Copyright by David James Anderson 2000 GIS-BASED HYDROLOGIC AND HYDRAULIC MODELING FOR FLOODPLAIN DELINEATION AT HIGHWAY RIVER CROSSINGS by David James Anderson, B.S. Thesis Presented to the Faculty of the Graduate School of the University of Texas at Austin in Partial Fulfillment of the Requirements for the Degree of Master of Science in Engineering The University of Texas at Austin December 2000 GIS-BASED HYDROLOGIC AND HYDRAULIC MODELING FOR FLOODPLAIN DELINEATION AT HIGHWAY RIVER CROSSINGS APPROVED BY SUPERVISING COMMITTEE: __________________________ David R. Maidment __________________________ Francisco Olivera iv ACKNOWLEDGEMENTS I would like to begin by thanking Dr. Francisco Olivera, without whom I would not have had the opportunity to work on such an interesting project, and whose insights proved invaluable as we met project objectives together. I would also like to thank Dr. David Maidment, who as a professor and mentor has allowed me to share his interests in the integration of water resources and GIS, and has reminded me that a teacher can make a difference. The Texas Department of Transportation funded this research, and Mr. Tony Schneider and Mr. David Stolpa at the Hydraulics Division in Austin deserve much gratitude for their support. Similarly, personnel in the district offices should also be thanked ? Mr. David Neumann and Mr. Lynn Passmore especially. Several members of the GIS research team at the Center for Research in Water Resources also merit my gratitude, especially Esteban Azagra and Eric Tate, who made significant strides in developing the tools presented in this thesis, and Dan Snead, who has worked alongside me in putting these tools to the test. Katherine Osborne and Jona Finndis Jonsdottir should also be thanked for tirelessly answering numerous questions and always providing encouragement when I needed it most. Finally, I?d like to thank my family for providing me the basic tools in life to recognize the importance of education, and the support necessary to help me realize my academic goals. ?For I can do all things through Him who strengthens me? - Phillipians 4:13. November 1, 2000 v ABSTRACT GIS-BASED HYDROLOGIC AND HYDRAULIC MODELING FOR FLOODPLAIN DELINEATION AT HIGHWAY RIVER CROSSINGS by David James Anderson, M.S.E. The University of Texas at Austin, 2000 SUPERVISOR: David R. Maidment The importance of the spatial variability inherent to a watershed contributing flow to highway drainage structures can greatly affect the time and resources dedicated to the design process, as well as the size and cost of the structure. Evaluating extreme storm events and the resulting floodplain is a time-consuming process that, in the past, has been accomplished by manually plotting the extent of the floodplain on paper maps. Automating this process, with the aid of geographical information systems (GIS), could result in significant time and resource savings in the design process. This research investigates the synthesis of previously developed hydrologic and hydraulic modeling tools for digital floodplain analysis at two locations - Castleman Creek (McClennan County, TX) and Pecan Bayou (Brown County, TX). The methodology proposed consists of site-specific terrain data development for hydrologic analysis and parameter extraction using CRWR-PrePro, terrain data development and floodplain delineation using CRWR-FloodMap and vi HEC-GeoRAS, and lumped parameter hydrologic modeling and steady flow hydraulic analysis using HEC-HMS and HEC-RAS. The results of the research indicate that although the availability of digital terrain data at an appropriate resolution may limit the application of these tools at small-scale sites such as are found at some highway river crossings, the methodology presented is an effective tool for representing the spatial variability of the watershed characteristics, integrating hydrologic and hydraulic modeling processes with GIS, and displaying an accurate floodplain map of the project site. vii TABLE OF CONTENTS TABLE OF CONTENTS......................................................................................vii LIST OF TABLES ................................................................................................. ix LIST OF FIGURES................................................................................................. x 1 INTRODUCTION........................................................................................... 1 1.1 Background ............................................................................................. 1 1.2 Site Selection........................................................................................... 2 1.3 Objectives................................................................................................ 6 1.4 Organization............................................................................................ 7 2 LITERATURE REVIEW................................................................................ 9 2.1 Digital Terrain Models............................................................................ 9 2.2 GIS in Hydrologic Modeling ................................................................ 