USF Libraries
USF Digital Collections

Rainfall-runoff modeling in humid shallow water table environments

MISSING IMAGE

Material Information

Title:
Rainfall-runoff modeling in humid shallow water table environments
Physical Description:
Book
Language:
English
Creator:
Hernandez, Tatiana X
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Runoff -- Computer simulation   ( lcsh )
Slopes (Physical geography)   ( lcsh )
hydrologic modeling
rainfall variability
variable source areas
Dissertations, Academic -- Civil Engineering -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Simulating the processes of rainfall and runoff are at the core of hydrologic modeling. Geomorphologic features, rainfall variability, soil types, and water table depths strongly influence hydrological process in Florida ecosystems. Topographic characteristics of the terrain define the stream paths and landscape. Alteration of these characteristics as a result of urban and/or agricultural developments, for example, can highly influence wetlands and river basin response. There are two predominant landforms in Florida: wetlands, where Variable Saturated Areas form near streams causing saturation excess runoff, and uplands where runoff is mainly generated by infiltration excess. The objective of this work is to analyze the impacts of geomorphologic and hydrologic characteristics on runoff mechanisms in humid environments such as Florida. In general, most research at the hillslope scale use hypothetical values of rainfall, sometimes non-realistic values, and single slope^forms to explain the geomorphic and hydrologic process on Variable Saturated Areas. In this thesis, the complexity of hillslope processes on actual Florida topography is assessed by coupling a Digital Elevation Model with a two-dimensional variable saturated-unsaturated flow model called HYDRUS-2D. Actual rainfall records and soil parameters from the Characterization Data for Selected Florida Soils, Soil Survey were used to evaluate hydrologic impacts. A commercial software package, River Tools was used to display and extract topographic information from the Digital Elevation Models.Results show that when inflitration excess runoff is dominant, infiltration and runoff are very sensitive to time resolution, especially for convective storms. When saturation excess occurs, runoff is not affected by rainfall intensity. However, saturated hydraulic conductivity, depth to the water table, slope and curvature highly influence the extent of Variable Saturated Areas. Results indicate runoff in shallow water table environments is produced mainly by subsurface storm runoff, running below the surface, except in hillslopes with concave curvature and mild slopes. Additionally, concave hillslopes generate more saturation excess runoff than straight and convex hillslopes.
Thesis:
Thesis (M.S.)--University of South Florida, 2001.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Tatiana X. Hernandez.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 123 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001413320
oclc - 47655549
notis - AJJ0736
usfldc doi - E14-SFE0000010
usfldc handle - e14.10
System ID:
SFS0024701:00001


This item is only available as the following downloads:


Full Text

PAGE 1

RAINFALL RUNOFF MODELING IN HUMID SHALLOW WATER TABLE ENV IRONMENTS Tatiana X. Hernandez T hesis submitted on the Faculty of the University of South Florida in partial fulfillment of the requirements for the degree of Master of Science I n Civil Engineering Mahmood H. Nachabe, Ph.D., P.E. Mark A. Ross, Ph.D., P.E. Jayantha Obeysekera, Ph.D., P.E. May 5, 2001 Tampa, Florida Keywords: Variable Source Areas, rainfall variability, hydrologic modeling Copyright 2001, Tatiana X. Hernandez

PAGE 2

DEDICATION This work is dedicated to my family, Oscar, Blanca, Kelly, and Mauricio for their love, support and patience. To all of them for encouraging me each step along the way. Los quiero mucho

PAGE 3

ACKNOWLEDGMENTS I like to thank Professor Rudolf Henning for providing the rainfall data, whic h is part of a study being conducted at the University of South Florida, Electrical Engineering Department and NASA. I want to acknowledge the South Florida Water Management District for providing funds to support this research. I also want to thank Jeff Vomacka for providing field measurement data. His assistance is gratefully acknowledged. I wish to thank Dr. Mahmood Nachabe, Dr. Jayantha Obeysekera and Dr. Mark Ross, the members of my committee for taking the time and effort to ensure my work was of quality. I would particularly like to thank Dr. Nachabe for his continuous guidance, and encouragement. His support and wise counsel were invaluable. Finally, I want to thank my family and friends for being always there for me. Thank you all.

PAGE 4

i TABLE OF CONTENTS LIST OF FIGURES ............................................................................................................ iii LIST OF TABLES ............................................................................................................. vii ABSTRAC T ...................................................................................................................... viii CHAPTER 1. INTRODUCTION .......................................................................................1 1.1. Background ..........................................................................................................1 1.2. Objective and Scope ............................................................................................3 1.3. Rainfall Variability ..............................................................................................4 1.4. Variable Saturated Areas .....................................................................................8 CHAPTER 2. RAINFALL ANALYSIS ...........................................................................15 2.1. Rainfall Data ......................................................................................................16 2.2. Rainfall Model Description ...............................................................................17 2.3. Statistical Analysis .............................................................................................20 2.4. Temporal Variability of Rainfall .......................................................................26 CHAPTER 3. INFILTRATION EXCESS RUNOFF .......................................................33 3.1. HYDRUS 1D .....................................................................................................34 3.2. Impacts to Depth to the Water Table .................................................................38 3.3. Impacts to Temporal Variability of Rainfall .....................................................41 CHAPTER 4. DIGITAL ELEVATION MODELS (DEM) .............................................46 4.1. DEM Data ..........................................................................................................48 4.2. River Tools 2.0 ..................................................................................................51 CHAPTER 5. SATURATION EXCESS RUNOFF .........................................................53

PAGE 5

ii 5.1. HYDRUS 2D Model ..........................................................................................54 5.2. Reference Simulation .........................................................................................56 5.3. Hydrologic Characteristics ................................................................................62 5.4. Geomorphologic Characteristics .......................................................................64 5.5. Simulation Results .............................................................................................66 CHAPTER 6. CONCLUSIONS ........................................................................................73 REFERENCES ..................................................................................................................75 APPENDICES ...................................................................................................................80 Appendix A. Subroutine Rainfall_Minute ..................................................................81 Appendix B. Subroutine Event_Minute .....................................................................86 Appendix C. Output File from the Event_Minute Subroutine ...................................89 Appendix D. Florida Soil Survey for Myakka Fine Sand ..........................................90 Appendix E. HYDRUS 1D Graphs for Temporal Variability of Rainfall .................91 Appendix F. HYDRUS 2D Graphs for Reference Simulation ...................................99 Appendix G. Water Retention Curves for Actual Soil Data and Fitter Parameters. ..102 Appendix H. Hyetographs for Convective and Frontal Storms ................................103 Appendix I. Initial Water Table Leve l. .....................................................................105 Appendix J. Graphic Results for Soils Sensitivity Analysis .....................................106 Appendix K. Graphic Results for Rainfall Sensitivity Analysis ..............................107 Appendix L. Graphic Results for Water Table Depths .............................................109 Appendix M. Graphic Results for Slope Sensitivity ................................................110 Appendix N. VSA Extent for Concave Curvature ....................................................111

PAGE 6

iii LIST OF FIGURES Figure 1. Mechanisms of infiltration and runoff. .............................................................2 Figure 2. Radar image for a frontal storm in Florida. Note: values are in dBA ..............6 Figure 3. Radar image for a convective storm in Florida. Note: value s are in dBA .......7 Figure 4. Convective storms from a rainfall gauge located at the University of South Florida, Tampa campus ...................................................................................15 Figure 5. Frontal storms from a rainfall gauge located at the University of South Florida, Tampa campus ...................................................................................16 Figure 6. Cumulative frequency distribution for storm duration ....................................23 Figure 7. Cumulative frequency distribution for storm depth ........................................24 Figure 8. Cumulative frequency distribution for rainfall intensity .................................26 Figure 9. Hyetograph for 1 minute time resolution of rainfall .......................................27 Figure 10. Hyetograph for 3 minute time resolution of rainfal l .......................................27 Figure 11. Hyetograph for 5 minute time resolution of rainfall .......................................28 Figure 12. Hyetograph for 10 minute time resolution of rainfall .....................................28 Figure 13. Hyetograph for 15 minute time resolution of rainfall .....................................28 Figure 14. Hyetograph for 30 minute time resolu tion of rainfall .....................................29 Figure 15. Hyetograph for 60 minute time resolution of rainfall .....................................29 Figure 16. Variance vs time resolution (minutes) in natural scale ...................................31

PAGE 7

iv Figure 17. Variance vs time resolution (minutes) in Log scale ........................................31 Figure 18. Static initial water content profile for a convective storm ..............................35 Figure 19. Pressure head at the surface for convective storm ..........................................36 Figure 20. Rainfall infiltration and runoff vs depth to water table ..................................40 Figure 21. Percentage error in runoff depth against time resolution of rainfall ...............45 Figure 22. TIN and contour lines ......................................................................................49 Figure 23. 1 foot contour map for Lithia near the Tampa Bay Regional Reservoir ........50 Figure 24. Shaded relief of the DEM at Doe Creek near a proposed Reservoir ...............52 Figure 25. Reference simulation profile at the reservoir site, Doe Creek ........................56 Figure 26. Initial water table level for the cross section ...................................................57 Figure 27. Van Genuchten fitted water retention curve ...................................................58 Figure 28. Water content vs head obtained by HYDRUS 2D ..........................................59 Figure 29. 2 D mesh for cross section at the reservoir site ..............................................60 Figure 30. Boundary conditions for the cross section domain .........................................61 Figure 31. Profile for 0.5%, 1.1%, and 2.5% slopes .........................................................64 Figure 32. Concave profiles for three negative radius of curvature. ..................................65 Figure 33. Convex profiles for three negative radius of curvature ...................................65 Figure 34. Divergence parameter for HYDRUS 2D ........................................................66 Figure 35. Water retention curves for different soils used in simulations ........................67 Figure 36. Pressure head distribution for the reference simulation ..................................69 Figure 37. Shaded relief of the Lithia DEM with several flow paths ...............................72 Figure 38. Sample copy of actual soil hydraulic parameters for Myakka fine sand ........90 Figure 39. Rainfall rate vs time ........................................................................................91

PAGE 8

v Figure 40. Water retention curve ......................................................................................91 Figure 41. Water content vs depth (1 meter) ....................................................................92 Figure 42. Inf iltration rate vs time (1 meter) .....................................................................92 Figure 43. Cumulative infiltration depth vs time (1 meter) ..............................................92 Figure 44. Water content vs d epth (2 meters) ...................................................................93 Figure 45. Infiltration rate vs time (2 meters) ...................................................................93 Figure 46. Cumulative infiltration depth vs time (2 meter s) ............................................93 Figure 47. Water content vs depth (5 meters) ...................................................................94 Figure 48. Infiltration rate vs time (5 meters) ...................................................................94 Figure 49. Cumulative infiltration depth vs time (5 meters) ............................................94 Figure 50. Rainfall rate for convective storm... ................................................................95 Figure 51. Pressure head at the surface (1 meter) ..............................................................95 Figure 52. Water content vs depth (1 meter) ....................................................................96 Figure 53. Infiltration rate vs time (1 meter). .....................................................................96 Figure 54. Cumulative infiltration depth vs time (1 meter) ..............................................96 Figure 55. W ater content vs depth (2 meters) ...................................................................97 Figure 56. Infiltration rate vs time (2 meters) ...................................................................97 Figure 57. Cumulative infiltration depth vs time (2 meters) ............................................97 Figure 58. Water content vs depth (5 meters) ...................................................................98 Figure 59. Infiltration rate vs time (5 meters) ...................................................................98 Figure 60. Cumulative infiltration depth vs time (5 meters) ............................................98 Figure 61. Profiles of reference simulations from HYDRUS 2D .....................................99 Figure 62. Pressure heads at the surface boundary. .......................................................100

PAGE 9

vi Figure 63. Subsurface storm flux during rainfall ............................................................100 Figure 64. Cumulative subsurface storm flow during rainfall ........................................100 Figure 65. Subsurface flux after rainfall .........................................................................101 Figure 66. Cumulative seepage flux after rainfall ..........................................................101 Figure 67. Seepage flux after rainfall .............................................................................101 Figu re 68. Fitter van Genuchten parameters for Myakka fine sand ................................102 Figure 69. Fitter van Genuchten parameters for Mulat fine sand.. .................................102 Figure 70. Fitter van Genuchten parameters for Millhopper fine sand ..........................102 Figure 71. Convective storms used for sensitivity analysis ............................................103 Figure 72. Frontal storms used for sensitivity analysis ..................................................104 Figure 73. Initial water table level for sensitivity analysis .............................................105 Figure 74. Subsurface fluxes for 3 soil types .................................................................106 Figure 75. Saturation excess fluxes for 3 soil types .......................................................106 Fi gure 76. Comparison of fluxes for Miakka fine sand ..................................................106 Figure 77. Saturation excess fluxes for convective stroms .............................................107 Figure 78 Saturation excess fluxes for frontal storms ...................................................107 Figure 79. Subsurface fluxes for convective storms .......................................................108 Figure 80. Subsurfac e fluxes for frontal storms .............................................................108 Figure 81. Subsurface flux after rainfall stops ................................................................109 Figure 82. Seepage flux after rainfall sto ps ....................................................................109 Figure 83. Cumulative surface flux for different slopes .................................................110 Figure 84. Cumulative subsurface flux ...........................................................................110 Figure 85. Extent of VSA after rainfall for concave profiles .........................................111

PAGE 10

vii LIST OF TABLES Table 1 Basic statistics for 1 minute total rainfall depth and duration in Tampa. ..........21 Table 2 Basic statistics for actual and 1 minute rainfall intensities in Tampa, Fl. .........22 Table 3 Main statistics for aggregated hyetographs of a convective storm ....................29 Table 4 Main statistics for aggregated hyetographs of a frontal storm ...........................30 Table 5 Total runoff for different depths to the water table using HYDRUS 1D ..........38 Table 6 Runoff for convective and frontal storm using HYDRUS 1D ...........................41 Table 7 Error in runoff depth for different time steps .....................................................43 Table 8. Soil Parameter changes in a n, and K s ............................................................. 67 Table 9 Changes in runoff cumulative flux es because of variable rainfall intensity ......68 Table 10 Runoff cumulative fluxes due to changes in the initia l water table depth .......69 Table 11 Runoff cumulative fluxes for different slopes in the profiles ..........................70 Table 12 Runoff cumulative fluxes for different curvatures ...........................................71 Table 13 Sample of ou tput file obtained using the event_minute subroutine .................89

PAGE 11

RAINFALL RUNOFF MODELING IN HUMID SHALLOW WATER TABLE ENVIRONMENTS Tatiana X. Hernandez ( ABSTRACT) Simulating the processes of rainfall and runoff are at the core of hydrologic modeling. Geomorphologic features, rainfall variability, soil types, and wa ter table depths strongly influence hydrological process in Florida ecosystems. Topographic characteristics of the terrain define the stream paths and landscape. Alteration of these characteristics as a result of urban and/or agricultural developments, for example, can highly influence wetlands and river basin response. There are two predominant landforms in Florida: wetlands, where Variable Saturated Areas form near streams causing saturation excess runoff, and uplands where runoff is mainly generated by infiltration excess. The objective of this work is to analyze the impacts of geomorphologic and hydrologic characteristics on runoff mechanisms in humid environments such as Florida. In general, most research at the hillslope scale use hypothetical va lues of rainfall, sometimes non realistic values, and single slope forms to explain the geomorphic and hydrologic process on Variable Saturated Areas. In this thesis, the complexity of hillslope processes on actual Florida topography is assessed by coupli ng a Digital Elevation Model with a two dimensional variable saturated unsaturated flow model called HYDRUS 2D. Actual rainfall records and soil parameters from the Characterization Data for Selected Florida Soils, Soil Survey were used to evaluate hydrol ogic impacts. A commercial software package, River Tools was used to display and extract topographic information from the Digital Elevation Models.

PAGE 12

Results show that when infiltration excess runoff is dominant, infiltration and runoff are very sensitive to time resolution, especially for convective storms. When saturation excess occurs, runoff is not affected by rainfall intensity. However, saturated hydraulic conductivity, depth to the water table, slope and curvature highly influence the extent of Var iable Saturated Areas. Results indicate runoff in shallow water table environments is produced mainly by subsurface storm runoff, running below the surface, except in hillslopes with concave curvature and mild slopes. The extent of Variable Saturated Are as on mild slope topographies is larger than for steep slopes. Additionally, concave hillslopes generate more saturation excess runoff than straight and convex hillslopes.

PAGE 13

1 CHAPTER 1. INTRODUCTION 1.1. Background Simulating the processes of rainfall, infiltration, and runoff is a primary objective of hydrologic modeling. These processes are influenced by the temporal and spatial variability of rainfall, geomorphologic features of the terrain, soil properties, and depth to the water table. Adequate prediction of runoff is essential for hydrologic modeling required for floo d forecasting, storm water management plans, reservoir management, development, and permitting. In the last decades infiltration excess overland flow, also called Hortonian Runoff, was the traditional concept applied for analysis of runoff response. La ter studies showed that Hortonian runoff mechanism was not applicable to all watersheds ( Dunne and Black, 1970a ). In south and central Florida, runoff or overland flow can be generated by two mechanisms: infiltration ex c ess and saturation excess runoff. Infiltration excess occurs when the rainfall intensity exceeds the infiltration capacity of the soil. This type of runoff usually occurs in deep water table environments and is influenced by the temporal variability of a ctual rainfall. In contrast, saturation excess runoff occurs when the total infiltration depth exceeds the soil water storage capacity above the water table. This mechanism is more likely to develop in a shallow water table

PAGE 14

2 environment, where infiltratio n depth fills the storage capacity of the soil, and excess saturation spills into the surface as runoff. Figure 1a) depicts the infiltration excess runoff mechanism for a vertical profile with depth against water content. During rainfall, water infiltrat es into the soil increasing the water content at the surface. From Figure 1a), the moisture content profiles increase with time. Downward movement of the water into the soil takes place until saturation at the surface is reached. At this time, ponding a t the surface occurs. For a time t less than the ponding time, tp the infiltration rate is equal to the rainfall rate and greater than the infiltration capacity of the soil. a) b) Figure 1 Mechanisms o f infiltration and runoff. Source: R. A. Freeze, A Stochastic Conceptual Analysis of Rainfall Runoff Process on a Hillslope, Water Resources Research. 16(2): 395, 1980.

