative cover types (grass and forest) in low and high ET periods yield some interesting observations. It was shown that expressing soil moisture in this manner eliminates field-scale variability due to difference in soil moisture retention and better represents the antecedent soil moisture condition. 4) The present version of IHM is not totally compatible with the new relative storage definition. The upper zone algorithm s in HSPF appear to be suited only for describing depression storage effects.
106 5) Calibration testing of the IHM shows that the model, with parameter adjustments, can reproduce reasonable cumula tive behavior including distribution of ET, but reproduction of daily be havior is somewhat poorer. The testing and calibration of the IHM model for the vertical moisture retention and flux behavior was made for two land covers : grassed and forested, at a field site in West-Central Florida. The model was reasonably calibrated for a two-and-a-half-year simulation period (January, 2002 through July, 2 004) to describe in filtration (and thus runoff), ET distribution and water-table fluctuation (a nd thus recharge). Calibration results compared reasonably well with obser ved processes with reasonable parameter estimates from physical soil measurements, reported characterization data and only limited parameter adjustment. However, results are much better for multi-day and cumulative behavior than daily. 6) Several key parameters were tested in th e sensitivity analysis : saturation, field capacity, capillary fringe, root zone, soil infiltration index and nominal depression storage. The model shows strong sensitivity to soil properties and less sensitivity to plant properties (perhaps unique to this shallow-water-table-data setting). Thus, further testing in deeper water-table environments is strongly warranted. 7) Excess pressure was shown to be peri odically important by greatly reducing infiltration (thereby increasing runoff) and water-table observati ons. The extent and perseverance of air pressuri zation appears to be greatest w ith deeper water tables and with more intense rainfall. 8) The time-scale of observed excess air pressure ranged over many days (not just hours) which has been pr eviously shown. The ideal gas law approach, used to
107 understand the behavior, generally supports th e magnitude of observe d air pressurization differences but not the durations. The current study employed field data and numerical modeling, using HYDRUS1D to quantify the variation of air pressuriza tion values. The observations of water table in the field data departed significantly on o ccasions from the theoretical values using a calibrated RichardsÂ’ equation solution (and also IHM simulation results). Antecedent conditions were found to be very important in controlling air pr essurization. Deeper water table and higher rainfall intensity seemed to be the most prevalent conditions generating excess air pressurization. A simple analysis based on the ideal gas law was also done to help understand air pressurization effects. Interest ingly, the duration of excess ai r pressures was inconsistent between observations and model predictions Field observations of prolonged excess pressures (many days) can not be supported by ideal gas law analys is assuming uniform pressures in the vadose zone. This suggests th at air pressures in the vadose zone may be strongly non-uniform but, this is only a hypothesis at this ti me. Further investigation is strongly warranted. 9) Investigation shows small variability in vadose zone moisture content dictates whether ET will affect the water table, even in shallow water table conditions. Moisture conditions for water-table depths less than1/2 m appears to be near equilibrium consistent with the literature but belo w 1 m are highly variable. 10) The values of infiltration index, INFILT used are reasonably close to reported values from soil tests (Carlisle et al 1989), contrasting Aly (2005) exceptionally high values and traditional very low valu es used for regional HSPF only models.