13 2.3 GIS in Hydraulic Modeling................................................................... 16 2.4 Synthesis of GIS-Based Hydrologic and Hydraulic Modeling for Floodplain Delineation.......................................................................... 19 3 DATA............................................................................................................ 20 3.1 Castleman Creek ................................................................................... 21 3.2 Pecan Bayou.......................................................................................... 34 4 METHODOLOGY........................................................................................ 52 4.1 Site Specific Terrain Data Development............................................... 53 4.2 GIS-based Hydrologic Parameter Extraction........................................ 68 4.3 GIS-based Hydraulic Geometry Extraction .......................................... 73 4.4 Hydrologic Modeling ............................................................................ 82 4.5 Hydraulic Modeling .............................................................................. 85 4.6 Floodplain Delineation.......................................................................... 88 5 IMPLEMENTATION PROCEDURES ........................................................ 93 5.1 Castleman Creek ................................................................................... 93 5.2 Pecan Bayou........................................................................................ 119 viii 6 RESULTS.................................................................................................... 144 6.1 Castleman Creek ................................................................................. 144 6.2 Pecan Bayou........................................................................................ 160 7 CONCLUSIONS AND RECOMMENDATIONS ..................................... 169 7.1 Conclusions ......................................................................................... 170 7.2 Recommendations ............................................................................... 177 APPENDIX A: Data Dictionary ........................... Error! Bookmark not defined. APPENDIX B: ArcView GIS Scripts ................... Error! Bookmark not defined. Appendix B.1: CRWR-FloodMap Amended ScriptsError! Bookmark not defined. Appendix B.2: Miscellaneous Scripts ............... Error! Bookmark not defined. APPENDIX C: Castleman Creek Hydrologic and Hydraulic ModelingError! Bookmark not defined. Appendix C.1: Existing Hydrologic Modeling DataError! Bookmark not defined. Appendix C.2: Hydrologic Modeling Results... Error! Bookmark not defined. Appendix C.3: Existing Hydraulic Modeling DataError! Bookmark not defined. Appendix C.4: Hydraulic Modeling Results..... Error! Bookmark not defined. APPENDIX D: Pecan Bayou Hydrologic and Hydraulic ModelingError! Bookmark not defined. Appendix D.1: Existing Hydrologic Modeling DataError! Bookmark not defined. Appendix D.2: Hydrologic Modeling Results . Error! Bookmark not defined. Appendix D.3: Existing Hydraulic Modeling .. Error! Bookmark not defined. Appendix D.4: Hydraulic Modeling Results.... Error! Bookmark not defined. REFERENCES...................................................... Error! Bookmark not defined. VITA ..................................................................... Error! Bookmark not defined. ix LIST OF TABLES Table 1-1 Summary of Floodplain Delineation Methodology......................................... 7 Table 3-1 Point Coverage of Hydrologic and Hydraulic Gauging Sites ...................... 41 Table 4-1 HEC-HMS Basin Component Summary........................................................ 83 Table 4-2 HEC-HMS Precipitation Component Summary........................................... 84 Table 6-1 Naming Conventions for Hydrologic Elements of Castleman Creek Watershed without SCS Flood Control Structures ..............................................146 Table 6-2 Comparisons of Watershed Areas for Castleman Creek without SCS Flood Control Structures.....................................................................................................147 Table 6-3 Comparisons of Loss Rates for Castleman Creek without SCS Flood Control Structures.....................................................................................................147 Table 6-4 Comparisons of Lag Times for Castleman Creek without SCS Flood Control Structures.....................................................................................................148 Table 6-5 Comparisons of Routing Parameters for Castleman Creek without SCS Flood Control Structures .........................................................................................149 Table 6-6 Comparisons of Peak Flow Values for Various Storm Return Periods for Castleman Creek without SCS Flood Control Structures ...................................151 Table 6-7 Comparison of Time-to-Peak for Various Storm Return Period for Castleman Creek without SCS Flood Control Structures ...................................151 Table 6-8 Range of Peak Flows for Differing Stream Velocities and Muskingum X Values..........................................................................................................................152 Table 6-9 Comparisons of Peak Flow Values for Various Storm Return Periods for Castleman Creek with SCS Flood Control Structures.........................................154 Table 6-10 Comparison of Time-to-Peak for Various Storm Return Period for Castleman Creek with SCS Flood Control Structures.........................................155 x LIST OF FIGURES Figure 1-1 Location Map of Castleman Creek Watershed............................................... 