PAGE 15

3 After ponding takes place, infiltration rate decreases, overland flow or runoff inc reases and can be estimated as the difference between infiltration rate and infiltration capacity. Notice that the water table level in this case is deep and is not affected by rainfall or infiltration processes. Saturation excess runoff is explained in Figure 1b). The profile of depth against moisture content shows how the moisture content in the soil increases with time. Since the water table is shallow, the storage capacity of the soil is small. Therefore, water going into the soil raises the water table levels filling the porous media and reaching surface saturation from below. For a time t less than the time for surface saturation, the infiltration rate is equal to the rainfall rate and there is not runoff. For a time t equal or greater than th e time for surface saturation, the storage capacity of the soil is full and infiltration stops. Thus, all the rain runs off. During infiltration, some of the water may also run laterally below the ground surface. This flow is called subsurface flow and it is a faster flow that runs in the vadose zone until it enters a stream. 1.2. Objective and Scope The purpose of this thesis is to analyze these two mechanisms of runoff and determine how geomorphologic characteristics and hydrologic processes impact these s hallow water table environments. This work is divided into six chapters. The first chapter is the introduction and contains a background of the runoff mechanisms including infiltration excess and saturation excess runoff. It explains the need to go into fine time resolution of rainfall when infiltration excess runoff is prevalent and the importance of

PAGE 16

4 topographic changes of the landscape in variable saturated areas dominated by saturation excess runoff. Chapter 2 contains the analysis of rainfall data. It explains what type of rainfall data is used and how it is aggregated. A statistical analysis of rainfall and how variability is lost when coarse time resolution of rainfall is used are comprised in this chapter as well. Chapter 3 describes HYDRUS 1D, a variable saturated model for 1 dimensional water movement in the soil. This chapter includes simulations with HYDRUS 1D to analyze the impacts in runoff as a result of temporal variability in rainfall and depth to the water table. Chapter 4 contains a description of Digital Elevation Models (DEM) and how their features can be utilized in variable saturated water movement models to describe topographic features of actual terrains. Chapter 5 of this thesis explains the saturation excess runoff process a nd the sensitivity analysis of geomorphologic and hydrologic characteristics that have an effect on Variable Saturated Areas (VSA). The complexity of hillslope processes on realistic non simple slopes is analyzed here. The development of techniques for t he identification, description, or extent of variable saturated areas is also included. And last of all, Chapter 6 summarizes the conclusions of this work. 1.3. Rainfall Variability The estimation of infiltration excess runoff is highly dependent on rainfall variability. Often hydrologic models such as SWMM and HSPF used to simulate the hydrologic processes at the large scale watershed scale and use large time resolution of rainfall time series. Therefore, the effects introduced by the internal structure o f rainfall are ignored. Gammeter and Fankhauser ( 1998 ) have expressed the necessity of

PAGE 17

5 finer resolution to capture the temporal variability of rain. In their work, they developed a program that disaggregates 10 minute obs erved rainfall data into 1 minute time step. Due to the lack of natural, fine time resolution rainfall data, hydrologists have been forced to create stochastic models to simulate the temporal structure of rainfall. Arnold and Williams ( 1989 ) developed a stochastic model to generate 0.5 hr rainfall intensities from daily rainfall series. The temporal distribution of rainfall generated in this study is exponentially distributed within the storm. Woolhiser and Goodrich (198 8) developed a stochastic model to disaggregate daily rainfall into short period rainfall intensity. They showed that the disaggregation method is superior to the constant and isosceles method. The constant intensity rainfall pattern was not recommended in their study. Flanagan et al. (1988) used a programmable rainfall simulator to reproduce several storm intensity patterns. This study showed that runoff is greatly affected by rainfall intensity patterns. Rodriguez Iturbe and Eagleson (1987) created a precipitation model to reproduce rainfall fields. These models in general simulate rainfall variability under limited conditions, and cannot be used for all types of rainfall patterns due to the complex variability of rainfall in time. Some studies of r ainfall runoff have been done using actual rainfall data. Agnese and Bagarello (1997) studied the temporal resolution of natural storms on the infiltration predicted by the Chu model (Agnese and Bagarello, 1997) on both peak and time to peak intensity. R esults of their study showed that the influence of the temporal resolution of rainfall on infiltration prediction was highly dependent on soil type, especially for soils

PAGE 18

6 with intermediate hydraulic conductivity. For these soils, an increase in the tempora l resolution of rainfall led to a decrease of runoff prediction. When coarser resolution was used, predicted variability was underestimated. In general, variable rainfall intensities generate greater peak discharge than constant rainfall intensities (Sin gh, 1997). Currently, most rainfall data used for storm water modeling is collected daily or hourly by rainfall gauges distributed around Florida. Recently, rainfall data from NEXRAD images having a resolution of 15 minutes are becoming available for mod eling. Figure 2 and 3 depict radar images of total daily rainfall depth for two storms in Florida. The first graph shows a frontal storm that occurred on January 10, 1999. This storm is uniformly distributed over a large domain with an average intensity of 7 dba (0.4 inches). Figure 2 Radar image for a frontal storm in Florida. Note: values are in dBA

PAGE 19

7 Figure 3 Radar image for a convective storm in Florida. Note: values are i n dBA Figure 3, on the other hand, shows a convective storm that occurred on June 26, 1999. Total rainfall depth for this storm is scattered over the domain coverage area. The storm has spots where total daily rainfall was about 15 dba (1.2 inches). Ra infall data at a finer resolution is rarely collected, and if collected, it is not distributed over a large domain. Several studies have been conducted to determine the effect of rainfall variability on runoff (Lamb, 1999; Loague, 1988; Sepulveda, 1997; M illy and Eagleson, 1988). The complex structure of rainfall has been found to exhibit a large degree of temporal variability under natural conditions, which substantially affects the downward water movement in the vadose zone. Other studies have shown th at accuracy of hydrologic

PAGE 20

8 models can be improved by using fine time step resolution of rainfall data in the determination of infiltration excess runoff (Finnerty et. al., 1997). In order to determine infiltration and runoff, it is necessary to know how mu ch water infiltrates in the soil. Ground water movement in the soil is governed by Richards equation. Assuming vertical downward ground water movement on a homogeneous soil, the infiltration process can be simulated with the one dimensional Richards equat ion. Because estimates of Hortonian runoff depends on rainfall intensity, hyetographs with a fine time resolution are usually required to capture rainfall variability. 1.4. Variable Saturated Areas The other mechanism of runoff, which develops in shallow wate r table depths, is saturation excess runoff. Dunne and Black (1970a) studied the runoff generation mechanisms on the Sleepers River Experimental Watershed near Danville, Vermont. Dunne and Black (1970a) noticed that surface runoff was produced on hillsid e by saturation from below. This led to the concept of saturation excess overland flow (Kirkby and Chorley 1967; Dunne, 1978) which take place when water fills up the storage capacity of the soil with the water table reaching the surface, and producing ru noff. This type of overland flow is likely to occur close to the stream banks where water table is shallow and a seepage face develops. Along with the concept of saturation excess overland flow are Variable Saturated Areas (VSA). Several authors studie d Variable Saturated Areas since the 60s and observed that VSA were the major contributor of overland flow especially in humid environments

PAGE 21

9 (Hewlett, 1961; Hewlett and Hibbert, 1963; Dunne and Black, 1970a,b; Freeze, 1972a,b; and Kirkby, 1978). Hewlett ( 1961) first described the concept of variable source areas as saturation excess runoff production. Then, Hewlett and Hibbert (1963) studied the contribution of runoff by extension of variable source areas. According to Dunne and Black (1970a,b), the vari able saturated areas were controlled by topography, antecedent water content, soil moisture storage capacity, and rainfall intensity. A detail study about the impacts of these controlling characteristics was not included in his work. Dunne and Black (197 0a,b) used a real watershed of steep slopes and poorly drained downslope. Here, variable saturated areas were produced by overland flow due to precipitation and return flow, or the subsurface flow that seeps through the surface and becomes overland flow. In thin soils, gentle concave downslopes with wide valley bottoms, runoff is dominated by direct precipitation and return flow (Kirkby, 1978). On an investigation of runoff generation at the hillslope scale, Freeze (1972a,b) used a small range of possibl e combinations of soils, slopes, and rainfall patterns and noticed that the most sensitive factor influencing runoff was the saturated hydraulic conductivity. In his study, rainfall intensity, duration, soil thickness, and slope did not have a major effec t on runoff. Numerous researchers have also developed and used hydrologic models to generate overland flow. Some of these models simulate the hydrologic processes at the watershed scale (OLoughlin, 1986; Stagnitti et al., 1992, Wigmosta et al., 1994; St eenhuis, 1995; Evans et al., 1999) and some others at the hillslope scale (Loague, 1988; Govindaraju and Kavvas, 1991; Salvucci and Entekhabi, 1995; Ogden and Watts, 2000).

PAGE 22

10 OLoughlin (1986) used a 3 dimensional topographic model as a function of hillslop e height to predict growth and contraction of VSA. His analyses assume runoff saturation caused by subsurface flow. This model does not account for transient changes in saturated boundaries because steady state conditions were part of the assumptions. S tagnitti et al. (1992) developed a mathematical model to predict hillslope and watershed discharge in a real basin with shallow water table depths and average slope. In their work, rapid subsurface flow traveling along the granite ledge produced runoff. Overland flow was not significant. Wigmosta et al. (1994) developed a spatially distributed hydrology vegetated model that generates runoff by saturation excess on the snowmelt dominated mountain basin. Digital Elevation Model (DEM) data is used here to m odel topographic characteristics. The extent of VSA was not discussed in this study. Steenhuis (1995) used the SCS approach to calculate runoff produced merely by VSA. This simplistic method do not account for temporal variation, rainfall intensity or l andscape formations. New attempts to describe Variable Saturated Areas have been done at the hillslope scale. For example, Loague (1988) used the Freeze stochastic rainfall simulator model SCRRS to create different rainfall intensities and soil hydraulic properties and to produce overland flow. He observed that saturated hydraulic conductivity was more important than rainfall intensity for convex topography. Govindaraju and Kavvas (1991) developed a physics based deterministic distributed model by coupl ing overland flow, stream flow, and subsurface flow to study the saturated areas near streams. They analyzed the effect of hydraulic conductivity, rainfall intensity, antecedent moisture

PAGE 23

11 content, and slope. Landscape convergence and divergence, as well as concave and convex hillslopes were not considered in their work. Salvucci and Entekhabi (1995) described an example application of a statistical dynamical approach that coupled saturated and unsaturated flow to illustrate the sensitivity of hydrologic process to variations in topographic and climatic conditions. They used a non real hillslope on a semi humid climate of certain length, soil depth, and slope with no curvature. Some variations of these parameters were analyzed for an equilibrium steady st ate condition. Results of this study showed that a decrease in soil depth (shallow water tables) increased evaporation, decreased baseflow, and produced more runoff. An increase in soil depth increased recharge and produced no surface runoff. Convergenc e and divergence hillslopes were analyzed here as defined by the convergence ratio of the contour length at the water divide to the length at the stream. Ogden and Watts (2000) used a 2 dimensional variable saturated subsurface model (VS2D) under unifor m rainfall rate. They analyzed the importance of hillslope properties and rainfall on the temporal evolution of variable saturated areas. In their study, Ogden and Watts (2000) simulated the response of a non convergent, constant slope hillslope. They s howed that hillslopes with shallow water tables, constant rainfall intensity, smaller slope angles, longer hillslope lengths, and smaller saturated hydraulic conductivity are more susceptible to variable saturated area formations. Neither hillslope curvat ure, nor landscape convergence was included in this investigation.

PAGE 24

12 Rainfall variability seems to have an impact in the infiltration runoff process near streams with shallow water tables. VSA expand and contract as a result of temporal variation in rainfa ll. During rainfall events the VSA expand. Then, when rainfall stops water is taken by evapotranspiration, redistribute in the soil, and seeps as subsurface drainage. As a consequence, the variable saturated areas shrink. When rainfall intensity increa ses, the variable saturated areas increase (Govindaraju and Kavvas, 1991). Topographic attributes such as slope and curvature are important factors in defining the extent of VSA. Mild slope topographies have larger VSA than steep slopes. Therefore, mor e water infiltrates in the soil raising the water table levels and saturating the ground surface. Two types of hillslope curvature are studied in this work: concave and convex hillslope. Concave hillslopes curve upward whereas convex hillslopes curve do wnward. Freeze (1972b) found out that on convex hillslopes with deeply incised channels and high saturated hydraulic conductivity subsurface flow is very significant. On the other hand, for concave hillslopes with low saturated hydraulic conductivity run off generation comes from direct precipitation rather than from subsurface flow. He also pointed out that saturated hydraulic conductivity has major effects on runoff. Kirkby (1978) observed that topography with concave slopes have a larger contributing area. Convergence and divergence is another landscape characteristic that affects variable saturated areas. On convergent hillslopes, the flow area reduces downstream and a larger source area sustains a smaller discharge area at the stream. In contrast, the flow area on divergent hillslopes increases downhill approaching the stream with a larger

PAGE 25

13 discharge area. Topographic settings like convergence and divergence are important in Florida because VSA are controlled by their surrounding upland landscapes. There has not been relevant research on convergence topography and how it affects the runoff response. Recently, Salvucci and Entekhabi (1995) studied the hillslope fluxes as a result of landscape convergence for long time response. They defined conver gence as the ratio of contour length at the water divide to contour length at the stream. Their findings demonstrated that VSA increase for convergent landscape and tend to generate more runoff. Philip (1996) looked at the convergence and divergence pert urbations on infiltration at the hillslope. He found these perturbations to be negligible small. Initial conditions of the soil also affect the extent of variable saturated areas because they are highly influenced by the initial position of the water tab le. For shallow water tables, the initial moisture content of the soil is high, the storage capacity of the soil decreases and saturation excess occurs faster. As a result, the VSA extend more rapidly. This does not happen during the dry season. During winter, the water table position is usually deeper. Therefore, the antecedent moisture content of the soil is lower and the soil has more storage capacity. Under these conditions the soil takes longer to be fully saturated. Stream shape and incision d epth also affect VSA. Large incision depth increases the area of the subsurface in contact with the stream. Therefore, the contribution of subsurface water to the stream increases (Govindaraju and Kavvas, 1991).

PAGE 26

14 In general, most research at the hillslop e scale has only studied single slope forms to explain the hydrologic process on variable saturated areas. This work analyzes natural topography in Florida and the impacts in saturation excess runoff due to changes in geomorphologic properties. Flat slop es usually with shallow water table depths and numerous wetlands, some of them isolated, characterize Florida landscapes. Wetlands are saturated areas that shrink during the dry season and expand in the wet season as part of their hydro periods. Wetlands are considered important and valuable ecosystems because they maintain biodiversity and wildlife habitat, provide flood protection, stream flow maintenance, and recharge control. Furthermore, wetland ecosystems are very important for hydrologic modeling, flood forecasting, reservoir management, wetlands conservation plans, soil pollution, etc. In forested wetlands near streams, Variable Saturated Areas form due to saturation excess runoff mechanism. Alteration of topographic characteristics of the terra in can highly influence wetlands and the river basin response. The extent of these variable saturated areas is controlled by several factors such as landscape changes, rainfall variability, topography, landscape convergence, initial moisture conditions, a nd stream sinuosity and incision depth. Therefore, impacts of geomorphologic and hydrologic features on runoff in humid shallow water tables should be studied.