108 11) Three plant coefficient distributions were investigated with findings suggesting that the plant coe fficient distribution has a str ong effect on lower zone and groundwater ET. The value obtained from th e field observation in the manner of Rahgozar (2006) yielded the best performance fo r lower zone ET. The decline shown in July found through this method appears to be anomalous. The model ar tificially resets the plant coefficient to near 1 for this period to account for direct soil evaporation at near potential values because the wate r table was at or near land surface for most of this time. It was also found that th ere is considerable bene fit in deriving the plant coefficient from observed data in the me thod of Rahgozar (2006), at least for areas similar to the study area and periods when the water table is not at land surface. 12) Simulation results indicate comprehe nsive integrated hydrological models such as IHM can reproduce water-table behavi or, soil moisture dist ribution incorporating field-scale variability and ET distribution and thus provide valuable predictive capability for continuous runoff or recharge studies. In this study, the appli cability of IHM for two different land covers in shallow water table (WTd <2 m) settings were shown. Reproducing ET: both total ET as well as ET di stribution was very important for runoff and recharge predictions. The model perfor med reasonably well for vertical moisture distribution, water table and runoff predicti on, especially with regard to multi-day and cumulative behavior. Several limitations and recommendations are offered: 1) At present, IHM cannot be parameteri zed to represent the shallow vadose zone (A horizon soils) as the uppe r zone. Upper zone storage in HSPF appears to be only
109 conceptualized as depression storage (e.g., ther e is no direct infiltration to the upper zone, only initial abstraction capt ure). Further study, code enhancements and testing are warranted. 2) Conceptual changes are recommended to allow differentiation of the upper and lower soil zones. Perhaps changes to the in filtration model are also warranted so that infiltration is more a function of a shallow soil state as well as possible conceptual changes to ET behavior in a model with upper and lower vadose zone. 3) Actual and relative soil moisture stor age was not compared to observations because of the problem with field measurem ents of the true elevation of saturation (periodic air entrapment deviations of obser ved water table). A strategy for addressing this occurrence needs to be derived and further investig ation of the behavior and constraints of air pressuriza tion in shallow water table hydrology needs to be made. 4) In shallow-water-table settings, brief peri ods of air entrapment play a role in controlling infiltration, runoff and observed de pth to water table and are not adequately simulated by IHM, RichardsÂ’ equation solutio n or any other known integrated surfacegroundwater model. 5) There are still IHM model parameters with high uncertainties, especially under different calibration conditions (e.g., d eep water table). Ther efore further fieldscale testing in different envi ronments is strongly warranted, especially in deeper water table settings.
110 REFERENCES Abbott, M. B., Bathurst, J.C., Cunge, J.A., OÂ’Connell, P.E. and Rasmussen, J. (1986a). Â“An Introduction to the Eu ropean Hydrological System Â– Systeme Hydrologique Europeen, SHE. 1 History and Philosophy of a Physically-based Distributed Modeling System.Â” J. Hydrol. 87, 45-59. Abbott, M. B., Bathurst, J.C., Cunge, J.A., OÂ’Connell, P.E. and Rasmussen, J. (1986b). Â“An Introduction to the Eu ropean Hydrological System Â– System Hydrologique Europeen, SHE. 2 Structure of a Physic ally-based Distribu ted Modeling System.Â” J. Hydrol. 87, 61-77. Adrian, D.D and J.B.Franzini (1966). Â“Imped ance to Infiltration by Pressure Build-Up Ahead of the Wetting Front.Â” Journal of Geophysical Research 71, 5857-5862. Aly, Alla (2005). Â“The Integrated Hudr ologic Model: A FieldScale Application.Â” Reston, VA ASCE/EWRI 0-7844-0792-4. Bicknell, B.R., Imhoff, J.C., Kittle, J.L. Jr, J obes, T. H., Donigian, A.S., Jr and Johanson, R.C. (2001). Â“Hydrological Simulation Program Â– FORTRAN (HSPF): UserÂ’s Manual for Version 12.Â” U.S. Environm ental Protection Agency, Athens, GA. < http://www.epa.gov/waterscience/basins/bsnsdocs.html >. Bidlake, W.M., Woodham, W.M., Lopez, M. A. (1993). Â“Evapotranspiration from Areas of Native Vegetation in West-Central Fl orida.Â” U. S. Geological Survey OpenFile Report 93-415, U. S. Geologi cal Survey, Tallahassee, FL. Carlisle, V.W., Sodek, F., Collins, M., Hammond, L.C. and Harris, W.G. (1989). Â“Characterization Data for Se lected Florida Soils.Â” Inst of Food and Agric. Sci., Univ. of Florida, Gainesville. Crawford, N.H. and Linsley, R.K. (1966). Â“D igital Simulation in Hydrology Â– Stanford Watershed Model IV,Â” Dept. of Civil Engine ering, Stanford University, Stanford, Calif. Crosbie, R.S., P.Binning, and J.D.Kalma (2005) Â“A Time Series Approach to Inferring Groundwater Recharge Using the Wa ter Table Fluctuation Method.Â” Water Resources Research 41 W01008, doi: 10, 1029/2004WR003077.