3 Figure 1-2 Location Map of Pecan Bayou Watershed...................................................... 5 Figure 3-1 TIN Elevation Data for US 77 in Castleman Creek Watershed................ 22 Figure 3-2 Castleman Creek Stream Network ................................................................. 23 Figure 3-3 Locations of SCS Flood Control Structures in Castleman Creek Watershed..................................................................................................................... 24 Figure 3-4 Comparison of Regional Regression Equations for Castleman Creek ..... 26 Figure 3-5 HEC-RAS Schematic of Surveyed Cross-Sections on Castleman Creek . 28 Figure 3-6 Discrepancies between DOQ, Photogrammetric Coordinates, and Texas Road Coverage............................................................................................................. 31 Figure 3-7 Artificial Wall Integrated in the DEM along US 77..................................... 32 Figure 3-8 SCS Curve Number Grid Coverage at the Castleman Creek Watershed . 33 Figure 3-9 Example Microstation ? Drawing for Pecan Bayou Watershed.................. 35 Figure 3-10 Pecan Bayou Stream Network ...................................................................... 37 Figure 3-11 Precipitation Data Recorded at NOAA Cooperative Station 419817 at Winchell, TX in Pecan Bayou Watershed................................................................ 38 Figure 3-12 Location of NOAA Cooperative Station 419817 in Pecan Bayou Watershed..................................................................................................................... 39 Figure 3-13 Recorded Lake Brownwood Elevation Data during December 1991 Storm Event................................................................................................................. 40 Figure 3-14 HEC-RAS Schematic of Surveyed Cross-Sections on Pecan Bayou ...... 42 Figure 3-15 Locations of Bridges of Interest in Pecan Bayou Watershed................... 43 Figure 3-16 Extent of Pecan Bayou NHD route.rch Coverage...................................... 44 Figure 3-17 Clipped NHD route.rch Coverage................................................................. 44 Figure 3-18 Attribute Table for NHD route.drain Coverage........................................... 45 Figure 3-19 Manual Editing of NHD route.rch Coverage with DRG........................... 46 Figure 3-20 Areal-Reduction Factor Equations for Dallas, TX.................................... 48 Figure 3-21 Comparison of Recorded Incremental Precipitation to Precipitation Adjusted with ARF ..................................................................................................... 49 Figure 3-22 Lake Brownwood Discharge Hydrograph .................................................. 50 Figure 3-23 Rating Curve at USGS Gage 08143600 near Mullin, TX ......................... 51 Figure 4-1 Schematic of Floodplain Delineation Methodology.................................... 53 Figure 4-2 CRWR-PrePro Implementation Procedures................................................. 54 Figure 4-3 Fill Sinks Algorithm.......................................................................................... 55 Figure 4-4 Raster-Based Functions for Terrain Analysis ............................................... 56 Figure 4-5 Terrain Development for Hydrologic Analysis............................................ 58 Figure 4-6 CRWR-FloodMap Menu.................................................................................. 59 xi Figure 4-7 Terrain Development for Floodplain Delineation....................................... 60 Figure 4-8 Surveyed and Interpolated Cross-Sections.................................................... 61 Figure 4-9 HEC-RAS Output File and Imported ArcView GIS Table....................... 62 Figure 4-10 Cross-Section Orientation Selection Menu................................................. 64 Figure 4-11 Interpolated Cross-Section Discrepancies .................................................. 65 Figure 4-12 Cross-Sections Inadequately Defined for Expected Water Surface Profile66 Figure 4-13 Longest Flowpath Calculation ...................................................................... 70 Figure 4-14 Prism and Wedge Storage in Muskingum Routing.................................... 71 Figure 4-15 GIS-Based Hydraulic Geometry Extraction............................................... 73 Figure 4-16 HEC-GeoRAS Flow Diagram ...................................................................... 75 Figure 4-17 HEC-GeoRAS preRAS Methodology.......................................................... 