PAGE 27

15 CHAPTER 2. RAINFALL ANALYSIS Rainfall directly influences surface and subsurface runo ff, streamflow, and groundwater recharge. Two types of mechanisms characterize Florida rainfall: frontal and convective. Frontal precipitation usually occurs when a cold front from the north results in lifting of air masses. Frontal precipitation genera tes longer periods of rain and small intensities usually during the dry season, November through March. In contrast, convective rainfall during the wet season (May through September) is characterized by short storm duration with fairly high rainfall inten sities. The months of April and October excluded above are transition periods that can be either convective, frontal or both. Two thirds of the rainfall in Central Florida are due to convective thunderstorms originating inland during the summer. Figure 4 and Figure 5 shows a typical rainfall hyetograph for a convective and a frontal storm, respectively, obtained from a gauge station in the Tampa area. Figure 4 Convective storms from a rainfall gauge located at the Univers ity of South Florida, Tampa campus Rainfall Hyetograph Sep 20, 1998 Convective type of rain 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8 9 10 11 12 Time (hr) Intensity (in/hr) Total depth: 5.37 in Duration: 3.12 hr Average intensity: 1.72 in/hr

PAGE 28

16 Figure 5 Frontal storms from a rainfall gauge located at the University of South Florida, Tampa campus The complex structure of these rainfall events cannot be captured when rainfall da ta is collected at large time resolution intervals such as daily and hourly time steps. In order to study the temporal variability of rainfall, actual precipitation has been used and converted to a 1 minute time series. A statistical analysis of intensit y of rain, total duration, and depth of rainfall storm is carried out here and is explained below. 2.1. Rainfall Data Collecting data at a small time resolution or if possible continuous data is crucial in this research. Currently, data is collected at 15 min ute or greater time steps. Rainfall data at a fine resolution is rarely collected or is not collected over a large domain. Moreover, the data have to be collected over a long period of time in order to perform a statistic analysis of the data. Fine time resolution data is required to analyze the temporal variability of rainfall in Florida. Accordingly, a suitable source for this research is a local rainfall gauge placed in Tampa. The rainfall data employed was collected from the roof of the Engineering Building II at the University of South Florida. Data is obtained from a Rainfall Hyetograph Dec 27, 1997 Frontal type of rain 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8 9 10 11 12 Time (hr) Intensity (in/hr) Total depth: 4.10 in Duration: 12.75 hr Average intensity: 0.32 in/hr

PAGE 29

17 tipping bucket gauge, which records time to tip with a depth resolution of 0.01 inches per tip. Water flows directly and without time delay from the funnel to the bucket. Time inte rvals between consecutive tips are recorded. The rainfall records used dated from March 1995 to September 1998, which is about four years of continuous rainfall data. Long term time series of rainfall at this resolution are not recorded. Furthermore, th ey are not needed because of the scale of this research. The tipping bucket gauge is part of a study being conducted at the University of South Florida, Electrical Engineering Department and NASA. 2.2. Rainfall Model Description A model to disaggregate the co ntinuous rainfall data was developed. This is a model created with Visual Basic code and is divided in 2 subroutines: Rainfall_minute and Event_minute The model creates 1 minute time series, separate storms, and calculates total rainfall depths and dur ations for each storms. Rainfall_minute is a subroutine that converts the actual data in 1 minute time resolution. The purpose of this subroutine is to create a time series of 1 minute time resolution from continuous rainfall data files. The program re ads date, time, and time to tip in seconds from a text file and calculates the intensity of rain and cumulative depth of rain every minute. 1 minute intervals are linearly interpolated to find the intensity and cumulative rainfall depth up to that period. The source code for this subroutine can be found in Appendix A. The steps used by the 1 minute time step rainfall program include :

PAGE 30

18 Input file: it is a comma separated ASCII file (.csv) with three columns including date, time and time to tip in seconds. The program reads and accumulates time to tip. If time to tip is less than 5 minutes, then it starts accumulating the time and depth of rain. If time to tip for a single tip is greater than 5 minutes (no rain within 5 minutes or more), then it assigns to that period the same time to tip as the next period that tips in less than 5 minutes. As a result, intensities generated by the rainfall model are not small than 0.002 in/min (0.01 inches / 5 minutes). Long periods with no tips have intensities equal to zero. Linear Interpolation: Once the file is read and values are cumulated, the subroutine interpolates linearly every minute and calculates the cumulative rainfall depth in inches as well as the intensity for each minute. When the time to tip is less than 1 minute the program averages intensities by distributing rainfall depth over the minute. Output file: this is a comma separated ASCII file with a text extension suffix (.txt) and contains the following columns: Date and time, time in minutes, inten sity in in/hr every minute, and cumulative depth of rainfall in inches. Event_minute is the subroutine that separates storms. The 1 minute time step rainfall data were previously separated into convective and frontal activity. Convective storms are stor ms during the months of May through September. Frontal storms are the

PAGE 31

19 storms during the months of November through March. The months of April and October were separated manually by looking at their rainfall intensities and duration. Once the rainfall da ta are divided into convective and frontal activity, they are entered into the model in order to separate storms events. Convective storms are separated using 1 hour inter event period of no rain (Bosch and Davis, 1999). That is, every time there is a pe riod greater than 1 hour with no rain there is a new convective storm. There are no criteria established to separate frontal storms. However, experience with actual data shows that frontal storms have long durations and occur about every 8 hours or more. Therefore, 8 hour inter event period is used for frontal storms. The source code for the Event_minute subroutine can be found in Appendix B. The steps used for the subroutine to separate storms include Input file: it is a comma separated ASCII file ( .txt) generated previously by the Rainfall_minute subroutine. This file contains four columns including date and time, time every minute, intensity in in/hr, and cumulative depth of rainfall in inches. The subroutine distinguishes between frontal or conve ctive activity. Therefore, only one of these criteria can be used at a time. Calculations: Once the file is read, cumulative depth of rain and duration for each storm are calculated. Rainfall depth is cumulated each minute until there is a period of 1 hour or more with no rainfall. In that case, the program records an event or storm and print to file the cumulative depth and duration up to that time.

PAGE 32

20 Then, a new storm begins and these variables are reinitiated and start cumulating depth of rain and du ration again all the way to the end of the file. Output file: this is a comma separated ASCII file with a text extension suffix (.txt) that contains the following output columns: date and time, time every minute, intensity in in/hr, cumulative depth of ra in in inches, total duration for each storm in hours, and total rainfall depth for each storm in inches. Appendix C includes a sample of the output file generated by this subroutine. The actual rainfall data were converted to 1 minute time series. Then, the time series were separated in convective and frontal storms. It was observed that some of the convective activity occurred during the dry months were frontal type of storms were expected. Therefore, these storms were separated manually to avoid error s. 2.3. Statistical Analysis From the four years of actual record data with 1 min time resolution, 409 storms were generated. 316 storms were convective and 93 were considered frontal. Statistical analysis for these data was done. Basic statistics such as m ean, standard deviation, coefficient of variation, median, kurtosis, and skewness were calculated for the 1 minute time series. The basic statistics for the rainfall data are shown in Table 1. Rainfall statistics shown in Table 1 have been computed for d epth of rainfall and duration of storms. From Table 1, it can be concluded that mean total rainfall depth and duration for frontal storms are greater than for convective storms. The mean rainfall depth for frontal storms is 0.94 inches compared to 0.31 i nches for convective storms.

PAGE 33

21 Table 1 Basic statistics for 1 minute total rainfall depth and duration in Tampa All storms 1 minute Total Depth (in) Values All storms 1 minute Total Duration (hr) Values Mean 0.457 Mean 2 .269 Standard Error 0.040 Standard Error 0.222 Median 0.160 Median 0.739 Mode 0.020 Mode 0.050 Standard Deviation 0.799 Standard Deviation 4.497 Sample Variance 0.638 Sample Variance 20.226 Kurtosis 24.792 Kurtosis 18.574 Skewness 4.186 Skewn ess 3.891 Range 7.800 Range 35.204 Minimum 0.010 Minimum 0.033 Maximum 7.810 Maximum 35.237 Sum 187.040 Sum 928.161 Count 409 Count 409 Frontal 1 minute Total Depth (in) Values Convective 1 minute Total Depth (in) Values Mean 0.943 Mean 0.314 Standard Error 0.128 Standard Error 0.030 Median 0.500 Median 0.130 Mode 0.350 Mode 0.020 Standard Deviation 1.236 Standard Deviation 0.539 Sample Variance 1.529 Sample Variance 0.290 Kurtosis 9.991 Kurtosis 33.431 Skewness 2.688 Skewn ess 4.783 Range 7.790 Range 5.380 Minimum 0.020 Minimum 0.010 Maximum 7.810 Maximum 5.390 Sum 87.680 Sum 99.360 Count 93 Count 316 Frontal 1 minute Total Duration (hr) Values Convective 1 minute Total Duration (hr) Values Mean 7.243 Me an 0.806 Standard Error 0.764 Standard Error 0.054 Median 4.454 Median 0.488 Mode 1.351 Mode 0.050 Standard Deviation 7.365 Standard Deviation 0.952 Sample Variance 54.236 Sample Variance 0.906 Kurtosis 3.092 Kurtosis 10.212 Skewness 1.650 Sk ewness 2.653 Range 35.170 Range 7.168 Minimum 0.067 Minimum 0.033 Maximum 35.237 Maximum 7.201 Sum 673.564 Sum 254.597 Count 93 Count 316

PAGE 34

22 Table 2 Basic statistics for actual and 1 minute rainfall intensities in Tampa, Fl. 1 minute Intensities (in/hr) for 4 years Actual data intensities (in/hr) for 4 years Mean 0.527 Mean 1.572 Standard Error 0.005 Standard Error 0.011 Median 0.239 Median 1.100 Mode 0.122 Mode 4.105 Standard Deviation 0.734 Standard Deviation 1.570 Sample Variance 0.538 Sample Variance 2.466 Kurtosis 12.516 Kurtosis 5.611 Skewness 3.186 Skewness 1.675 Range 8.119 Range 17.734 Minimum 0.000 Minimum 0.000 Maximum 8.119 Maximum 17.734 Sum 11222.398 Sum 30141.103 Count 213 07 Count 19176 Frontal 95 98 actual data intensity (in/hr) Convective 95 98 actual data intensity (in/hr) Mean 1.250 Mean 1.853 Standard Error 0.014 Standard Error 0.017 Median 0.752 Median 1.478 Mode 2.875 Mode 2.963 Standard Deviation 1 .371 Standard Deviation 1.676 Sample Variance 1.879 Sample Variance 2.809 Kurtosis 5.062 Kurtosis 5.603 Skewness 1.802 Skewness 1.555 Range 16.216 Range 17.734 Minimum 0.000 Minimum 0.000 Maximum 16.216 Maximum 17.734 Sum 11176.852 Sum 18964 .252 Count 8944 Count 10232 Frontal 95 98 1 minute data intensity (in/hr) Convective 95 98 1 minute data intensity (in/hr) Mean 0.431 Mean 0.656 Standard Error 0.005 Standard Error 0.009 Median 0.219 Median 0.272 Mode 0.122 Mode 0.011 S tandard Deviation 0.586 Standard Deviation 0.878 Sample Variance 0.344 Sample Variance 0.771 Kurtosis 18.315 Kurtosis 7.992 Skewness 3.748 Skewness 2.613 Range 6.643 Range 8.119 Minimum 0.001 Minimum 0.000 Maximum 6.644 Maximum 8.119 Sum 5260 .799 Sum 5961.600 Count 12220 Count 9087

PAGE 35

23 Table 1 also confirms that frontal storms have longer durations than convective storms. Frontal storm mean duration for this rainfall record is 7.24 hr whereas the mean duration for convective storms is 0.8 hr Rainfall intensities were also compared with the actual intensities for the 4 years of records and are shown in Table 2. From this table, it can be noticed that the rainfall intensities calculated from the 1 minute time series are considerably small th an the actual rainfall intensities. This is because there were a lot of small intensities created when the rainfall data were aggregated into 1 minute time steps. These small non realistic intensities created a bias on the time series that makes the avera ges smaller. Cumulative Frequency Distribution 1min / 4 year record for storm duration 0% 20% 40% 60% 80% 100% 0 10 20 30 40 Storm duration (hr) Cumulative Frequency distribution (%) All storms Convective Frontal Figure 6 Cumulative frequency distribution for storm duration Figure 6 and Figure 7 depict the cumulative frequency distributions for total duration and depth of storms respectively. It is clear that convective storms have smaller storm duration with 90% of the storms having durations less than 2 hours. In contrast,

PAGE 36

24 90% of the frontal storms have durations of about 17 hours or less. Storm rainfall depth on the other hand, is more sim ilar for convective and frontal storms, being greater for frontal storms. Looking at 90% of the storms, the rainfall depth for convective storms is 1 inch or less and the rainfall depth for frontal storms is approximately 2.5 inches. Cumulative Frequency Distribution 1min / 4 year record for storm depth 0% 20% 40% 60% 80% 100% 0 2 4 6 8 10 Storm depth (in) Cumulative Frequency distribution (%) All storms Convective Frontal Figure 7 Cumulative frequency distribution for storm depth For each storm, total rainfall depth, storm duration, maximum and average intensity are determined. In conclusion, statistics showed that convective storms hav e more variability, shorter duration and lower total rainfall depth than frontal storms. On the other hand, frontal storms are more uniform distributed, have longer duration and larger total rainfall depth. Figure 8 shows the cumulative frequency distrib ution of intensities for the four years of data. The graph illustrates different cumulative frequency distributions of 1 minute time series for all 4 years of rainfall intensities. With the purpose of reducing the

PAGE 37

25 bias in the intensities when the 1 minut e time series where created, intensities less than 0.25 in/hr were deleted from the record and frequency distributions where calculated again. The mean rainfall intensity for all record is 0.53 in/hr. After ignoring rainfall intensities less that 0.25 in /hr, the mean intensity is 0.93 in/hr. The same was done but now ignoring rainfall intensities less than 0.50 in/hr. As a result, the mean intensity increased to 1.41 in/hr, which compared better with the mean intensity for the actual record of 1.57 in/h r. From Figure 8, after ignoring all intensities less than 0.50 in/hr, this graph matches closer the graph for the actual (raw) rainfall intensities. All small intensities were not necessary created by the rainfall model. Some of these intensities could have been captured by the rainfall gauge as part of a moving storm that was either approaching or moving away from the gauge. Intensity (in/hr) for 4 years rainfall data 0.00% 20.00% 40.00% 60.00% 80.00% 100.00% 120.00% 0 1 2 3 4 5 6 7 8 9 10 Intensity (in/hr) Cummulative Frequency Distribution (%) Raw data 1-min rainfall intensities intensities >0.25 in/hr intensities >0.5 in/hr Figure 8 Cumulative frequency distribution for rainfall intensity

PAGE 38

26 2.4. Tempor al Variability of Rainfall Trying to describe the temporal and spatial variability of actual precipitation is one of the objectives of hydrologist. Several authors have demonstrated that runoff is sensitive to this variability. From the four years of re cords, two rainfall storms, one convective and one frontal were selected to analyze the temporal variability of these storms and how details are lost when coarse time resolution of rainfall data is used. The 1 minute time step hyetographs for the storms a re illustrated on Figure 4 and Figure 5. These hyetographs were aggregated in different time steps including 3, 5, 10, 15, 30 minutes and 1 hour. Hyetographs of the convective storm were created for these time resolutions and are shown in the figures bel ow. The convective storm selected had a total duration of 3.12 hours and rainfall depth of 5.37 inches. The frontal storm had a total duration of 12.75 hours and rainfall depth of 4.1 inches. Basic statistics such as mean, standard deviation, and varian ce were calculated for the aggregated hyetographs. Figure 9 Hyetograph for 1 minute time resolution of rainfall Hyetograph for 1-minute time step 0.00 2.00 4.00 6.00 1 31 61 91 121 151 Time (min) Intensity (cm/min)

PAGE 39

27 Figure 10 Hyetograph for 3 minute time resolution of rainfall Fi gure 11 Hyetograph for 5 minute time resolution of rainfall Figure 12 Hyetograph for 10 minute time resolution of rainfall Hyetograph for 3-minute time step 0.00 2.00 4.00 6.00 3 33 63 93 123 153 Time (min) Intensity (cm/min) Hyetograph for 5-minute time step 0.00 2.00 4.00 6.00 5 60 115 170 Time (min) Intensity (cm/min) Hyetograph for 10-minute time step 0.00 2.00 4.00 6.00 10 60 110 160 Time (min) Intensity (cm/min)

PAGE 40

28 Figure 13 Hyetograph for 15 minute time resolution of rainfall Figure 14 Hyetograph for 30 minute time resolution of rainfall Figure 15 Hyetograph for 60 minute time resolution of rainfall Hyetograph for 15-minute time step 0.00 2.00 4.00 6.00 15 60 105 150 Time (min) Intensity (cm/min) Hyetograph for 30-minutes time step 0.00 2.00 4.00 6.00 30 60 90 120 150 180 Time (min) Intensity (cm/min) Hyetograph for 60-minutes time step 0.00 2.00 4.00 6.00 60 120 180 Time (min) Intensity (cm/min)

PAGE 41

29 From these hyetographs, it is clear how temporal variability in the intensities for convective storms is lost. The hyetograph at 1 min time resolution shows the rainfall variability during the 3 hour rainfall storm. The average intensity is 0.08 cm/min but there is considerable diffe rence with respect to the mean. Table 3 and Table 4 show the mean, standard deviation, and variance for the convective and frontal storms respectively. This variability is being lost when the time resolution increases from 1 min to 60 min. When intensiti es are aggregated in coarser time steps, they are smoothed out and the intensities become more uniform distributed over time. For comparison, the intensities for the 1 minute time resolution hyetograph shows intensities as high as 6 cm/min, whereas the in tensities for the 1 hour time resolution are less that 3 cm/min. As we move to coarser time resolution, the details in rainfall distribution are lost. The hyetograph at 60 min time resolution doesn't show the real distribution of rainfall on time; it sho ws a uniform storm with an average intensity of 0.08 cm/min. The mean does not change when the time resolution changes. It is constant whereas the standard deviation and variance decrease when the time resolution increases. Therefore, large time resolut ion (60min, 30 min, 15 min) has less variation with respect to the mean. Table 3 Main statistics for aggregated hyetographs of a convective storm Time step (minutes) Mean Standard deviation Variance 1 0.0757 0.0648 0.0042 3 0 .0757 0.0627 0.0039 5 0.0757 0.0613 0.0038 10 0.0757 0.0581 0.0034 15 0.0757 0.0545 0.0030 30 0.0757 0.0472 0.0022 60 0.0757 0.0312 0.0010

PAGE 42

30 Table 4 Main statistics for aggregated hyetographs of a frontal storm Time step (mi nutes) Mean Standard deviation Variance 1 0.0133 0.0256 0.0007 3 0.0133 0.0243 0.0006 5 0.0133 0.0228 0.0005 10 0.0133 0.0209 0.0004 15 0.0133 0.0211 0.0004 30 0.0133 0.0163 0.0003 60 0.0133 0.0131 0.0002 The variance vs time resolution graphs bel ow illustrate how much details are being lost when the time resolution of the rainfall time series is coarse. For example, looking at the graph below for convective storms, the variance when small time steps are used is larger than the variance for coarse r time resolution. As a result, more details in the temporal variability of rainfall can be captured using small or fine time steps. Plots for variance and log variance against time resolution are shown in Figures 16 and 17. Variance vs Time resolution 0.000 0.001 0.001 0.002 0.002 0.003 0.003 0.004 0.004 0.005 0.005 0 10 20 30 40 50 60 70 Time resolution (min) Variance frontal convective Figure 16 Variance vs time resolution (minutes) in natural scale

PAGE 43

31 Variance vs Time resolution 0.00010 0.00100 0.01000 1 10 100 Log (Time resolution, minutes) Log (Variance) frontal convective Figure 17 Variance vs time resolution (minutes) in Log scale Comparing the variance for frontal storms wit h the variance for convective storms, it can be concluded that frontal storms are more uniform and their variability is smaller. From the graph in natural scale, the variance for convective storms is about 4 times larger than the variance for frontal stor ms. Figure 17 demonstrates, in logarithmic scale, how many details are captured in the temporal variability of rainfall. Notice that the pattern of the curves in this graph shows a rapid decrease in the variance for time resolutions larger than 10 minute s. It explains the loss in details for large time resolution rainfall series.