111 Donigian, A. S., Jr. and H. H. Davis, Jr. (1978). Â“UserÂ’s Manual for Agricultural Runoff Management (ARM) ModelÂ” EPA-600/3-78-080. U.S. E nvironmental Protection Agency, Athens, GA.Donigian, A. S. Danish Hydraulic Institute (DHI). (1998) Â“MIKE SHE Water Movement User Guide and Technical Reference Manual, Edition 1.1.Â” < http://www.dhisoftware.com/mikeshe/ >. Fielland, C. E., Ross, M. A. (1991) "Impr oved HSPF Infiltration Calibration Procedure with a Linked GIS", 1991 National Confer ence on Hydraulic Engineering and International Symposium on Groundwater, Nashville, Tennessee, July 1991. Garrote, L., and R. L. Bras. (1995). Â“A Distributed Model For Real-Time Flood Forecasting Using Digita l Elevation Models.Â” Journal of Hydrology 167, 279306. Geurink, Jeffrey S and Mark. A. Ross. ( 2006) Â“Time-Step Dependency of Infiltration Errors in the HSPF Model.Â” ASCE, J. Hydrologic Engrg ., Vol. 11, No. 4. Green, W.H., and C.A.Ampt (1911).Â”Studies on Soil Physics, 1, Flow of Air and Water through Soils.Â” Journal of Agricultural Science 4, 1-24. Hammecker, C, A. C. D. Antonino, J. L. Maeght, and P. Boivin (2003). Â“Experimental And Numerical Study of Wate r Flow in Soil under Irriga tion in Northern Senegal: Evidence of Air Entrapment.Â” European Journal of Soil Science 54, 491Â–503. Harbaugh, A.W., and McDonald, M.G. ( 1996). Â“Programmer's Documentation for MODFLOW-96, An Update to the U.S. Geological Survey Modular FiniteDifference Ground-Water Flow Model.Â” U.S. Geological Survey Open-File Report 96-486, Reston, Va. Heliotis, F.D. and C.B. Dewitt (1987). Â“Rap id Water Table Response to Rainfall in a Northern Peatland Ecosystem.Â” Water Resource Bulletin, 23(6), 10111016. Healy, R.W., and P.G. Cook (2002). Â“Using Gr oundwater Levels to Estimate Recharge.Â” Hydrogeology Journal 10, 91-109. Hernandez,T, M.Nachabe, M.Ross, and J. Obeysekera (2003). Â“ Modeling Runoff from Variable Source Areas in Humid, Sh allow Water Table Environments.Â” Journal of the American Water Resources Association 39(1), 75-85. Heillel, D (1998). Environmental Soil Physics, Academic Press, San Diego, CA.