76 Figure 4-18 HEC-GeoRAS River ID Tool ...................................................................... 77 Figure 4-19 HEC-GeoRAS Flowpath ID Tool............................................................... 78 Figure 4-20 Creating the Land Use Table......................................................................... 79 Figure 4-21 HEC-GeoRAS Theme Setup Dialog Box................................................... 80 Figure 4-22 Hydrologic Modeling Using HEC-HMS..................................................... 82 Figure 4-23 Components of HEC-HMS Model.............................................................. 83 Figure 4-24 Hydraulic Modeling with HEC-RAS ........................................................... 85 Figure 4-25 Stream Cross-Section Schematic .................................................................. 87 Figure 4-26 Floodplain Delineation Using HEC-GeoRAS ........................................... 89 Figure 4-27 HEC-GeoRAS postRAS Methodology ........................................................ 89 Figure 4-28 Schematic of Water Surface and Terrain TINs .......................................... 91 Figure 5-1 Summary of CRWR-PrePro Terrain Development for Hydrologic Analysis at Castleman Creek Watershed.................................................................. 94 Figure 5-2 Additional Outlets Selected to Define Extents of HEC-RAS Cross- Section Data................................................................................................................. 95 Figure 5-3 Reversed and Interpolated HEC-RAS Cross-Sections................................ 96 Figure 5-4 Interpolated Cross-Sections in HEC-RAS.................................................... 97 Figure 5-5 CRWR-FloodMap Import Table .................................................................... 98 Figure 5-6 3D Cross-Section and Stream Centerline Themes....................................... 99 Figure 5-7 Procedure for Editing 3D Shapefiles...........................................................100 Figure 5-8 3D Point and PolylineZ Data for TIN Construction................................100 Figure 5-9 TIN Created from CRWR-FloodMap Georeferencing ............................101 Figure 5-10 Curve Number Grid.....................................................................................102 Figure 5-11 CRWR-PrePro Parameter Table.................................................................103 Figure 5-12 Longest Flow Path Grid ..............................................................................104 Figure 5-13 CRWR-PrePro Attribute Table for Each Hydrologic Element.............104 Figure 5-14 CRWR-PrePro HEC-HMS Schematic.......................................................105 Figure 5-15 HEC-GeoRAS preRAS Themes .................................................................106 Figure 5-16 HEC-GeoRAS Theme Selection Menu.....................................................107 Figure 5-17 HEC-HMS Project Definition....................................................................107 xii Figure 5-18 HEC-HMS Subbasin Editor .......................................................................108 Figure 5-19 HEC-HMS Schematic..................................................................................109 Figure 5-20 HEC-HMS Transform Parameter Editor .................................................110 Figure 5-21 HEC-HMS User-Specified Hyetograph Data ..........................................111 Figure 5-22 Application of Unique Hyetograph for Each Subbasin..........................111 Figure 5-23 HEC-HMS Control Specifications.............................................................112 Figure 5-24 HEC-HMS Simulation Manager.................................................................113 Figure 5-25 Example of Discharge Hydrograph at US 77 Bridge ..............................113 Figure 5-26 Importing GIS Data into HEC-RAS.........................................................114 Figure 5-27 Bank Relocation in HEC-RAS....................................................................115 Figure 5-28 HEC-RAS Cross-Section Schematic..........................................................115 Figure 5-29 HEC-RAS Flow Data Editor......................................................................116 Figure 5-30 HEC-GeoRAS Setup for RAS Post-processor ........................................117 Figure 5-31 Floodplain Grid for 100-year Storm with SCS Flood Control Structures118 Figure 5-32 3D Representation of 100-year storm event at Castleman Creek..........118 Figure 5-33 Summary of CRWR-PrePro Terrain Development for Hydrologic Analysis at Pecan Bayou...........................................................................................121 Figure 5-34 Addition of Interpolated Cross-Sections to Pecan Bayou Model .........122 Figure 5-35 Cross-Sections Imported in Pecan Bayou GIS Model Using CRWR- FloodMap...................................................................................................................123 Figure 5-36 Microstation ? Drawing Coverage...............................................................123 Figure 5-37 NED DEM Appended to Contour