PAGE 44

32 CHAPTER 3. INFILTRATION EXCESS RUNOFF Infiltration excess runoff or Hortonian Runoff depends on generation of surface runoff by rainfall excess. Infiltration excess ru noff occurs when the rainfall intensity exceeds the infiltration capacity of the soil. Runoff is calculated as the difference between rainfall intensity and infiltration capacity. This type of runoff occurs especially in soils with deep water table level s. After ponding, the time at which ground surface reaches saturation, water starts running off while part of it may still infiltrate. Due to the temporal variability of rainfall, there are periods when the rainfall intensity is less than the saturated h ydraulic conductivity or periods of no rain at all. These periods are called hiatus (Ogden and Sagharian, 1997) and can be of any duration. During hiatus, water infiltrating into the soil moves downward due to gravity and capillary drive. When the hiatu s periods are long (3 hours or more), water redistribute in the soil and the moisture content decreases with time. The model used here accounts for redistribution of the water content in the soil. The infiltration runoff process for deep water table en vironments can be impacted by temporal variability of rainfall, as well as depth to the water table. For instance, convective storms having short bursts of high rainfall intensities can generate more

PAGE 45

33 infiltration excess runoff when fine time resolution of rainfall is used. Frontal storms on the other hand, lasting longer periods with low intensities, have a slower response. 3.1. HYDRUS 1D A commercial software package, HYDRUS 1D, for simulating the one dimensional movement of water in variably saturated med ia was utilized to study the impacts of temporal variability of rainfall on runoff. This program was developed by U.S. Salinity Laboratory, U.S. Department of Agriculture, Agriculture Research Service (Simunek and van Genuchten, 1998). HYDRUS 1D is an in teractive graphics based user interface developed in support of the computer model HYDRUS (FORTRAN source code). HYDRUS 1D may be used to simulate one dimensional water flow, heat, and solute transport in variably saturated soils. For this, HYDRUS uses t he Richards equation to predict water movement between the soil surface and the groundwater table. The user interface includes data pre processing and graphical presentation of the output results in Microsoft Windows environment. Data pre processing invo lves specification of all necessary parameters to successfully run the FORTRAN source code, discretization of the soil profile into finite elements, and definition of the vertical distribution of hydraulic and other parameters characterizing the soil profi le (Simunek and van Genuchten, 1998)

PAGE 46

34 HYDRUS 1D module contains the main program unit and the overall computational environment of the system. This module controls execution of the program and determines which other optional modules are necessary for a pa rticular application. The module contains both the pre processing and post processing units. The pre processing unit includes specification of all parameters for the FORTRAN code. At first, the processes to be simulated are selected (water flow, chemica l transport, heat transport, root growth, and/or root water uptake). Length unit, depth and inclination of the soil profile to be analyzed, and the number of materials to be used are specified. The time and space scales are selected, and boundary conditi ons are entered among many other operations. A small catalog of soil hydraulic parameters based on soil texture is part of the pre processing unit. The user can choose between direct simulation and inverse solution, where parameter optimization can be car ried out. Figure 18 Static initial water content profile for a convective storm -200 -150 -100 -50 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Theta [-] Initial wc 1 hr 2 hr 3 hr

PAGE 47

35 The post processing unit consists of simple x y graphics for graphical presentation of the soil hydraulic properties, time changes of a particular variable at selected observation points in the profile, and actual or cumulative water fluxes across the upper and lower boundaries. Water content and pressure head profiles can be obtained. Examples of these profiles are shown in Figures 18 and 19 where variable rainfall rate was used. Figure 18 explains the redistribution process of water content in the soil for a storm that lasted about 3 hours. The soil depth for this example is 200 meters. Initial static water content was assumed for t his simulation (black solid line). Notice that water redistributes in the soil during 2 and 3 hours. The water content profile for 3 hours has smaller water content at the surface and moved downward with time. This process takes place when there is no ra infall or when the rainfall is smaller than the saturated hydraulic conductivity. Figure 19 Pressure head at the surface for convective storm -200 -150 -100 -50 0 0 50 100 150 200 250 Time [min]

PAGE 48

36 Distribution of the pressure head with time for infiltration excess runoff c an be explained in Figure 19. The simulation parameters for this graph are the same as in Figure 18. In this graph, when the pressure head is zero, saturation at the surface occurs. This might have been caused by a burst of high rainfall intensity, satu rating the surface. Therefore, the infiltration rate decreases, water infiltrates into the soil, and pressure head decreases. When another burst of high intensity comes or when the soil saturates due to accumulation of water at the surface, the pressure head increases until zero value, which is saturation. After 140 minutes approximately, rainfall stops, water drains and pressure head decreases. The profile module is conformed by the discretization of the flow domain in a graphical mode and specificatio n of domain properties. It is an external module which may be used to discretize a one dimensional soil profile into finite difference cells. The module also helps a user to specify the initial conditions in the pressure head and the water content, as we ll as the spatial distribution of other parameters characterizing the soil profile (e.g., spatial distribution of soil materials and root water uptake parameter) and/or observation nodes. All parameters in this module are specified in a graphical environm ent. HYDRUS 1D solves numerically the Richards equation for variably saturated water flow in one dimension (Simunek and van Genuchten, 1998). The flow equation incorporates a sink term to account for water uptake by plant roots. Richards equation for one dimensional vertical flow with one dependent variable is

PAGE 49

37 Where, q (z,t) is the water content of the soil (c m 3 of water/c m 3 of soil) h( q ) is the matric potential head z is the depth of the soil K ( q ) is the unsaturated hydraulic conductivity. t is the t ime The program considers prescribed head and flux boundaries, boundaries controlled by atmospheric conditions such as rate of application at the surface or evapotranspiration, and free and deep drainage boundary conditions assuming transient flow. It ha s three different models to calculate the hydraulic conductivity: van Genuchten (1980), Brooks and Corey (1964), and van Genuchten modified. Hysteresis in the soil hydraulic properties can be specified as well. 3.2. Impacts to Depth to the Water Table To anal yze the effect of depth to the water table on runoff and infiltration, the infiltration process was simulated utilizing HYDRUS 1D. The assumptions used for this simulation follow. A convective storm selected from the actual rainfall record with 3.1 hour duration and total depth of 13.6 cm was used. The time resolution of the storm is 1 minute time step. This storm is a representative observed storm in this region. Soil properties of a representative soil in Florida (Mulat fine sand) obtained from the s oil survey was used. d d d d = d d 1 ) ) )( ) (( ) ( z h ( K z t t z q q q

PAGE 50

38 Initial static water content distribution was assumed. Van Genuchten (1980) parameters were applied to estimate the soil hydraulic properties. The residual water content for this soil was 0.02, saturated water content of 0.39, alpha parameter of 0.03 1/cm and n equal to 2.041 (n= 1/(1 m)). The upper boundary condition was variable rainfall rate at the surface, and the lower boundary condition was zero flux at the bottom. Five simulations were made using different depth to the water table. They include 5, 2, 1, 0.5, and 0.1 meters below the ground surface. Time to ponding and time to start of saturation were calculated as well. Table 5 Total runoff for different depths to the water table using HYDRUS 1D Depth to the water table Time to ponding Time to excess saturation Total infiltration Total runoff 5 meters 45.13 min -10.7 cm 2.9 cm 2 meters 45.1 min -10.5 cm 3.1 cm 1 meter 44.4 min -9.96 cm 3.64 cm 50 centimeters 43 min 68.5 min 3.81 cm 9.79 cm 10 centimeters 2.8 min 0.052 cm 13.55 cm The results show that when the depth to the water table increases, the total depth of infiltration increases and the total depth of runoff decreases (Figure 20). Thus, for shallow water table there was more runoff and less infiltration in the soil. This happens in shallow water table where the type of runoff dominating is excess saturation runoff. So the soil pores fill up quickly, and more rainfall becomes runoff. In contrast, for deep

PAGE 51

39 water table, the am ount of water that infiltrate was greater, and there was less runoff. In these water table environments, infiltration excess runoff occurs. Therefore, the soil reaches ponding and infiltration occurs at a smaller rate than the rainfall rate. See Figure 20. Rainfall, Infiltration, and Runoff vs Depth to Water Table 02468 10121416 0100200300400500600 Depth to water table (cm)Total Depth (cm) Total depthof rainfall Total depthof infiltration Total depthof runoff Figure 20 Rainfall, infiltration and runoff vs depth to water table According to these simulations, some of the impacts of runoff as a result of convective or frontal type of rain are described below. Convective rainfall storms tend to have short bursts of high intensity especially at the beginning of the storm. As a result, in shallow water tables, this type of rain can cause the soil to reach the storage capacity of the soil faster thus poten tially producing saturation excess runoff. For soils with deeper water table, runoff due to convective storms can only be produced by infiltration excess. When coarser

PAGE 52

40 time resolution is used, the intensities are smoothed out, surface saturation takes lo nger, and runoff is underestimated. Frontal storms have longer periods of time usually with low rainfall intensities. Rainfall variability for these storms is small and does not significantly impact runoff response. Precipitation is more distributed and uniform over the duration of the storm. When the water table is shallow, a long duration storm might end up filling the storage capacity of the soil resulting in excess saturation runoff. The same can occur for deep water table levels, although it is not very common. 3.3. Impacts to Temporal Variability of Rainfall Under natural conditions, the complex structure of rainfall exhibits a large degree of temporal variability. This variability substantially affects the downward water movement in the vadose zone. Many studies have shown that accuracy of hydrologic models can be improved by using finer time resolution of rainfall data in the determination of infiltration excess (Finnerty et al., 1997). Again, HYDRUS 1D, a model that simulates water movement in th e soil, was used to model the process of infiltration and excess rainfall. HYDRUS 1D solves Richards equation in one dimension, assuming vertical downward water movement on a homogeneous soil. To analyze the impacts of temporal variability of rainfall, a convective and a frontal storm were selected (See Figure 4 and Figure 5 for hyetographs of the storms). The frontal storm with duration of 12.75 hours was used in the numerical simulation with

PAGE 53

41 HYDRUS 1D. Parameters utilized in the simulation included in itial static water content distribution, boundary conditions at the surface accounted for variable rainfall rate, and no flux boundary condition was assumed at the water table, or lower boundary. Van Genuchten parameters were used to model the hydraulic p roperties of the soil. Three scenarios including 1m, 2m, and 5m depths to the water table were tested with simulations lasting 12.75 hours, the duration of the frontal storm. The following table compares the runoff obtained from previous simulations usin g the convective storm shown in Figure 4 with the runoff obtained from the simulations of this frontal storm using 1 minute time steps. The total storm depth for the convective storm was 13.64 cm and for the frontal storm was 10.41 cm. Table 6 Runoff for convective and frontal storm using HYDRUS 1D Depth to the water table Total runoff Convective storm Total runoff Frontal storm % Runoff Convective % Runoff Frontal 5 meters 2.9 0.2 21.3 1.9 2 meters 3.1 0.2 22.7 1.9 1 meter 3 .64 0.3 26.7 2.9 The percentage of runoff was obtained by dividing the total runoff by the total depth of rain for each storm. From the table, the percentage of runoff depth to total storm depth generated from convective storms was much greater than the runoff percentage of frontal storms. This can be explained by the high intensities during the convective storm. Furthermore, when the water table is deeper, there is no percentage difference in runoff

PAGE 54

42 for frontal storms (5 and 2 meters with % runoff equ al to 1.9). Results indicated that convective storms generate more runoff than frontal storms, and the runoff response is more sensitive to temporal variability of rainfall. Appendix E contains the Hydrus 1D graphs for the convective and frontal simulati ons with depth to the water table of 5 m, 2 m, and 1 m. The temporal variability of rainfall, especially for convective storms, cannot be captured if coarser time resolution (e.g. hourly rainfall) is used. Runoff response for simulations of 6 aggregated convective rainfall hyetographs are shown on Table 7. The hyetographs are depicted in Figures 9 to 15. HYDRUS 1D simulations also included four depths to the water table: 0.5, 1, 2, and 5 meters. For Hortonian type of runoff, the time resolution greatly affects runoff response. The error in runoff depth increases when the time resolution increases. For example, for 5 meters depth to the water table, the error in runoff depth is 65.5% using 60 min time steps. In contrast, for a time resolution of 5 min utes, the error is only 6.9%. The next section discusses in more detail how this error also increases when the depth to the water table increases. In Table 7, for a time resolution of 30 minutes, the error in runoff with a depth to the water table of 1 m eter is 36.8% whereas for 5 meters the error is 51.7%. Notice that there are no runoff depth errors for the simulation using shallow soil depths of 0.5 meters. For shallow soil depths, infiltration runoff modeling is not sensitive to the time resolution of rainfall data because runoff is generated by saturation excess and not by infiltration excess.

PAGE 55

43 Table 7 Error in runoff depth for different time steps Time INFILTRATION DEPTH (centimeters) (min) 5 meters 2 meters 1 meter 50 centimeters 1 10.7 10.5 9.96 3.81 3 10.8 10.6 10.1 3.81 5 10.9 10.6 10.1 3.81 10 11.2 11.0 10.5 3.81 15 11.3 11.1 10.5 3.81 30 12.2 11.9 11.3 3.81 60 12.6 12.4 12.0 3.81 Time RUNOFF DEPTH (centimeters) (min) 5 meters 2 meters 1 meter 50 centime ters 1 2.9 3.1 3.64 9.79 3 2.8 3.0 3.5 9.79 5 2.7 3.0 3.5 9.79 10 2.4 2.6 3.1 9.79 15 2.3 2.5 3.1 9.79 30 1.4 1.7 2.3 9.79 60 1.0 1.2 1.6 9.79 Time ERROR IN RUNOFF DEPTH (%) (min) 5 meters 2 meters 1 meter 50 centimeters 1 0.0 0.0 0.0 0.0 3 3 .4 3.2 3.8 0.0 5 6.9 3.2 3.8 0.0 10 17.2 16.1 14.8 0.0 15 20.7 19.4 14.8 0.0 30 51.7 45.2 36.8 0.0 60 65.5 61.3 56.0 0.0 It can be concluded that for fine sand soils with deep water table levels, where Hortonian runoff dominates, the time resolution suitable for infiltration runoff modeling seems to be less than 15 minutes. However, the rainfall time step used should be 5 minutes or less leading to errors of around 10%. Figure 21 depicts the percentage of error in runoff depth estimated by the mode l against time resolution of the rainfall data. The curves for three different water table depths are shown. These curves are very close

PAGE 56

44 % error in runoff depth vs Time resolution 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Time resolution (min) % error in runoff depth 5 meters 2 meters 1 meter which confirms that depth to the water table is not significant when Hortonian runoff is the dominant runoff process. From this graph, the error in runoff estimated increases when time resolution of rainfall increases. Figure 21 Percentage error in runoff depth against time resolution of rainfall Lastly, in hydrologic modeling, high rainfall intensities generated especially by convective storms are smoothed when using large time resolution rainfall data. In consequence, runoff is underestimated. Results show that, where Hortonian runoff is dominant, infiltration and runoff are very sensitive to time resolution. For finer time resolution of rainfall, runoff generated by the model increased.

PAGE 57

45 CHAPTER 4. DIGITAL ELEVATION MODELS (DEM) Digital Elevation Models or DEMs are digital records of terrain elevations in a raster form f or ground positions at regularly spaced horizontal intervals. DEMs produce accurate topographic maps and surface models with grids and contours. They are used in a number of applications in the earth, environmental and engineering sciences, and have prov ed to be an important method for modeling and analysis of spatial topographic information. There are many application domains where DEMs are utilized. Some of these applications include civil engineering, earth sciences analysis, planning and resource man agement, surveying and photogrammetry, global topography data, and military applications. The use of DEMs to extract the geomorphologic attributes of the terrain to develop parameters for hydrologic models has not being widely used. DEMs with both high horizontal and vertical resolution are required to detect and measure critical geomorphic features, such as hillslope angles, slopes, stream channels, etc. The DEMs used here are derived from United States Geological Survey (USGS ). The U.S. Geological Su rvey produces three primary types of digital elevation model data for the United States. They are:

PAGE 58

46 7.5 minute DEM data are produced in 7.5 minute units, which correspond to 1:24,000 and 1:25,000 scale topographic quadrangle map series for all of the Unit ed States and its territories. 7.5 minute DEM data consist of a regular array of elevations referenced horizontally on the Universal Transverse Mercator (UTM) coordinate system of the North American Datum of 1927 (NAD 27). These data are stored as profiles with 30 meter spacing along and between each profile. The vertical accuracy of 7.5 minute DEMs is equal to or better than 15 meters. 30 minute (2 Arc second) DEM data which correspond to the east half or west half of the USGS 30 by 60 minute topograph ic quadrangle map series for the conterminous United States and Hawaii. It covers 30 minute by 30 minute areas where each 30 minute unit is produced and distributed as four 15 by 15 minute cells representing one half of a 1:100,000 scale map. 30 minute DE M data have the same characteristics as the 15 minute DEM data (available for Alaska) except that the spacing of elevations along and between each profile is 2 arc seconds. 1 degree DEM data are produced by the Defense Mapping Agency in 1 degree by 1 degr ee units, which correspond to the east half or west half of USGS 1 by 2 degree topographic quadrangle maps series, for all the United States and its territories. 1 degree DEM data consist of a regular array of elevations referenced horizontally using the geographic (latitude/longitude) coordinate system of the World Geodetic System 1972 Datum. Spacing of the elevations along and between each profile are 3 arc seconds (approximately 90 meters) with 1,201

PAGE 59

47 elevations per profile. The 1 degree DEMs are also referred to as 3 arc second or 1:250,000 scale DEM data. The accuracy of a DEM is dependent upon the level of detail of the source and the spatial resolution, that is the grid spacing used to sample that source. The primary limiting factor for the level of detail of the source is the scale of the source materials. The proper selection of grid spacing determines the level of content that may be extracted from a given source during digitization. The source material for 7.5 minute DEM are 5 feet contour qu adrangle maps. This vertical scale is not appropriate to capture the topographic details needed at the hillslope scale. Therefore, aerial photography with 1 foot contours of the area studied here was used to create the DEM. To accurately describe topogra phic characteristics at the hillslope scale, DEMs with good spatial resolution are needed. 4.1. DEM Data DEMs have proved useful in different areas. Therefore, they are used in this work to develop actual geomorphologic parameters for hydrologic modeling in cluding slope, curvature, landscape convergence, and other topographic attributes. The study area is located Southeast of Tampa at a proposed reservoir site in Hillsborough County, Florida. Doe Creek, one of the tributaries of the Alafia River basin cros ses the study area. Currently, there are no USGS DEMs for the Alafia River basin with a vertical resolution better than 7.5 minute. Hence, a DEM was created for this area.