112 Hooghoudt,S.B (1947). Â“Waaarnemingen van grondwaterstanden voor de landbouw. Commissie voor Hydrologisch TNO.Â” Verlag en Technische Bijeenkomsten 1-6: 185-201. HydroGeoLogic Inc. (2003). Â“MODHMS: A Comprehensive MODFLOW-Based Hydrologic Modeling System, Version 2.1, Code Documentation and UserÂ’s guide.Â” HydroGeoLogic Inc., Herndon, Virginia. < http://www.modhms.com/software.htm >. Ivanov, V.Y. (2002). Â“A continuous Real-time Interactive Basin Simulator.Â” M.S. Thesis, Massachusetts Institute of Technology. Jensen, M.E. and Haise, H.R. (1963). Â“E stimating Evapotranspiration from Solar RadiationÂ”. J. Irrig. Drainage Div. ASCE 89: 15-41. LaRoche J, van der Staay, GWM, Partensky F, Ducret A, Aebersold R, Li R, Golden SS, Hiller RG, Wrench PM, Larkum AWD and Green BR. (1996).Â”Independent Evolution of the Prochlorophyt e and Green-plant Chlorophyll a/b LightHarvesting ProteinÂ”. Proc Natl Acad Sc,i USA 93: 15244Â–15248. Latifi, H., S. N. Prasad, and O. J. Helweg (1994). Â“Air Entrapment and Water Infiltration in Two-Layered Soil Column.Â” J. Irrig. Drain. Eng., 120, 871Â–891. MacVicar, T.K. (1981). Â“Frequency Analysis of Rainfall Maximums for Central and South Florida.Â” South Fla. Water Manage. Dist. Tech. Publ. No. 81-3. McWhorter, D.B. (1971). Â“Infiltration Affected by the Flow of Air.Â” Colorado State University Hydrological Paper 49, Fort Collins, CO. Meyboom, P (1967). Â“Groundwater St udies in the Assiniboine River Drainage: Part II, Hydrologic Characteristics of Phreat ophytic Vegetation in Zouth-central Saskatchewan.Â” Geological Survey of Canada Bulletin, 139. Morel-Seytoux, H.J., and J.Khanji (1974). Â“Der ivation of an Equati on of Infiltration.Â” Water Resources Research, 10, 795-800. Morel-Seytoux,H.J, and J.Khanji (1975). Â“Equation of Infiltration with Compression and Counter flow Effects.Â” Hydrological Science Journal 20, 505-517. Myers, R. L. and Ewel, J. J. (1991). Ecosystems of Florida University of Central Florida Press, Orlando. FL.
113 Nachabe, M., N. Shah, M.A Ross, and J. Vomacka (2005). Â“Evapotranspiration of Two Vegetation Covers in a Shallo w Water Table Environment.Â” Soil Science Society of America Journal, 69:492-499. Nachabe, M.H., C. Masek, J.Obeysekra ( 2004). Â“Observations and Modeling of Profile Soil Water Storage above a Shallow Water Table.Â” Soil Science Society of America Journal, 68:719-724. Natural Resources Conservation Service (N RCS), Â“Lesson Plans: A Soil Profile.Â” < http://soils.usda.gov/education/ resources/k_12/le ssons/profile/ >. Panday, S. and Huyakorn, P.S. (2004). Â“A Fully Coupled Physica lly-Based SpatiallyDistributed Model for Evaluati ng Surface/Subsurface Flow.Â” Advances in Water Resour. 27, 361Â–382. Parlange, J. Y., and D. E. Hill (1979). Â“A ir and Water Movement in Porous Media: Compressibility Effects.Â” Soil Science 127, 257Â–263. Philip J.R. (1957). Â“The Theory of Infiltrati on, 4, Sorptivity and Algebraic Infiltration Equations,Â” Soil Science 84(3), 257-264. Powers, R.M., Davis, P.R., and Ross, M.A. (1989). Â“Modeling the Hydrologic Impacts of Phosphate Mining and Reclama tion in Central Florida,Â” AWRA 25th Annual Conference on Water: Laws and Management, Tampa, FL. Rahgozar, M., Ross, M., and Said, A. (2005) Â“Estimation of Evapotranspiration and Hydrologic Parameters Using Conti nuous Soil Moisture and Water Table Monitoring.Â” Hydrological Science and Technology, 21(1-4), 119-142. Rahgozar (2006). Â“Estimation of Evapotrans piration using Continuous Soil Moisture Measurement.Â” Dissertation, Depart ment of Civil and Environmental Engineering, University of South Florida. Refsgaard, J.C. and Storm, B. (1995). Â“M IKE SHE in: Computer Models of Watershed Hydrology.Â” V. P. Singh ed., Water Resour ces Publications, Colorado, USA, 809846. Richards.L.A (1931). Â“Capillary Conduction of Liquids through Porous Mediums.Â” Journal of Applied Physics 1(5), 318-333. Ross, M., Geurink, J., Aly, Al, Tara, P., Trout, K and Jobes, T. (2004). Â“Integrated Hydrologic Model (IHM)-Volume I: Theo ry Manual.Â” Water Resource Group, Dept. of Civil Engineering, Univ ersity of South Florida, USA. < http://hspf.com/pub/him/IHM_Theory_Manual.pdf >.