Data..................................................125 Figure 5-38 BuffElev Script Methodology........................................................................126 Figure 5-39 Terrain Data Utilized to Generate Pecan Bayou TIN.............................126 Figure 5-40 Pecan Bayou Terrain TIN ...........................................................................127 Figure 5-41 CRWR-PrePro Parameter Table for Pecan Bayou ..................................128 Figure 5-42 CRWR-PrePro Attribute Table for Each Hydrologic Element in Pecan Bayou ..........................................................................................................................129 Figure 5-43 CRWR-PrePro HEC-HMS Schematic.......................................................130 Figure 5-44 Original Pecan Bayou HEC-GeoRAS Cross-Sections............................131 Figure 5-45 Revised HEC-GeoRAS Cross-Sections for Pecan Bayou ......................132 Figure 5-46 Pecan Bayou Table Relating Manning?s n to Land Use...........................132 Figure 5-47 HEC-GeoRAS Theme Selection Menu for Pecan Bayou ......................133 Figure 5-48 HEC-HMS Basin Schematic for Pecan Bayou.........................................134 Figure 5-49 User-Specified Hyetograph Precipitation Data ........................................135 Figure 5-50 Lake Brownwood Spillway Discharge Data..............................................135 Figure 5-51 USGS Gage 08143600 Recorded Flow Data............................................136 Figure 5-52 Pecan Bayou Control Specifications ..........................................................137 Figure 5-53 Comparison of Calculated and Observed Flow Data at USGS Gage 08143600 ....................................................................................................................137 Figure 5-54 Summary Table of Discharge Hydrograph at FM 2126 Bridge.............138 xiii Figure 5-55 HEC-RAS Bridge Cross-Sections with Building Data Incorporated into the TIN.......................................................................................................................139 Figure 5-56 HEC-RAS Cross-Section Schematic for Pecan Bayou ...........................140 Figure 5-57 Extraction of Flow Data from HEC-DSS for Pecan Bayou..................140 Figure 5-58 Hourly Flow Data for Pecan Bayou...........................................................141 Figure 5-59 HEC-GeoRAS Setup for RAS Post-processor for Pecan Bayou..........142 Figure 5-60 Christmas 1991 Flood ..................................................................................143 Figure 6-1 Comparison of HEC-1 and HEC-HMS Stream Network Schematic Diagrams without SCS Flood Control Structures................................................145 Figure 6-2 Comparisons of Flowpath Lengths for Selected Subbasins in Castleman Creek without SCS Flood Control Structures.......................................................149 Figure 6-3 Comparisons of Peak Flows for Various Storm Return Periods for Castleman Creek without SCS Flood Control Structures ...................................150 Figure 6-4 Comparison of HEC-1 and HEC-HMS Stream Network Schematic Diagrams with SCS Flood Control Structures......................................................153 Figure 6-5 Comparisons of Peak Flows for Various Storm Return Periods for Castleman Creek with SCS Flood Control Structures.........................................154 Figure 6-6 Comparison of Castleman Creek Stage for Proposed US 77 Bridge ......156 Figure 6-7 Comparisons of Cross-Sections at Station 2000 for Castleman Creek with SCS Flood Control Structures.................................................................................158 Figure 6-8 Comparison of TINs with and without Re-sampled Cross-Sections......159 Figure 6-9 100-year Floodplain as Determined by HEC-HMS and HEC-RAS Hydrologic and Hydraulic modeling ......................................................................159 Figure 6-10 Comparison of Floodplain Generated from TxDOT and HEC-HMS Flow Data...................................................................................................................160 Figure 6-11 Area Summary (mi 2 )of Subbasins Contributing Flow to USGS Gage 08143600 ....................................................................................................................162 Figure 6-12 Cumulative Flow Summary of Subbasins Contributing Flow to USGS Gage 08143600..........................................................................................................162 Figure 6-13 Comparison of Precipitation, Spillway Discharge, and Observed Flow at USGS Gage 08143600..............................................................................................163 Figure 6-14 Observed and Modeled Discharge Hydrograph at USGS Gage 08143600164 Figure 6-15 Comparison of HEC-RAS and HEC-GeoRAS Cross-Sections............165 Figure 6-16 Summary of HEC-RAS Water Surface Profiles at each Cross-Section 166 Figure 6-17 Water Surface Profile for Pecan Bayou at FM 2126 Bridge...................167 Figure 6-18 Christmas 1991 Floodplain on Pecan Bayou............................................168 Figure 7-1 Schematic of Floodplain