PAGE 60

48 The Digital Elevation Model was built using a 1 foot aerial photography map from the Southwest Florida Water Management District. DEMs based on 1 foot contours in South Florida can be found for small areas. Usually, they are 30 meters horizontal resolution. The aerial sheet number corresponds to section 2, township 31 S, and range 2 1 E with a scale of 1 = 200. The 1 foot contours for this map were digitized using AutoCAD and imported into Arcview to create the grid file needed for River Tools. The AutoCAD drawing included contour lines, elevation points, and breaklines or lines t hat represent a sharp break in slope. This drawing was imported into Arview GIS to create the grid file. A Triangular Irregular Network (TIN) was created to represent the surface of the map. A TIN partitions a surface into a set of small, contiguous, no n overlapping triangles. An elevation value is recorded for each triangle node. Elevations between triangles can also be interpolated. Figure 22 TIN and contour lines

PAGE 61

49 Figure 22 depicts a surface built using the TIN method. The red lines are the contours and the blue lines are breaklines. The last ones were added to improve accuracy when the TIN was being created, so the triangles were not interpolated within the same contour lines as if it were a flat area. The grid for t he map was then produced using the TIN and specifying the cell size. 5 meters by 5 meters cells were used for the grid. Data files containing the grid were imported into River Tools to visualize and analyze the domain. The AutoCAD drawing with 1 foot co ntour lines is displayed in Figure 23. Breaklines and elevation points are also shown. The colors in the figure indicate different elevations. Light pink represents the uplands or higher elevations and dark red represents lower elevations (streams). F igure 23 1 foot contour map for Lithia near the Tampa Bay Regional Reservoir

PAGE 62

50 4.2. River Tools 2.0 RiverTools 2.0 (Research Systems, 1999), an application for DEMs and river network analysis is used to import, display, analyze, and ex tract information from DEMs data. RiverTools uses an Interactive Data Language for data analysis and visualization that runs on Windows. Shaded relief graphs of the DEMs can be displayed by RiverTools as raster images where every pixel in a DEM is assign ed a color that depends on its elevation and brightness. Other information such as profiles and corresponding flow paths, stream order, contributing area to the stream, basin delineation, slope and curvature, flow distance, etc., can be extracted from Riv erTools. River Tools allows you to import and prepare the input data from different sources. Information such as measurements can be extracted from the input data. Flow grid and river network can be extracted to display and analyze flow paths. Visuali zation of the DEM and display of the spatial properties of the objects contained in the DEM are some of the main tools of this program. Shaded relief of the study areas can be displayed a long with profiles, flow paths, and river channel properties. Stat istical analysis of the many measurements is also available. Elevation values are typically gridded in one of two pixel geometries, fixed length or fixed angle. In the first way, the x size and y size of each grid cell or pixel has some fixed length. For example, the pixels for the DEM used in this work measure 5 meters on a side. In the second way, the x size and y size of each grid cell or pixel has a fixed, angular measure. For example, the pixels in a USGS 1 Degree DEM measure 3 arcseconds on a side. A shaded relief of the DEM used in this study is shown in Figure 24.

PAGE 63

51 Figure 24 Shaded relief of the DEM at Doe Creek near a proposed Reservoir Flow grid was extracted from the DEM study in this work. Upstream areas, downstr eam slope, and downstream curvature were pulled out from the grid. These features are necessary to display the line profiles, flow paths, and reach information so geomorphologic characteristics like slope, curvature, and divergence can be obtained. The c oupling of a Digital Elevation Model with a variable saturated model has not been done before and it has several advantages. Topographic features like slope, curvature, convergence, and divergence can be specified more accurately. Flow paths and their re spective profiles can be easily obtained. These geomorphologic features of the terrain are very important in the extent of variable saturated areas near streams and need to be study in more detail. At the small scale hillslope scale coupling DEMs with rainfall runoff models takes into account the variations of saturated areas due to changes in geomorphologic and hydrologic features.

PAGE 64

52 CHAPTER 5. SATURATION EXCESS RUNOFF Saturation excess runoff occurs when soil saturation from below causes runoff a t the surface. During this process, water infiltrates in the soil and fills the water storage capacity of the soil. Then, the soil gets saturated and all rainfall runs off. This scenario occurs especially in soils with high initial water content, shallow water table, or by accumulated volume of infiltrated water during a storm. If saturation excess occurs, the runoff rate is equal to the rainfall rate (Maidment, 1993) Topographic features such as slope, curvature, and degree of convergence have an im pact on the extent of variable saturated areas that form by saturation excess mechanisms. Most research at the hillslope scale use hypothetical values, and sometimes non realistic values, to represent the hillslope properties needed. In order to capture the topographic variations of Florida environments, digital terrain data should be used. To analysis the conditions explained above, River Tools, a Digital Elevation Model (Research Systems, 1999), is coupled with HYDRUS 2D, a two dimensional variable sat urated unsaturated model that simulates water flow in variably saturated soils. River Tools provides analysis of geomorphologic characteristics of the region such as slope, concave and convex curvature, convergence, and divergence. After extracting these landscape attributes from DEMs, they are incorporated into HYDRUS 2D to describe the upper boundary of physical domain. Sensitivity analysis is done on landscape and

PAGE 65

53 hydrologic parameters. Actual rainfall data was used in the simulations as well with the purpose of looking at the impacts due to temporal variability of rainfall. 5.1. HYDRUS 2D Model HYDRUS 2D is an interactive graphics based user interface that runs in Windows and executes HYDRUS2, which is the Fortran code. It includes SWMS_2D which is a two dimensional finite element model used for the analysis of water flow and solute transport in variable saturated porous media. The model has the same features of Hydrus 1D plus a mesh generator for unstructured finite element grids, MESHGEN_2D, which has the capability of simulating flow on irregular boundaries. HYDRUS 2D was developed by U.S. Salinity Laboratory, U.S. Department of Agriculture, Agriculture Research Service (Simunek and van Genuchten, 1999). In HYDRUS 2D, flow and transport can occur in the horizontal plane, vertical plane, or axisymmetrical vertical region (a three dimensional region exhibiting radial symmetry about the vertical axis). The finite element variable saturated flow model used in this work solves the Richards equation in t wo dimensions shown below Where q (t) is the water content of the soil (c m 3 of water/c m 3 of soil), h( q ) is the matric potential head, z is the depth of the soil, x is the distance from the stream, K ( q ) is the unsaturated hydraulic conductivity, and t is the time. dd d d h (q) d d d d = d dq t ( K x z h q ( K z t q ) ( ) ) ( 1 ) ) )( )( q ( )(

PAGE 66

54 A vertical plane was us ed for most of the simulation except for runs involving convergence and divergence. Analysis of convergence was achieved by using axisymmetrical vertical flow. Several boundary conditions can be described with HYDRUS 2D including prescribed and variable pressure head, prescribed and variable flux, no flux, free drainage, seepage and atmospheric conditions. A combination of polylines, arcs, circles, and splines define the specifications of the flow region. Domain geometry as well as initial conditions ca n also be imported from files (Simunek and van Genuchten, 1999). Some of the applications of HYDRUS 2D include irrigation management, seasonal simulation of water flow and plant response, deep percolation, seepage in highway design, lake basin recharge, g roundwater aquifer and stream interaction, and environmental impacts of drawdown of shallow water tables. Input preparation for HYDRUS 2D includes soil hydraulic parameters, initial and boundary conditions, and mesh generation. The code allows users to s elect three types of models to describe the soil hydraulic properties: van Genuchten, Brooks and Corey, and modified van Genuchten type equations. In this study, soil properties are chosen from the Characterization Data for Selected Florida Soil (Carlisle et al., 1989) and van Genuchten parameters are fitted. Boundary conditions at the surface were defined as atmospheric boundary to simulate rainfall falling uniformly over the domain during rainfall, and seepage face to allow water to go out of the do main surface as saturation excess after rainfall stops. At the stream, the boundary condition was constant pressure head equal to

PAGE 67

55 the level of the water table at the stream. All other boundaries were no flow boundaries. Initial conditions were specified pressure head corresponding to an equilibrium profile. Initial water table depth for the reference simulation was sloping. The water table was assumed curved down to the stream and was almost parallel to the ground surface. 5.2. Reference Simulation A refe rence simulation designed with field data was used as the starting point for all simulations and from which the parameters were changed one at a time to analyze their sensitivity. The reference hillslope topography was taken from a DEM created from a 1 fo ot aerial photography map of Southwest Florida Water Management District. The Digital Elevation Model from Lithia, at Doe Creek is located in the southeastern portion of Hillsborough County, Florida. The reference simulation consisted of a small stream o n Doe Creek with an average hillslope of 1.1 %. The ground surface was a mix slope conformed by a concave downslope with a convex upslope. Several profiles were taken from the DEM on that specific site, and one with an average slope and overland flow pla ne was chosen. The hillslope profile had an overland flow plane of 387 meters (1271 ft). The stage at the stream was 0.6096 meters (2 ft). Incision depth in the stream was 1.07 m (3.5 ft), and the depth of the soil from the streambed was 2.1336 meters (7 ft). Figure 25 Reference simulation profile at the reservoir site, Doe Creek

PAGE 68

56 Figure 25 shows the profile used for the reference simulation with the domain measurements. This profile is vertically exaggerated for better vie w and it is not to actual scale. The actual stream was intermittent and the regular depth of the water level at the stream varied around 2 feet or more during summer. Initial water table depth on the profile was not flat. It curved down to the stream st age, which was 2 ft, and follows a path almost parallel to the ground surface as illustrated in Figure 26. The colors in Figure 26 represent the pressure head distribution in the profiles for the initial conditions. Notice that the line at which the pres sure head is zero (line inside profile) corresponds to the water table level. Figure 26 Initial water table level for the cross section A van Genuchten (1980) soil hydraulic model was utilized with a Myakka fine sand (layer 3 ) soil type, predominant at the reservoir site (Carlisle, 1989). This hydraulic model was used because it explained the total water retention curve, including water retention characteristic at the air entry pressure. Myakka fine sand soil parameters were obtained from the Characterization Data for Selected Florida Soils, Soil Survey (1989). These parameters included a saturated hydraulic conductivity of 0.128 m/hr. The van

PAGE 69

57 Genuchten parameters alpha ( a ) and n ( n = 1/ (1 m )) were fitted to the actual wat er retention curve of Myakka fine sand to use these parameters in the model. Appendix D shows a copy of these parameters obtained from the Soil Survey. Figure 27 depicts the Myakka fine sand water retention curve and the fitted van Genuchten curve. The saturated water content was 0.479 and the residual water content was 0.020. The van Genuchten curve was fitted using a spreadsheet, and alpha ( a ) was found equal to 0.105, m equal to 0.205, and n equal to 1.26. 0.0000.1000.2000.3000.4000.500 050100150200 250300 350 Capillary Pressure in cm of water Water Content in cm3 of water/cm3 of soil Measured watercontent Van Genuchten Figure 27 Van Genuchten fitted water retention curve For comparison, the water retention curve calculated by HYDRUS 2D using the van Genuchten soil hydraulic model for the fitted parameters is illustrated in Figure 28. The water retention c haracteristic curve describes the ability of the soil to release and store water. It is defined as the relationship between water content and matric potential ( Maidment, 1993) Notice that this type of fine sand does not have a capillary fringe or it is very small, which means that the water drains quickly in the soil. For the simulations,

PAGE 70

58 homogeneous soil with vertical flow was assumed. Hysteresis was not taken into account. 0.0 0.1 0.2 0.3 0.4 0.5 0 0.5 1 1.5 2 2.5 3 3.5 |h| [m] Figure 28 Water content vs head obtained by HYD RUS 2D The rainfall rate was constant for 2.25 hours with an intensity of 0.013 m/hr. Rainfall intensity and duration were obtained from actual rainfall data taken from the roof of the Engineering Building at the University of South Florida. The analysi s was based on convective storms since most of the rainfall in Florida is owing to convective activity. The mean duration for convective storms was 1.16 hours and the mean rainfall depth was 0.51 inches. Small storms created by the rainfall model with to tal depth less than 0.1 inches were ignored. The standard deviation for storm duration was 1.09 hours and for rainfall depth was 0.64 inches. The standard deviation was added to the mean depth and duration to increase the total rainfall depth and produce some runoff. As a result, the average rainfall depth and duration (mean plus standard deviation) used in this work were

PAGE 71

59 1.15 inches and 2.25 hours respectively. The rainfall rate was considerably small compared to the saturated hydraulic conductivity, t hus Hortonian runoff did not occur. HYDRUS 2D is a finite element model that has the ability to generate a mesh of irregular elements improving discretization for the domain. The model was discretized into a finite element mesh of irregular elements, w here boundary conditions, initial conditions, and parameter values discussed previously were assigned. The resolution of the mesh closer to the stream was finer to better approximate the fluxes moving across this area, to capture the irregularities of the ground surface, and to have a more accurate solution (See Figure 29). The mesh was generated using the automatic features of the model. For the reference simulation, the mesh had a total of 3427 mesh points with 6553 mesh triangles, which were created o f irregular sizes and shapes. Figure 29 2 D mesh for cross section at the reservoir site Boundary conditions on the domain included no flow boundary condition at the bottom of the profile and on the sides of the domain. The left side corresponds to the boundary below the streambed and the right side corresponds to the water divide at the upslope. Therefore the assumption of no flow at these boundaries is applicable. At the stream or water surface, constant pressure head bou ndary condition was applied with

PAGE 72

60 linear distribution in the pressure head. Along the land surface, atmospheric boundary conditions were used during rainfall. Atmospheric boundary conditions account for precipitation, evaporation, transpiration, and drain age fluxes. These fluxes are assumed uniformly distributed along the domain. In this analysis, precipitation was the only flux considered. After rainfall stops, seepage boundary condition at the ground surface was used. Under this condition, water is a llowed to exfiltrate (leave) the domain at the surface and run off towards the stream. Figure 30 illustrates the boundary conditions used during rainfall. When saturation at the surface was reached, HYDRUS 2D changed the boundary conditions at the surfac e domain to prescribed pressure head equal to zero. Figure 30 Boundary conditions for the cross section domain The simulations were divided into two runs. The first run was during the rainfall period with atmospheric boundar y conditions at the surface. The second run used the output of the first run as the initial conditions in the pressure head and had a seepage face surface boundary condition, so water came out of the domain as saturation excess proceeded. HYDRUS 2D graph s including the potential atmospheric, actual atmospheric, subsurface, and seepage fluxes were utilized for the sensitivity analysis. Since runoff can

PAGE 73

61 be produced by saturation excess runoff or subsurface storm runoff, different graphs were used to estima te these fluxes. Saturation excess runoff was calculated on the first run of the simulations as the difference between the potential and actual atmospheric fluxes (total rainfall minus infiltration). The subsurface graphs were used to calculate the later al subsurface flux going to the stream. The seepage graphs helped in calculating the amount of water going into the stream as saturation excess after the storm stops on the second run of the simulations. The 2 dimensional profiles displayed by HYDRUS 2D were used to estimate the extent and shrinkage of variable saturated areas, as well as the water table position with time. HYDRUS 2D graphs for the reference simulation are illustrated in Appendix F. These graphics include subsurface and seepage fluxes d uring and after rainfall. Extent of VSA is also shown in this appendix. After defining the reference simulation, some hydrologic properties and geomorphologic features were tested to analyze their sensitivity to saturation excess runoff. Following are t he parameters that were modified here and their ranges of variation. 5.3. Hydrologic Characteristics Hydrologic characteristics that can impact the extent of variable saturated areas are soil parameters, rainfall variability, and initial water table depth. A sensitivity analysis was performed in order to define the effects in VSA as a result of a change in these characteristics.

PAGE 74

62 Soil parameters: Several types of soil were used: Myakka fine sand, Mulat fine sand, and Millhopper fine sand were taken from the C haracterization Data for Selected Florida Soils, Soil Survey (Carlisle, 1989). Other soils used were sand, sandy loam, and loamy sand taken from the HYDRUS 2D soil catalog for the USDA texture soil classification. Saturated hydraulic conductivity, Ks, ra nged from 0.044 m/hr to 0.297 m/hr. The van Genuchten parameters ? (alpha) and n along with Ks were varied. Water retention curves with fitted van Genuchten parameters for actual soil data including Myakka fine sand, Mulat fine sand, and Millhopper fine sand can be found in Appendix G. Rainfall rate: Constant rainfall rate and variable rainfall rate obtained from actual rainfall records were used in the simulations. Rainfall rate was less than the saturated hydraulic conductivity in order to avoid inf iltration excess runoff. A series of storms, three convective and two frontal, were used in the model. The hyetographs for these storms are added to the Appendix H. Initial conditions for the water table: Wet and dry soils given by the depth to the wat er table level were sampled. For these simulations, the initial water table was assumed flat. Dry soils started with a dry stream whereas wet soils were assumed to be fully saturated with the water table at the ground surface on the downslope profile. S hallow water table depths ranging from 1 foot above the streambed to 4 feet above the streambed were analyzed to account for dry and wet season respectively. Appendix I contains the profiles for these water table depths.