114 Ross, M., Geurink, J., Said, A., Aly, A and Tara, P. (2005a). Â“Evapotranspiration Conceptualization in the HSPF -MODFLOW Integrated Models.Â” JAWRA 41(5), 1013-1025. Ross, M., Trout, K., Tara, P., Said, A and Geurink, J. (2005b). Â“A New Discretization Scheme for Integrated Surface and Groundwater Modeling.Â” Hydrological Science and Technology 21(1-4), 143-156. Ross, M.A., Tara P.D., and Geurink J.S. ( 1998). Â“Review of the SDI Integrated Model ISGW,Â” prepared for Southwest Florida Water Management District, University of South Florida, Tampa, FL. Ross, M., Tara, P., Geurink, J., and Stew art, M. (1997). Â“FIPR Hydrologic Model Users Manual and Technical Documentation.Â” prepared for Florida Institute of Phosphate Research and Southwest Florid a Water Management District, Center for Modeling Hydrologic and Aquatic Systems, CMHAS Water Resources Report, FIPR.97.03, University of South Florida, Tampa, FL. Ross, M., Geurink, J. and J. Zhang (2007). Â“Modeling Vadose Zone Moisture Dynamics with the Integrated Hydrol ogic Model (IHM), Part I: Theoretical Basis.Â” ASCE, J. Hydrologic Engrg ., under review. Ross, M.A., and Tara, P.D. (1993). Â“Int egrated Hydrologic Modeling with GIS,Â” American Society of Civil Engineering,Â” Journal of Water Resources Planning and Management Vol. 119, No. 2. SDI Environmental Services Inc. (1999). Â“U pdate to Integrated Hydrologic Model of Central Northern Tampa Bay Area: IS GW/CNTB Model Run No. 121,Â” prepared for Tampa Bay Water, 87 p. plus appendices. Shah, N., and M.Ross (2006). Â“Variability in Specific Yield for Different Wetting and Drying Conditions.Â” Vadose Zone Journal, under review. Shah, N., Zhang, J. and M.Ross (2006). Â“L ong Term Air Entrapment Affecting Runoff and Water Table Observations.Â” Water Resour. Res., under Review. Simunek,J., and M.Th. van Genuchten (1999). Manual of HY DRUS-2D Computer Program for Simulation Water Flow, Heat and Solute Transport in Variably Saturated Porous Media. U.S. Departme nt of Agriculture, Riverside, CA. Simunek, J., M. Th. van Genuchten and M. Sejna (2005). Â“The HYDRUS-1D Software Package for Simulating the Movement of Water, Heat, and Multiple Solutes in Variably Saturated Media.Â” Version 3.0, HYDRUS Software Series 1, Department of Environmental Sciences, University of California Riverside, Riverside, California, USA, 270 pp.