PAGE 75

63 5.4. Geomorphologic Characteristics Nat ural topography extracted from Digital Elevation Models was coupled with the variable saturated model, HYDRUS 2D, to address the changes in the extent of variable saturated areas due to geomorphologic features including slope, curvature, and divergence. T he topographic features analyzed here included flat and steep slopes, convex and concave curvature, and landscape divergence. Flat initial water table at 3.5 feet above the streambed was assumed throughout these simulations. Some of the simulations used average values of curvature in order to see the response when small changes were applied. Straight slope: refers to flat and steep slopes with zero curvature. Profiles taken from DEMs showed that hillslopes slopes are less that 2.5% average. Therefore, straight slopes were changed from 0.5% to 2.5%. Figure 31 illustrates the three slopes utilized. Figure 31 Profile for 0.5%, 1.1%, and 2.5% slopes Curvature: refers to the degree of curvature of the ground surface boundary. Curvature can be positive or negative for convex and concave profiles,

PAGE 76

64 respectively. The following graphs show the three different radii evaluated for each curvature (Figure 32 and Figure 33). Figure 32 Concave profiles for three negative radius of curvature The degree of curvature is attained by the radius of curvature. Slightly changes in the radius of curvature for concave slopes were enough to demonstrate its impact in runoff generation. The radii for concave slopes we re 14871.1 m, 17148.7 m, and 19474 m. The radiuses for convex slopes were 35349.4 m, 15928.8 m, and 11779.1 m. Figure 33 Convex profiles for three negative radius of curvature Landscape divergence is described by the ratio of downslope width to upslope width using the axisymmetrical vertical flow option in HYDRUS 2D. Flow through divergent hillslopes was simulated as radial flow going away from the symmetric axis with the axisymmetrical axes coinciding with the left side of t he

PAGE 77

65 domain. For these simulations, the domain was reverted to have the stream at the right side. A discharge area with divergent flow is shown in Figure 34. Figure 34 Divergence parameter for HYDRUS 2D From Figure 34, assuming concentric radius we have 2 21 12211 S S r = r S r q S r == qWhere r 2 is the radius from the center to the stream, r 1 is the radius from the center to the water divide of the profile, S 2 is the length of the stream (arc length), and S 1 is the length of the water divi de. The degree of convergence and divergence is defined as the ratio of r 2/r1 or S 2/ S1. If this ratio is greater than one, the flow in the discharge area diverges. On the other hand, if the ratio is smaller than one, the flow in the discharge area conver ges. 5.5. Simulation Results Different types of soils predominant in the area were used. Figure 35 displays the water retention curve obtained from actual data for the soils used in these simulations. Notice that even for fine sand, the water retention curve differs significantly (Myakka fine sand vs Mulat fine sand). Table 8 shows the extent of VSA and values of the fluxes

PAGE 78

66 when soil parameters n q and Ks were changed. All fluxes were compared with the reference simulation for Myakka fine sand (layer 3). From these simulations subsurface runoff and saturation excess were seen to be very sensitive to the van Genuchten paramete n and the saturated hydraulic conductivity, Ks. For the same type of soil, when n increased, saturation excess decreased. In this case, subsurface storm flow also decreased but it was not very significant. On the other hand, when the saturated hydraulic conductivity increased, subsurface storm flow increased considerably whereas saturation excess runoff was not affected. 0 50 100150200250300350 0.0000.1000.2000.3000.4000.500 Water Content in cm3 of water/cm3 of soil Capillary Pressure in cm of water Myakka fine sand Millhopper fine sand Mulat fine sand Sandy loam Loamy sand Sand Figure 35 Water retention curves for different soils used in simulations Sandy loam soils, with very low hydraulic conductivity for example, produced less percentage of total subsurface flux. In Table 8 obser ve that combinations of these parameters produced saturation excess runoff only when the parameter n was decreased. Thus, VSA extended largely for these cases and seepage was observed. Also, the amount of water running as saturation excess is substantial ly smaller compared to the amount of subsurface flow. The values highlighted on Table 8 are parameters that were changed for these particular simulations.

PAGE 79

Soil Properties Fluxes (meters2) Types of soil used q r q s a n Ks Extent of VSA (m) Qsur1 Qsub1 Qsur2 Qsub2 Tsur (%) Tsub (%) Maykka Fine Sand (layer 3). Reference simulation 0.02 0.479 9.5238 1.26 0.128 35.5 0.1 0.0374 0.0356 1.5 1.2 13.605 0.02 0.479 9.5238 2.28 0.128 0 0 0.0303 0 1.36 0 12.304 Mulat Fine Sand (layer 4) 0.04 0.427 33.33 2.22 0.263 0 0 0.0604 0 2.44 0 22.127 Millhopper Fine Sand (layer 1) 0.04 0.442 26.316 2 0.105 0 0 0.0267 0 1.09 0 9.8823 Sand 0.045 0.43 14.5 2.68 0.297 0 0 0.0675 0.0389 2.87 0.344248 25.996 0.045 0.43 9.5238 2.68 0.297 0 0 0.0694 0 2.92 0 26.455 0.045 0.43 14.5 1.26 0.297 33.2 0.1 0.0802 0.0389 3.04 1.229204 27.612 0.045 0.43 14.5 2.68 0.128 0 0 0.0318 0 1.34 0 12.14 Sandy loam 0.065 0.41 7.5 1.89 0.0442 31.6 0.05 0.0123 0.00384 0.603 0.47646 5.4451 0.065 0.41 12.4 1.89 0.0442 0.01 0.0119 0.000527 0.577 0.181655 5.2115 0.065 0.41 7.5 2.28 0.0442 0.01 0.0118 0.00109 0.581 0.186637 5.246 0.065 0.41 7.5 1.89 0.146 0.01 0.0411 0.00697 1.64 0.238673 14.877 Loamy sand 0.057 0.41 12.4 2.28 0.146 0 0 0.0359 0 1.56 0 14.123 0.057 0.41 9.5238 2.28 0.146 0.01 0.0365 .000094 0.0159 0.089325 0.4637 0.057 0.41 12.4 1.26 0.146 36.4 0.12 0.0457 0.151 2.08 2.39823 18.812 0.057 0.41 12.4 2.28 0.297 0 0 0.0692 0 2.93 0 26.542 q r Residual water content q s Saturated w a ter content a alpha n n Ks Saturated hydraulic conductivity Qsur1 Cumulative saturation excess runoff during rainfall Qsub1 Cumulative subsurface flux during rainfall Qsur2 Cumulative saturation exc ess runoff and seepage after rainfall Qsub2 Cumulative subsurface flux after rainfall Tsur (%) Total percetage of flux going as saturation excess Tsub (%) Total percetage of flux going as subsurface runoff Table 8 Soil Parameters changes in a, n, and Ks 67

PAGE 80

68 Constant rainfall rat e was used for the reference simulation. The storm had duration of 2.25 hours and constant intensity of 0.013 m/hr. To analyze the impacts of temporal variability, actual storms with variable rainfall intensities were tested. The storms used here were t hree convective storms with the same duration and total rainfall depth as the reference simulation, and two frontal storms with the same total depth as the reference simulation but durations of 5.62 and 3.8 hours. Table 9 Changes in runoff cumulative fluxes because of variable rainfall intensity Fluxes (meters 2 ) Rainfall variability Extent of VSA (m) Qsur1 Qsub1 Qsur2 Qsub2 Tsur (%) Tsub (%) Reference simulation, I = 0.013 m/hr, d = 2.25 hr 35.5 0.1 0.037 0.036 1.500 1.200 13 .605 Convective storm with high intensities at the beginning 35 0.1 0.043 0.126 2.010 2.000 18.170 Convective storm with high intensities in the middle 37 0.1 0.035 0.043 1.510 1.268 13.675 Convective storm with high intensities at the end 34 0.1 0.036 0.038 1.500 1.219 13.589 Frontal storm, d = 5.62 hr 33 0 0.100 0.040 1.490 0.350 14.070 Frontal storm, d = 3.8 hr 35.2 0.000 0.068 0.036 1.500 0.315 13.879 During rainfall, the VSA expanded. Then, when rainfall stopped the variable saturated areas shr unk. From Table 9, the extent of VSA changed slightly. Frontal storms did not produce saturation excess during rainfall, but there was some seepage after rainfall stopped. Convective storms with high intensity at the beginning of the storm produced more seepage and saturation excess after rainfall stopped. The fluxes for all other storms were relatively similar. At the small scale, temporal variability of rainfall seemed not to impact substantially the infiltration runoff process near the streams with shallow water tables. Appendix H contains the hyetographs for these storms.

PAGE 81

69 Different depths to the water table were simulated. Water table levels used were 1, 3, 3.5, and 4 feet above the streambed. The reference simulation had an initial flat water ta ble level at 3.5 feet from the streambed. For dry initial conditions, 1 foot water table depth was used. The water table was later increased to 4 feet where part of the soil surface was already saturated. Variable saturated areas expanded when the initi al water table depth was closer to the surface. A shallow water table depth of 4 feet above the streambed produced an extent in the VSA of 20.7 m, whereas a deep water table depth of 1 foot above the streambed did not produce saturation excess runoff. Ta ble 10 Runoff cumulative fluxes due to changes in the initial water table depth Fluxes (meters 2 ) Depth to the water table Extent of VSA (m) Qsur1 Qsub1 Qsur2 Qsub2 Tsur (%) Tsub (%) Reference simulation, water table at 3.5 f eet 19.4 0.200 0.014 0.000 0.261 1.770 2.434 Water table at 1 foot from streambed 0 0.000 0.009 0.000 0.155 0.000 1.449 Water table at 3 feet from streambed 9.5 0.000 0.057 0.000 0.363 0.000 3.719 Water table at 4 feet from streambed 20.7 0.300 0.009 0. 002 0.203 2.669 1.874 Figure 36 Pressure head distribution for the reference simulation

PAGE 82

70 Figure 36 illustrates the pressure head distribution in the reference simulation right after rainfall stops. The line that intercepts t he ground surface near the stream is equal to the zero pressure line, which represents the water table depth at 2.25 hours. Geomorphologic features like slope, curvature, and divergence were studied here. Profiles with slopes of 0.5%, 1.1%, and 2.5% were simulated. Initial water table level at 3.5 ft from the streambed was used for all simulations. Table 11 shows an increase in the size of VSA when slope of the cross section was increased. Table 11 Runoff cumulative fluxes f or different slopes in the profiles Fluxes (meters 2 ) Slopes with flat initial water table Extent of VSA (m) Qsur1 Qsub1 Qsur2 Qsub2 Tsur (%) Tsub (%) Reference simulation 19.6 0.2 0.0140 0.00159 0.261 1.784 2.4336 Straight slope of 0.5% 73.7 0.2 0.0 052 0.0199 0.154 1.946 1.4086 Straight slope of 1.1% 30.1 0.3 0.0053 0.0133 0.22 2.773 1.9942 Straight slope of 2.5% 12.8 0.1 0.0176 0.00279 0.279 0.910 2.6248 For instance, decreasing the slope of the profile to 0.5% caused a substantially increase of the variable saturated areas which expand up to 73.7 meters. Other fluxes did not have major changes. The degree of curvature was analyzed by taking different curved profiles representative of real topographic terrains. Three convective profiles and th ree concave profiles were studied in this work. Flat initial water table level at 3.5 ft feet above the streambed was assumed for the simulations of changes in curvature. Table 12

PAGE 83

71 summarizes the cumulative runoff fluxes obtained after changing the curvat ure of the profile. Notice that variable saturated areas were greater for concave slope profiles than convex slopes. Small changes in the degree of curvature for concave slopes produced significant extent of VSA and saturation excess runoff during and af ter rainfall. Table 12 Runoff cumulative fluxes for different curvatures Fluxes (meters 2 ) Types of curvature Extent of VSA (m) Qsur1 Qsub1 Qsur2 Qsub2 Tsur (%) Tsub (%) Reference simulation 19.6 0.2 0.014 0.002 0.261 1.784 2.434 Concave slope with r = 14871.1 meters 134.9 1.58 0.001 0.295 0.086 16.593 0.776 Concave slope with r = 19474 meters 94.02 1.25 0.0015 0.106 0.049 11.998 0.437 Concave slope with r = 28631.9 meters 63.89 0.3 0.002 0.076 0.107 2.446 0.966 Convex slope with r = 11779.1 meters 20.7 0.1 0.012 0.006 0.283 0.942 2.610 Convex slope with r = 15928.8 meters 14.5 0.1 0.015 0.005 0.292 0.925 2.720 Convex slope with r = 35349.4 meters 12 0.1 0.018 0.003 0.285 0.911 2.683 Slight changes occurred in th e saturation excess runoff generated by the model when convex slopes where changed. On the other hand, subsurface fluxes for convex curvature were higher than subsurface fluxes for concave curvature. From Table 12, it can be observed that the percentage of total saturation excess produced by concave curvature was greater that the percentage of total subsurface flux. Saturation excess runoff dominates in concave curvature profiles. Appendix N shows the extent of VSA on these concave profiles.

PAGE 84

72 Changes in the landscape such as convergence and divergence were also considered in this work. Divergence and convergence were assessed as the relationship between the radii of two concentric circles that represent the left (stream) and right (water divide) boundar ies of a cross section taken from a flow path. Figure 37 depicts different flow paths that converge and diverge. This analysis was based on a discharge area bounded by two flow paths that either converge or diverge towards the stream. Figure 37 Shaded relief of the Lithia DEM with several flow paths Simulations using divergence showed that when the radius of divergence increased, the area of possible saturation also increased. Therefore, the VSA extended farther from the st ream. For this DEM, most of the flow paths diverged towards the stream. Few flow paths showed convergence, which was not expected for the type of topography predominant in Florida.

PAGE 85

73 CHAPTER 6. CONCLUSIONS Infiltration excess, also called Hortonia n Runoff, and saturation excess are the runoff mechanisms predominant in Florida. These mechanisms are highly influenced by geomorphologic and hydrologic characteristics. Temporal variability of rainfall has been shown to impact the infiltration excess runoff process whereas hillslope topography, soil parameters, and depth to the water table have a greater impact on runoff generated by saturation excess mechanisms. When Hortonian runoff is dominant, results showed that infiltration and runoff were very sensitive to time resolution, especially for convective storms. Actual rainfall data with fine time resolution was used. Statistical analysis of rainfall showed that details in rainfall temporal distribution are lost when coarse time resolution was used. Consequently, high rainfall intensities generated by convective storms were smoothed out, and Hortonian runoff was underestimated. The error in runoff response increased when coarser resolution was used. Fine time resolution may, therefore, be used in e xisting hydrologic models to accurately simulate infiltration excess runoff. The time resolution suitable for infiltration runoff modeling on this fine sandy soil should be less than 5 minutes leading to manageable errors of around 10%.

PAGE 86

74 Runoff generated by saturation excess is more common in shallow water table environments, especially close to wetlands and streams. Under these conditions, Variable Saturated Areas (VSA) are present. VSA at the hillslope scale were assessed by coupling a Digital Elevation Model ( DEM) with the variable saturated water flow model, HYDRUS 2D. Geomorphologic characteristics of the hillslope including slope, curvature, and divergence were extracted from DEMs. Sensitivity of the model to these characteristics was attained with HYDRUS 2D, by running different simulations to generate overland flow. From these simulations, subsurface storm flow was very significant especially for hillslopes with relatively high saturated hydraulic conductivity soils like fine sands. On the other hand, hillslopes with flatter slopes and concave curvature showed more saturation excess runoff. The most important soil hydraulic parameter influencing runoff was found to be the saturated hydraulic conductivity and van Genuchten parameter n Subsurfac e flux increased when saturated hydraulic conductivity increased, and saturation excess increased when n decreased. Rainfall temporal variability did not influence the magnitude of Variable Saturated Areas. Depth to the water table, in contrast, generate d more saturation excess than subsurface flux when the water tables were high. Further verification and especially a field study to observe and measure these theoretical model results have not yet been made. There is a need for actual rainfall data wit h high temporal resolution, digital elevation data with high vertical resolution, and soil data for actual soils to be able to verify these results.

PAGE 87

75 REFERENCES Agnese, C., and V. Bagarello. 1997. Describing Rate Variability of Storms Events for I nfiltration Prediction. Transactions of the ASAE, Volume 40, No. 1, pp 61 70. Arnold, J. G. and J. R. Williams. 1989. Stochastic Generation of Internal Storm Structure at a Point. Transactions of the ASAE, Volume 32, No. 1, pp 161 167. Bosch, D. D., an d F. M. Davis. 1999. Rainfall Variability and Spatial Patterns for the Southeast. Proceedings of the Fourth International Conference on Precision Agriculture, pp 535 545. Brooks, R. H., and A. T. Corey. 1964. Hydraulic Properties of Porous Media. Hyd rology Paper 3, Colorado State University, Fort Collins, Colorado. Carlisle, V. W., F. Sodek, M. E. Collins, L. C. Hammond, and W. G. Harris. 1989. Characterization Data for Selected Florida Soils, Soil Survey. U.S. Department of Agriculture, Soil Cons ervation Service. Soil science research report number 89 1. Florida, pp 307. Dunne, T. and R. Black. 1970a. An Experimental Investigation of Runoff Production in Permeable Soils. Water Resources Research, Volume 6, No. 2, pp 478 490. Dunne, T. and R. Black. 1970b. Partial Area Contributions to Storm Runoff in a Small New England Watershed. Water Resources Research, Volume 6, No. 5, pp 1296 1311. Dunne, T. 1978. Field Studies of Hillslope Flow Processes. Hillslope Hydrology ed. M. J. Kirkby. Wile y Interscience, New York, pp 227 293.

PAGE 88

76 Evans, M., T. Burt, J. Holden, and J. Adamson. 1999. Runoff Generation and Water Table Fluctuations in Blanket peat: Evidence from UK Data Spanning the Dry Summer of 1995. Journal of Hydrology, Volume 221, pp 141 1 60. Finnerty, Bryce, M. B. Smith, D. Seo, V. Koren, and G. E. Moglen. 1997. Space Time Scale Sensitivity of the Sacramento Model to Radar Gage Precipitation Inputs. Journal of Hydrology, Volume 203, pp 21 38. Flanagan, D.C, G. R. Foster, and W.C. moden g. 1988. Storm Pattern Effects on Infiltration, Runoff, and Erosion. Transactions of the ASAE, Volume 31, No. 2, pp 414 420. Freeze, R. 1972b. Role of Subsurface Flow in Generating Surface Runoff 1. Base Flow Contributions to Channel Flow. Water Resou rces Research, Volume 8, No. 3, pp 609 623. Freeze, R. 1972a. Role of Subsurface Flow in Generating Surface Runoff 2. Upstream Source Areas. Water Resources Research, Volume 8, No. 5, pp 1272 1283. Freeze, R. A. 1980. A Stochastic Conceptual Analysis of Rainfall Runoff Process on a Hillslope. Water Resources Research, Volume 16, No. 2, pp 395. Gammeter, S. Burckhardt, and R. Fankhauser. 1998. Analysis of Rainfall Time Series with Regard to Temporal Disaggregation for the Use in Urban Hydrology. W ater Science Technology, Volume 37, No. 11, pp 65 72. Govindaraju, R. and M. Kavvas. 1991. Dynamics of Moving Boundary Overland Flows Over Infiltrating Surfaces at Hillslope. Water Resources Research, Volume 27, No. 8, pp 1885 1898. Hewlett. 1961. Som e Ideas about Storm Runoff and Baseflow. In: Process Studies in Hillslope Hydrology, ed. M. G. Anderson and T.P. Burt. John Wiley & Sons, England, pp 314. Hewlett, J. D. and A. Hibbert. 1963. Moisture and Energy Conditions on a Sloping Soil Mass Durin g Drainage. Journal of Geophysical Research, Volume 68, pp 1081 1087.

PAGE 89

77 Kirkby, M. J. 1978. Hillslope Hydrology Wiley Interscience, New York, pp 348. Kirkby, M. J. and R. Chorley. 1967. Throughflow, Overland Flow and Erosion. In: Process Studies in Hillslope Hydrology. M. G. Anderson and T.P. Burt. John Wiley & Sons, England, pp 175. Lamb, Robert. 1999. Calibration of a Conceptual Rainfall Runoff Model for Flood Frequency Estimation by Continuous Simulation. Water Resources Research, Volume 35, No. 10, pp 3103 3114. Loague, Keith. 1988. Impact of Rainfall and Soil Hydraulic Property Information on Runoff Predictions at the Hillslope Scale. Water Resources Research, Volume 24, No. 9, pp 1501 1510. Maidment, David R. 1993. Handbook of Hydro logy McGraw Hill, Inc. New York, NY. Milly, P. C. D., and P. S. Eagleson. 1988. Effect of Storm Scale on Surface Runoff Volume. Water Resources Research, Volume 24, No.4, pp 620 624. Ogden, Fred and Bahram Saghafian. 1997. Green and Ampt Infiltra tion with Redistribution. Journal of Irrigation and Drainage Engineering, Volume 123, No. 5, pp 386 393. Ogden, F. and B. Watts. 2000. Saturated Area Formation on Nonconvergent Hillslope Topography with Shallow Soils: A Numerical Investigation. Water Resources Research, Volume 36, No. 7, pp 1795 1804. OLoughlin, E. M. 1986. Prediction of Surface Saturation Zones in Natural Catchments by Topographic Analysis. Water Resources Research, Volume 22, No. 5, pp 794 804. Philip, J. R. 1996. Hillslope In filtration: Divergent and Convergent Slopes. Water Resources Research, Volume 27, No. 6, pp 1035 1040. Research Systems. 1999. River Tools 2.0 Users Guide. Research Systems, Interactive Data Language 5.2. ( http://www.rsinc.com/rivertools ).

PAGE 90

78 Rodriguez Iturbe, Ignacio and P. S. Eagleson. 1987. Mathematical Model of Rainstorm Events in Space and Time. Water Resources Research, Volume 23, No. 1, pp 180 190. Salvucci, G. and D. Entekhabi. 1995. Hillsl ope and Climatic Controls on Hydrologic Fluxes. Water Resources Research, Volume 31, No. 7, pp 1725 1739. Sepulveda, Nicasio. 1997. Application of Two Direct Runoff Prediction Methods in Puerto Rico. ASCE Journal of Hydrologic Engineering, Volume 2, No .1, pp 10 17. Simunek, J., M. Sejna, and M. Th. van Genuchten. 1998. The HYDRUS 1D Software Package for Simulating the One Dimensional Movement of Water, Heat, and Multiple Solutes in Variably Saturated Media. Version 2.0. US Salinity Laboratory, ARS/US DA. Riverside, California. Simunek, J., M. Sejna, and M. Th. van Genuchten. 1999. HYDRUS 2D, Simulating Water Flow, Heat, and Solute Transport in Two Dimensional Variably Saturated Media. Version 2.0. US Salinity Laboratory, ARS/USDA. Riverside, Califo rnia and International Ground Water Modeling Center, IGWMC TPS 53. Colorado School of Mines, Golden, Colorado. Singh, V. P. 1997. Effect of Spatial and Temporal Variability in Rainfall and Watershed Characteristics on Stream Flow Hydrograph. Transact ions of the ASAE, Volume 31, No. 2, pp 414 420. Stagnitti, F., J. Parlange, T. S. Steenhuis, M. B. Parlange, and C. W. Rose. 1992. A Mathematical Model of Hillslope and Watershed Discharge. Water Resources Research, Volume 28, No. 8, pp 2111 2122. Stee nhuis, T. S., M. Winchell, J. Rossing, J. A. Zollweg, and M. F. Walter. 1995. SCS Runoff Equation Revisited for Variable Source Runoff Areas. Journal of Irrigation and Drainage Engineering, Volume 121, No. 3, pp 234 238. Van Genuchten, M. Th. 1980. A Closed Form Equation for Predicting the Hydraulic Conductivity of Unsaturated Soils. Soil Science Society American Journal, Volume 44, pp 892 898.

PAGE 91

79 Wigmosta, M., L. Vail, and D. Lettenmaier. 1994. A Distributed Hydrology Vegetation Model for Complex Te rrain. Water Resources Research, Volume 30, No. 6, pp 1665 1679. Woolhiser, D. A., and D. C. Goodrich. 1988. Effect of Storm Rainfall Intensity Patterns on Surface Runoff. Journal of Hydrology, Volume 102, pp 335 354.

PAGE 92

80 APPENDICES

PAGE 93

81 Appendix A. Subroutine Rainfall_Minute 'Program: Discretization of Rainfall data every minute using Linear Interpolation and average 'Form name: frmrainfall minute 'Programmer: Tatiana X. Hernandez. 'Date: June 30 1999. Last Modified: May 27, 2000 'Purpose: This program is design to convert the continuous rainfall data in rainfall with a 1 minute resolution. The program reads date, time, and time to tip in seconds fr om a text file(.csv) and calculates the intensity of rain every minute. It interpolates when the data is missing and average when there is more than one point in a minute. It creates an output file (.txt) Option E xplicit Dim T(50000) As Single 'cumulative time in minute Dim Ttip(50000) As Single 'time from last tip in seconds Dim datedata(50000) As String 'date for each tip Dim timedata(50000) As String 'time for each tip Dim n As Integer 'total number of records Private Sub cmdinputfile_Click() 'Input a file using a dialog box. Read values of time to tip, date and time from a text file and calculate cumulative time in minutes. Dim i As Integer 'counter for tim e to tip Dim c As Integer 'counter for time to tip when they are longer than 5 minutes Dim filenum As Integer filenum = FreeFile Dialogfile.Filter = "all files(*.*)|*.*|comma delimited files(*.csv)|*.csv|text files(*.txt)|*.txt|" Dialogfile.FilterIndex = 2 Dialogfile.Action = 1 Dialogfile.CancelError = True Open Dialogfile.filename For Input As filenum i = 0 'Calculate cumulative time in minutes While Not EOF(1) i = i + 1 Input #filenum, datedata(i), timedata(i), Ttip(i) Wend Close #filenum n = i

PAGE 94

82 Appendix A. (Continued) 'i=1 is the heading or comment line with column's names at the begining of file For i = 2 To n T(1) = 0 'when there are no tips in a 5 minutes period or more, it skips them 'and use intensity for that period equal to intensity of next period If Ttip(i) / 60 > 5 And Ttip(i + 1) / 60 < 5 Then T(i) = Ttip(i + 1) / 60 ElseIf Ttip(i) / 60 > 5 And Ttip(i + 1) / 60 > 5 Then c = i Do Until Ttip(i) / 60 <= 5 i = i + 1 Loop T(c) = Ttip(i) / 60 i = c Else T(i) = Ttip(i) / 60 + T(i 1) End If Next i cmdcalculate.SetFocus End Sub Private Sub cmdcalculate_Click() 'Interpolate to find the cumulative depth every minute and intensity in in/min Dim i As Integer 'index for each tip Dim j As Integer 'index for time every minute Dim intold As Single 'intensity for period before Dim Wmin As Single 'cumulative depth of rain in inches Dim Wminold As Single 'c umulative depth of rain for period before Dim Wminnew As Single 'cumulative depth of rain for period after Dim intensity As Single 'intensity in in/min Dim Jmin As Integer 'time minute to minute for new value Dim secondata As Integer seconds for new value Dim minutedata As Integer 'minutes for new value Dim hourdata As Integer 'hours for new value Dim timedatamin As String 'date and time for new value Dim filename As String 'file name for input file Call output_file frmra infall_minute.MousePointer = 11 cmdexit.SetFocus j = 1; i = 1 Wminnew = 0

PAGE 95

83 Appendix A. (Continued) Wminold = 0 intensity = 0 Jmin = 0 Do While i < n 1 Do While i < n 1 i = i + 1 Wminold = Wminold + 0.01 Do While Abs(((Day(datedata(i + 1)) 1440 + Hour(timedata(i + 1)) 60 + Minute(timedata(i + 1)) + Second(timedata(i + 1)) / 60) (Day(datedata(i)) 1440 + Hour(timedata(i)) 60 + Minute(timedata(i)) + Second(timedata(i)) / 60))) < 5 'Interpolate when it has to create more than 1 value Do While j >= T(i) And j < T(i + 1) intold = Wmin Wmin = Wminold + 0.01 / (T(i + 1) T(i)) (j T(i)) intensity = (Wmin intold) Jmin = j secondata = Int((Jmin T(i)) 60 + 0 .5) + Second(timedata(i)) minutedata = Minute(timedata(i)) hourdata = Hour(timedata(i)) timedatamin = DateValue(datedata(i)) + TimeSerial(hourdata, minutedata, secondata) Write #1, timedatamin, Jmin, intensit y 60, Wmin j = j + 1 Loop If j >= T(i + 1) Then Exit Do 'linear regresion to insert values ElseIf j < T(i) Then Wminold = Wmin intensity = 0.01 / T(i) Wmin = Wminold + 0.01 / T(i) Jmin = j secondata = Second(timedata(i)) Int((T(i) Jmin) 60 + 0.5) minutedata = Minute(timedata(i)) hourdata = Hour(timedata(i)) timedatamin = DateValue(datedata(i)) + TimeSerial(hourdata, minutedata, secondata) Write #1, timedatamin, Jmin, intensity 60, Wmin j = j + 1 End If Wminold = 0.01 Loop

PAGE 96

84 Appendix A. (Continued) If Abs(((Day(datedata(i + 1)) 1440 + Hour(timedata (i + 1)) 60 + Minute(timedata(i + 1)) + Second(timedata(i + 1)) / 60) (Day(datedata(i)) 1440 + Hour(timedata(i)) 60 + Minute(timedata(i)) + Second(timedata(i)) / 60))) >= 5 Then If T(i) = T(i + 1) Then If T(i) = 0 Then Exit Do ElseIf T(i) <= 1 Then Wminold = 0 Wmin = 0 j = 1 Exit Do ElseIf T(i) > 1 Then Do While j <= T(i) Wminold = Wmin intensity = 0.01 / T(i) Wmin = Wminold + 0.01 / T(i) Jmin = j secondata = Second(timedata(i)) Int((T(i) Jmin) 60 + 0.5) minutedata = Minute(timedata(i)) hourda ta = Hour(timedata(i)) timedatamin = DateValue(datedata(i)) + TimeSerial(hourdata, minutedata, secondata) Write #1, timedatamin, Jmin, intensity 60, Wmin j = j + 1 Loop inten sity = 0.01 Wmin Wmin = 0.01 Jmin = j timedatamin = DateValue(datedata(i)) + TimeSerial(hourdata, minutedata + 1, secondata) End If 'when time to tip is < 1 min, distribute depth over the minute ElseIf j > T(i) Then If T(i) > 1 Then intensity = Wminold Wmin Wmin = Wminold Jmin = j secondata = Int((Jmin T(i)) 60 + 0.5) + Second(timedata(i)) minutedata = Minute(timedata(i)) hourdata = Hour(timedata(i)) timedatamin = DateValue(datedata(i)) + TimeSerial(hourdata, minutedata, secondata) Else

PAGE 97

85 Appendix A. (Continued) Wminold = 0 Wmin = 0 j = 1 Exit Do End If ElseIf j < T(i) Then Exit Do End If Write #1, timedatamin, Jmin, intensity 60, Wmin Wminold = 0 Wmin = 0 j = 1 End If Loop Loop Close #1 frmrainfall_minute.MousePointer = 0 cmdexit.SetFocus End Sub Sub output_file() 'This sub creates an output file (.txt) using a dialog box and copy 'date and time, time(min), intensities(in/hr), cummulative dept h of rain(in) Dialogfile.Filter = "all files(*.*)|*.*|text files(*.txt)|*.txt|comma delimited files(*.csv)|*.csv|" Dialogfile.FilterIndex = 3 Dialogfile.ShowSave Dialogfile.CancelError = True Open Dialogfile.filename For Append As #1 Write #1, "Date_and_T ime", "Time(min)", "intensity(in/hr)", "cummulative_depth(in)" End Sub Private Sub cmdexit_Click() End End Sub

PAGE 98

86 Appendix B. Subroutine Event_Minute 'Program: Separation of storms in Convective and Frontal 'Form name: frmevent minute 'Programmer : Tatiana X. Hernandez. 'Created: July 30 1999. Last Modified: May 26, 2000 'Purpose: This program is design to separate storms and calculate the duration and total depth of each storm. It used an inter event period of 1 hr for convective storms and 8 hours for frontal storms. The program reads date and time, time, intensity, and cummulative depth from a text file and calculates total depth of storm and duration of storms It creates a .txt output file Option Explicit Dim datetimedata(50000) As String 'date and time each minute Dim duration(50000) As Single 'duration of storm in minutes Dim depth(50000) As Single 'depth of storm in inches Dim intensity(5 0000) As Single 'intensity every minute (in/hr) Dim n As Integer 'total number of records Dim tstorm As Single 'variable to identify type of storm Private Sub cmdinputfile_Click() Dim i As Integer 'coun ter Dim typestorm As String 'variable to identify type of storm Dim c As String 'convective type storm Dim f As String 'frontal type storm 'Input a file using a dialog box dialogfile.Filter = "all files(*.*)|*.*|comma delimited files(*.csv)|*.csv|text files(*.txt)|*.txt|" dialogfile.FilterIndex = 2 dialogfile.ShowOpen dialogfile.CancelError = True Open dialogfile.FileName For Input As #1 'Distinguish between convective and frontal type of storm typestorm = InputBox("Type of sto rm (c for convective or f for frontal)", "Type of storm", "c") If typestorm = "c" Or typestorm = "C" Then tstorm = 60 'One hour dry inter event If typestorm = "f" Or typestorm = "F" Then tstorm = 480 'Eight hours dry inter event

PAGE 99

87 Appendix B. (Continued) 'Read values of date and time, time, intensity and cummulative depth from a text file i = 0 While Not EOF(1) i = i + 1 Input #1, datetimedata(i), duration(i), intensity(i), depth(i) Wend Close #1 datetimedata(i + 1) = 0 n = i cmdcalculate.SetFocus End Sub Private Sub cmdcalculate_Click() 'calculate cummulative depth of rain and duration for each storm Dim i As Integer 'index for each tip Dim j As Integer 'index for time every minute Dim deltat As Single 'difference in time (minutes) Dim olduration As Single 'cumulative duration of each storm Dim oldepth As Single 'cumulative depth of each storm Dim startstorm As String 'date and time and which the storm starts Call output_file frmevent_minute.MousePointe r = 11 cmdexit.SetFocus olduration = 0 oldepth = 0 'i=1 is the heading with the variable names startstorm = datetimedata(2) For i = 2 To n deltat = ((Day(datetimedata(i + 1)) 1440 + Hour(datetimedata(i + 1)) 60 + Minute(datetimedata(i + 1)) + Second(d atetimedata(i + 1)) / 60) (Day(datetimedata(i)) 1440 + Hour(datetimedata(i)) 60 + Minute(datetimedata(i)) + Second(datetimedata(i)) / 60)) If Abs(deltat) <= tstorm And duration(i) = 1 Then Do Write #3, datetimedata(i), duration(i), intens ity(i), depth(i), "", "" i = i + 1 If i >= n Then Exit Do Loop While duration(i + 1) <> 1 oldepth = oldepth + depth(i)

PAGE 100

88 Appendix B. (Continued) olduration = ((Day(datetimedata(i)) 1440 + Hour(datetimedata(i)) 60 + Minute(da tetimedata(i)) + Second(datetimedata(i)) / 60) (Day(startstorm) 1440 + Hour(startstorm) 60 + Minute(startstorm) + Second(startstorm) / 60)) + 1 i = i 1 ElseIf Abs(deltat) <= tstorm And duration(i) <> 1 Then Write #3, datetimedata(i), dura tion(i), intensity(i), depth(i), "", "" olduration = ((Day(datetimedata(i + 1)) 1440 + Hour(datetimedata(i + 1)) 60 + Minute(datetimedata(i + 1)) + Second(datetimedata(i + 1)) / 60) (Day(startstorm) 1440 + Hour(startstorm) 60 + Minute(startst orm) + Second(startstorm) / 60)) + 1 Else If duration(i) = 1 Then startstorm = datetimedata(i + 1) Else Write #3, datetimedata(i), duration(i), intensity(i), depth(i), olduration / 60, oldepth startstorm = datetimedata(i + 1) olduration = 0 oldepth = 0 End If End If Next i Close #3 frmevent_minute.MousePointer = 0 cmdexit.SetFocus End Sub Sub output_file() 'This sub creates an output file using a dialog box and copy 'the date and time, time(min), int ensities(in/hr), cummulative depth(in), total duration(hr), and total depht(in) of storm dialogfile.Filter = "all files(*.*)|*.*|text files(*.txt)|*.txt|comma delimited files(*.csv)|*.csv|" dialogfile.FilterIndex = 3 dialogfile.ShowSave dialogfile.CancelE rror = True Open dialogfile.FileName For Append As #3 Write #3, "date and time", "Time(min)", "Intensity(in/hr)", "Depth(in)", "Total_duration_of_storm(hr)", "Total_depth_of_storm(in)" End Sub Private Sub cmdexit_Click() End End Sub