115 Software Spotlight (2000). Â“Review of HYDRUS-2D.Â” Ground Water 38(1), 10-11. Sophocleous, M.A., Koelliker, J.K., Govindaraju, R.S., Birdie, T., Ramireddygari, S.R. and Perkins, S.P. (1999). Â“Integrated Numerical Modeling for Basin-Wide Water Management: The Case of the Rattlesnake Creek Basin in South-Central Kansas.Â” J. Hydrol., 214, 179-196. Touma, J., G. Vachaud, and J.-Y. Parlange (1984). Â“Air and Wate r Flow in a Sealed, Ponded Vertical Soil Column: Experiment and Model.Â” Soil Science, 137, 181Â– 187. Trout, K and Ross, M. (2005). Â“Intensive Hydrologic Data collection in a Small Watershed in West-Central Florida.Â” Hydrological Science and Technology 21(14), 187-197. Tucker GE, Lancaster ST, Gasparini NM, Br as RL, Rybarczyk SM. (2001). Â“An ObjectOriented Framework for Distributed Hydrologic and Geomorphologic Modeling Using Triangulated Ir regular Networks.Â” Computers & Geosciences 27(8): 959973. Vachaud, G., J.P. Gaudet, and V.Kuraz (1974) Â“Air and Water Flow During Ponded Infiltration in a Vertical Bound Column of Soil.Â” Journal of Hydrology 22:89108. van Genuchten, M.Th. (1978). Â“Calculating th e Unsaturated Hydraulic Conductivity with a New Closed-Form Analytical Model.Â” Research Report 78-WR-08, Water Research Program, Dept. of Civil Engin eering, Princeton University, Princeton, New Jersey, USA. Vivoni, E.R., Ivanov, V.Y., Bras, R.L., Entekhabi, D. (2003). Â“Generation of Triangulated Irregular Networks Ba sed on Hydrological Similarity.Â” J. Hydrologic Engrg ., 9(4), 288-302. Wang, Z., J. Feyen, D. R. Nielsen, and M. T. van Genuchten (1997). Â“Two-Phase Flow Infiltration Equations Accounting for Air Entrapment Effects.Â” Water Resources Research, 33(12), 2759-2767. Wang, Z., J. Feyen, D. R. Nielsen, and M. T. van Genuchten (1998).Â” Air Entrapment Effects on Infiltration Rate and Flow Instability.Â” Water Resources Research, 34(2), 213-222.
116 Waterstone Environmental Hydr ology and Engineering, Aqua Terra Consultants, Richard Allen, and John Wilson (2001). Â“Uncertainty and Sensitivity Analysis for the Central Northern Tampa Bay Area Inte grated Hydrologic Model: Phase IIÂ”. Weeks, E.P (2002). Â“The Li sse Effect Revisited.Â” Ground Water 40(6), 652-656. WEST Consultants Inc., Gartner Lee Lim ited and Aqua Terra Consultants (2001). Â“Scientific Review of the Integr ated Hydrologic Model ISGW/CNTB121Â”. Yan, J.J. and Smith, K.R. (1994). Â“Simula tion of Integrated Surface Water and Ground Water Systems Model Formulation.Â” Wa ter Resources Bulletin, 30(5), 1-12. Youngs,E.G., and A.J.Peck (1964). Â“Moisture Profile Development and Air Compression during Water Uptake by Bounded Porous Bodies, 1, Theoretical Introduction.Â” Soil Science 98, 290-294. Zhang, J. and M. A Ross (2006), Â“A 2-Layer Vadose Zone Model for SurfaceGroundwater Interactions.Â” J. Hydrologic Engrg ., Accepted.
117 ABOUT THE AUTHOR Jing Zhang was born and bred in P.R. Ch ina. She received her M.S degree in Civil Engineering (CE) at XiÂ’an University of Science and Technology with high reputation in the resear ch area of numerical method applie d in CE. Without satisfied her education, She entered United States in A ugust, 2002 to pursuit her another M.S degree in Civil and Environmental Engineering (CEE), specified in water resources area at University of Louisiana. Continuously, she tr ansferred to Universi ty of South Florida starting her Ph.D. program in CEE since Ja nuary, 2004 under the direction of Dr. Mark A. Ross. She has interest in the areas of hydrologic, hydraulic and water quality modeling. Recent research incl udes use of GIS in hydrologi c analysis, integration of surface and groundwater flow models. She comp leted all the requirements for the Ph.D. program in the spring of 2007.