PAGE 101

89 Appendix C. Output File from the Event_Minute Subroutine Table 13 Sample of output file obtained using the event_minute subroutine Date and time Time (min) Intensity (in/hr) Depth (in) Total duration of storm(hr) Total depth of storm(in) 9/3/1998 4:22:16 PM 1 3.35596 0.05593 9/3/1998 4:22: 49 PM 2 0.24404 0.06000 9/3/1998 4:33:26 PM 1 0.23415 0.00390 9/3/1998 4:34:26 PM 2 0.23415 0.00780 9/3/1998 4:35:26 PM 3 0.13171 0.01000 9/3/1998 5:10:26 PM 1 0.23415 0.00390 9/3/1998 5:11:26 PM 2 0.23415 0.00780 9/3/1998 5:1 2:26 PM 3 0.23415 0.01171 9/3/1998 5:13:26 PM 4 0.23415 0.01561 9/3/1998 5:14:26 PM 5 0.23415 0.01951 9/3/1998 5:15:53 PM 6 0.02927 0.02000 0.910 0.09 9/3/1998 9:25:50 PM 1 0.14388 0.00240 9/3/1998 9:26:50 PM 2 0.14388 0.00480 9/3 /1998 9:27:50 PM 3 0.14388 0.00719 9/3/1998 9:28:50 PM 4 0.14388 0.00959 9/3/1998 9:29:50 PM 5 0.14388 0.01199 9/3/1998 9:30:50 PM 6 0.14388 0.01439 9/3/1998 9:31:50 PM 7 0.14388 0.01679 9/3/1998 9:32:50 PM 8 0.14388 0.01918 9 /3/1998 9:34:11 PM 9 0.95590 0.03512 9/3/1998 9:35:17 PM 10 1.84484 0.06586 9/3/1998 9:36:09 PM 11 2.52603 0.10796 9/3/1998 9:37:03 PM 12 3.27978 0.16263 9/3/1998 9:38:02 PM 13 3.64796 0.22343 9/3/1998 9:39:02 PM 14 4.08619 0.29153 9/3/1998 9:40:04 PM 15 1.16778 0.31099 9/3/1998 9:41:20 PM 16 0.96482 0.32707 9/3/1998 9:42:05 PM 17 1.50248 0.35211 9/3/1998 9:43:11 PM 18 1.35361 0.37467 9/3/1998 9:43:47 PM 19 0.89941 0.38966 9/3/1998 9:44:58 PM 20 0.02010 0.39000 0.336 0.39 9/3/1998 11:51:33 PM 1 0.41176 0.00686 9/3/1998 11:52:33 PM 2 0.41176 0.01373 9/3/1998 11:53:05 PM 3 0.39842 0.02037 9/3/1998 11:54:05 PM 4 0.25605 0.02463 9/3/1998 11:55:05 PM 5 0.25605 0.02890 9/3/1998 11:55: 45 PM 6 0.06597 0.03000 9/4/1998 12:09:13 AM 1 1.69098 0.02818 9/4/1998 12:09:45 AM 2 1.06258 0.04589 9/4/1998 12:11:28 AM 3 0.36505 0.05198 9/4/1998 12:12:28 AM 4 0.25148 0.05617 9/4/1998 12:13:05 AM 5 0.38249 0.06254 9/4/199 8 12:13:45 AM 6 0.44741 0.07000 0.3866699 0.1 9/4/1998 10:37:04 AM 1 1.39737 0.02329 9/4/1998 10:38:06 AM 2 1.72140 0.05198 9/4/1998 10:39:10 AM 3 1.29374 0.07354 9/4/1998 10:40:16 AM 4 1.34253 0.09592 9/4/1998 10:41:26 AM 5 1.37308 0. 11880 9/4/1998 10:42:15 AM 6 0.90172 0.13383 9/4/1998 10:43:36 AM 7 0.76376 0.14656 9/4/1998 10:43:41 AM 8 0.47901 0.15454 9/4/1998 10:45:11 AM 9 0.32739 0.16000 9/4/1998 11:16:52 AM 1 0.53191 0.00887 9/4/1998 11:17:52 AM 2 0. 06809 0.01000 0.6966634 0.17

PAGE 102

90 Appendix D. Florida Soil Survey for Myakka Fine Sand Figure 38 Sample copy of actual soil hydraulic parameters for Myakka fine sand (Carlisle et al., 1989)

PAGE 103

91 Appendix E. HYDRUS 1D Graphs for Temporal Variability of Rainfall Frontal Storm 0.20 0.15 0.10 0.05 0.00 0 100 200 300 400 500 600 700 800 Time [minutes] Figure 39 Rainfall rate vs time 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.1 1 10 100 1000 1000 pressure head [cm] Figure 40 Water re tention curve

PAGE 104

92 Appendix E. (Continued) Figure 41 Water content vs depth (1 meter) Figure 42 Infiltration rate vs time (1 meter) Figure 43 Cumulative infiltration depth vs time (1 meter) 100 80 60 40 20 0 0.10 0.20 0.25 0.30 0.40 Theta [ ] 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 0 100 200 300 400 500 600 700 800 Time [min] 12 10 8 6 4 2 0 0 100 200 300 400 500 600 700 800 Time [min]

PAGE 105

93 Appendix E. (Continued) Figure 44 Water content vs depth (2 meters) Figure 45 Infiltration rate vs time (2 meters) Figure 46 Cumulative infiltration depth vs time (2 meters) 200 150 100 50 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Theta [ ] 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 0 100 200 300 400 500 600 700 800 Time [min] 12 10 8 6 4 2 0 0 100 200 300 400 500 600 700 800 Time [min]

PAGE 106

94 Appendix E. (Continued) Figure 47 Water content vs depth (5 meters) Figure 48 Infiltration rate vs time (5 meters) Figure 49 Cumulative infiltration depth vs time (5 meters) 500 400 300 200 100 0 0.0 0.1 0.2 0.3 0.4 Theta [ ] 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 0 100 200 300 400 500 600 700 800 Time [min] 12 10 8 6 4 2 0 0 100 200 300 400 500 600 700 800 Time [min]

PAGE 107

95 Appendix E. (Continued) Convective Storm 0.25 0.20 0.15 0.10 0.05 0.00 0 50 100 150 200 250 Time [min] Figure 50 Rain fall rate for comnvective storm 100 80 60 40 20 0 0 50 100 150 200 250 Time [min] Figure 51 Pressure head at the surface (1 meter)

PAGE 108

96 Appendix E. (Continued) 100 80 60 40 20 0 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Theta [ ] Figure 52 Water content vs depth (1 meter ) 0.20 0.15 0.10 0.05 0.00 0 50 100 150 200 250 Time [min] Figure 53 Infiltration rate vs time (1 meter) 10 8 6 4 2 0 0 50 100 150 200 250 Time [min] Figure 54 Cumulative infiltration depth vs time (1 meter)

PAGE 109

97 Appendix E. (Continued) 200 150 100 50 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Theta [ ] Figure 55 Water content vs depth (2 meters) 0.20 0.15 0.10 0.05 0.00 0 50 100 150 200 250 Time [min] Figure 56 Infiltration rate vs time (2 meters) 12 10 8 6 4 2 0 0 50 100 150 200 250 Time [min] Figure 57 Cumulat ive infiltration depth vs time (2 meters)

PAGE 110

98 Appendix E. (Continued) 500 400 300 200 100 0 0.0 0.1 0.2 0.3 0.4 Theta [ ] Figure 58 Water content vs depth (5 meters) 0.25 0.20 0.15 0.10 0.05 0.00 0 50 100 150 200 250 Time [min] Figure 59 Infiltration rate vs time (5 meters) 12 10 8 6 4 2 0 0 50 100 150 200 250 Time [min] Figure 60 Cumulative infiltration depth vs time (5 meters)

PAGE 111

99 -1 8 0 1 2 3 4 5 6 7 Appendix F. HYDRUS 2D Graphs for Reference Simulation Initial water table depth at t=0 Water table position after rainfall stops at t=2.25 hr ______________36.9 m_________ VSA Closer view of velocity vectors near the stream at t=2.25 hr Velocity vectors at t = 96 hr (4 days, end of simulation) Figure 61 Profiles of reference simulations from HYDRUS 2D

PAGE 112

100 Appendix F. (Continued) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 50 100 150 200 250 300 350 400 Length (meters) Figure 62 Pressure heads at the surface boundary 0.010 0.015 0.020 0.025 0.030 0.0 0.5 1.0 1.5 2.0 2.5 Time [hours] Figure 63 Subsurface s torm flux during rainfall 0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.0 0.5 1.0 1.5 2.0 2.5 Time [hours] Figure 64 Cumulative subsurface storm flow during rainfall

PAGE 113

101 Appendix F. (Continued) 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0 20 40 60 80 100 Time [hours] Figure 65 Subsurface flux after rain fall 0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0 20 40 60 80 100 Time [hours] Figure 66 Cumulative seepage flux after rainfall 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0 20 40 60 80 100 Time [hours] Figure 67 Seepage flux after rainfall

PAGE 114

102 Appendix G. Water Retention Curves for Actual Soil D ata and Fitter Parameters 0 50 100150200250300350 0.000 0.100 0.200 0.300 0.400 0.500 Water Content in cm3 of water/cm3 of soil Capillary Pressure in cm of water Measured watercontent Van Genutchen Figure 68 Fitter van Genuchten parameters for Myakka fine sand 0 50 100150200250300350 0.0000.1000.2000.3000.4000.500 Water Content in cm3 of water/cm3 of soil Capillary Pressure in cm of water Measured watercontent Van Genutchen Figure 69 Fitter van Genuchten parameters for Mulat fine sand 0 50 100150200250300350 0.0000.1000.200 0.3000.4000.500 Water Content in cm3 of water/cm3 of soil Capillary Pressure in cm of water Measured watercontent Van Genutchen Figure 70 Fitter van Genuchten parameters for Millhopper fine sand

PAGE 115

103 Appendix H. Hyetographs for Convective and Frontal Storms Convective storm (September 22, 1996) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.02 0.13 0.25 0.37 0.48 0.60 0.72 0.83 0.95 1.07 1.18 1.30 1.42 1.53 1.65 1.77 1.88 2.00 2.12 2.23 Time (hr) Intensity (m/hr) Convective Storm (June 25, 1995) 0 0.02 0.04 0.06 0.08 0.1 0.02 0.17 0.32 0.47 0.62 0.77 0.92 1.07 1.22 1.37 1.52 1.67 1.82 1.97 2.12 Time (hr) Intensity (m/hr) Convective storm (April 28, 1997) 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.02 0.17 0.32 0.47 0.62 0.77 0.92 1.07 1.22 1.37 1.52 1.67 1.82 1.97 2.12 Time (hr) Intensity (m/hr) Figure 71 Convective storms used for sensitivity analysis

PAGE 116

104 Appendix H. (Continued) Frontal storm (January 23/1998) 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 0.02 0.35 0.68 1.02 1.35 1.68 2.02 2.35 2.68 3.02 3.35 3.68 4.02 4.35 4.68 5.02 5.35 Time (hr) Intensity (m/hr) Frontal Storm (December 13, 1997) 0 0.005 0.01 0.015 0.02 0.025 0.03 0.02 0.23 0.45 0.67 0.88 1.1 1.32 1.53 1.75 1.97 2.18 2.4 2.62 2.83 3.05 3.27 3.48 3.7 Time (hr) Intensity (m/hr) Figure 72 Frontal storms used for sensitivity analysis

PAGE 117

105 Appendix I. Initial Water Table Level Curved initial water table depth for reference simulation Initial water table depth at 1 ft, 3 ft, and 4 ft Figure 73 Initial water table lev el for sensitivity analysis

PAGE 118

106 Appendix J. Graphic Results for Soils Sensitivity Analysis 0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0 20 40 60 80 100 Time (hours) Subsurface Flux (m2/hr) sand with n=1.26 sandy loam myakka fine sand Figure 74 Subsurface fluxes for 3 soil types 0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.018 0.020 0 2 4 6 8 10 Time (hours) Saturation Excess flux (m2/hr) sand with n=1.26 sandy loam myakka Figure 75 Saturation excess fluxes for 3 soil types 0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.018 0 20 40 60 80 100 120 Time (hours) Fluxes (m2/hr) Subsurface flux Surface flux Figure 76 Comparison of fluxes for Miakka fine sand

PAGE 119

107 Appendix K. Graphic Results for Rainfall Sensitivity Analysis 0.000 0.002 0.004 0.006 0.008 0.010 0.012 0 10 20 30 40 50 Tme (hours) Saturation excess flux (m2/hr) 1st storm 2nd storm 3rd storm Figure 77 Saturation excess fluxes for convective stroms 0.000 0.002 0.004 0.006 0.008 0.010 0.012 0 10 20 30 40 50 Time (hours) Saturation excess flux (m2/hr) storm with duration = 5.62 hr storm with duration = 3.8 hr Figure 78 Saturation excess fluxes for frontal storms

PAGE 120

108 Appendix K. (Continued) 0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.045 0.050 0 20 40 60 80 100 Time (hours) Subsurface flux (m2/hr) 1st storm 2nd storm 3rd storm Figure 79 Subsurface fluxes for convective storms 0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0 20 40 60 80 100 Time (hours) Subsurface flux (m2/hr) storm with duration = 5.62 hr storm with duration = 3.8 hr Figure 80 Subsurface fluxes for frontal storms

PAGE 121

109 Appendix L. Graphic Results for Water Table Depths 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0 20 40 60 80 100 Time (hours) Subsurface flux (m2/hr) wt = 1 ft wt = 3.5 ft wt = 4 ft Figure 81 Subsurface flux after rainfall stops 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0 5 10 15 20 Time (hours) Seepage flux (m2/hr) wt = 3.5 ft wt = 4 ft Figure 82 Seepage flux after rainfall stops

PAGE 122

110 Appendix M. Graphic Results for Slope Sensitivity 0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 0.900 1.000 0 20 40 60 80 100 Time (hours) Cummulative surface flux (m2/hr) Slope 0.5 % Slope 1.1 % Slope 2.5 % Figure 83 Cumulative surface flux for different slopes 0.000 1.000 2.000 3.000 4.000 5.000 6.000 7.000 8.000 9.000 10.000 0 20 40 60 80 100 Time (hours) Cummulative subsurface flux (m2/hr) Slope 0.5 % Slope 1.1 % Slope 2.5 % Figure 84 Cumulative subsurface flux

PAGE 123

111 Appendix N. VSA Extent for C oncave Curvature Concave profile with Radius of curvature equal to 14871.1 m Concave profile with Radius of curvature equal to 19474 m Concave profile with Radius of curvature equal to 28631.9 m Figure 85 Exten t of VSA after rainfall for concave profiles


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 001413320
003 fts
005 20061107134909.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 030703s2001 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000010
035
(OCoLC)47655549
9
AJJ0736
b SE
SFE0000010
040
FHM
c FHM
049
FHME
090
TA145
1 100
Hernandez, Tatiana X.
0 245
Rainfall-runoff modeling in humid shallow water table environments
h [electronic resource] /
by Tatiana X. Hernandez.
260
[Tampa, Fla.]:
University of South Florida,
2001.
502
Thesis (M.S.)--University of South Florida, 2001.
504
Includes bibliographical references.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 123 pages.
520
ABSTRACT: Simulating the processes of rainfall and runoff are at the core of hydrologic modeling. Geomorphologic features, rainfall variability, soil types, and water table depths strongly influence hydrological process in Florida ecosystems. Topographic characteristics of the terrain define the stream paths and landscape. Alteration of these characteristics as a result of urban and/or agricultural developments, for example, can highly influence wetlands and river basin response. There are two predominant landforms in Florida: wetlands, where Variable Saturated Areas form near streams causing saturation excess runoff, and uplands where runoff is mainly generated by infiltration excess. The objective of this work is to analyze the impacts of geomorphologic and hydrologic characteristics on runoff mechanisms in humid environments such as Florida. In general, most research at the hillslope scale use hypothetical values of rainfall, sometimes non-realistic values, and single slope^forms to explain the geomorphic and hydrologic process on Variable Saturated Areas. In this thesis, the complexity of hillslope processes on actual Florida topography is assessed by coupling a Digital Elevation Model with a two-dimensional variable saturated-unsaturated flow model called HYDRUS-2D. Actual rainfall records and soil parameters from the Characterization Data for Selected Florida Soils, Soil Survey were used to evaluate hydrologic impacts. A commercial software package, River Tools was used to display and extract topographic information from the Digital Elevation Models.Results show that when inflitration excess runoff is dominant, infiltration and runoff are very sensitive to time resolution, especially for convective storms. When saturation excess occurs, runoff is not affected by rainfall intensity. However, saturated hydraulic conductivity, depth to the water table, slope and curvature highly influence the extent of Variable Saturated Areas. Results indicate runoff in shallow water table environments is produced mainly by subsurface storm runoff, running below the surface, except in hillslopes with concave curvature and mild slopes. Additionally, concave hillslopes generate more saturation excess runoff than straight and convex hillslopes.
516
Text (Electronic thesis) in PDF format.
590
Adviser: Nachabe, Mahmood H.
650
Runoff
x Computer simulation.
Slopes (Physical geography)
690
Dissertations, Academic
z USF
Civil Engineering
Masters.
653
hydrologic modeling.
rainfall variability.
variable source areas.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.10