Delineation Methodology..................................169 Figure 7-2 Data Prioritization Flow Chart for Digital Terrain Development for Floodplain Delineation.............................................................................................172 Figure 7-3 Detailed View of Castleman Creek Channel Defined by TIN.................177 1 1 INTRODUCTION The design, construction, and maintenance of highway drainage structures are major expenditures for the Texas Department of Transportation (TxDOT) every year. One aspect of this design process involves a determination of the quantity of water expected to be conveyed by each structure. The peak flows associated with an extreme storm event can cause flooding of the areas adjacent to the structure and road. As practiced currently, hydrologic modeling is often used to calculate the quantity of runoff that is generated for each rainfall event that occurs in a particular watershed. Hydraulic modeling is also used to determine the water surface profiles that can be expected from the runoff calculated as a result of hydrologic modeling. Evaluating the resulting floodplain is a time-consuming process that, in the past, has been accomplished by manually plotting the extent of the floodplain on paper maps. Automating this process, with the aid of geographic information systems (GIS), could result in significant time and resource savings in the design process. 1.1 Background From 1996 to 1999, TxDOT funded the Center for Research in Water Resources (CRWR) at the University of Texas at Austin to develop hydrologic and hydraulic modeling tools for the purpose of floodplain delineation at highway river crossings. From 1999 to 2000, TxDOT funded CRWR to implement those tools on two case study projects. This research, supported and funded by TxDOT, investigates the possibility of combining existing GIS-based development tools, lumped parameter hydrologic and one-dimensional hydraulic models, and the visual 2 display capabilities of GIS to overcome the historical limitations of floodplain mapping. The focus of this thesis is the implementation of the above methodology at two existing TxDOT highway drainage structures to determine if site-specific data available at the district level, combined with state and national digital data, are sufficient to produce an accurate representation of the floodplain resulting from selected design storm events. 1.2 Site Selection The selection of the two study areas was made jointly by TxDOT engineers and researchers at CRWR in 1999. To evaluate the floodplain delineation methodology adequately, sites were selected that were comprised of differing watershed, channel, and drainage structure characteristics. Based on these requirements, the Castleman Creek watershed, located in McLennan County, Texas, and the Pecan Bayou watershed, located in Brown County, Texas, were selected. 1.2.1 CASTLEMAN CREEK The Castleman Creek watershed is located just south of the town of Robinson, Texas and drains to main and relief bridge structures on US Highway 77 (US 77) (Figure 1-1). 3 Figure 1-1 Location Map of Castleman Creek Watershed This watershed encompasses approximately 20.1 square miles (sq mi), or 52.1 square kilometers (sq km), and is bisected by Interstate 35 (I-35) near the small community of Hewitt, Texas. It contains three primary waterways; Crow Creek and Chambers Creek flow into Castleman Creek as it exits the watershed and proceeds to the Colorado River. Two Soil Conservation Service (SCS) flood control dams are located on Crow Creek and one flood control dam is located on Castleman Creek, although there are no flow gages associated with any of the creeks in the watershed. Agricultural land use dominates the watershed except for areas immediately adjacent to US 77, I-35, and the town of Hewitt. TxDOT is currently in the design phase of the main and relief bridge structures at US 77. Four cross-sections have been surveyed on Castleman Creek that cover approximately 400 meters upstream of the main bridge, and four cross- sections are also available that cover approximately 330 meters downstream of the bridge. High-resolution photogrammetrical survey data is also available for US 77 McClennan County 4 and areas immediately adjacent to the east and west of the highway as it traverses the creek crossings. In early 1999, TxDOT completed two HEC-1 hydrologic analyses on the watershed; a SCS Type 2 storm was applied to the watershed and routed to the bridge without considering the effects of the flood control structures and again when considering the effects of the dams. TxDOT also completed three HEC-RAS analyses (prior to the completion of this research) on the watershed: 1) without considering the effects of the dams on the existing bridges; 2) with considering the effects of the dams on the existing bridges; and 3) with considering the effects of the dams on the proposed bridge upgrades. This information provides an excellent opportunity for a comparison of the effects of the spatial variability of the watershed characteristics on the hydrologic response of the system, and provides adequate data for floodplain delineation using existing cross-sectional data. 1.2.2 PECAN BAYOU The portion of the Pecan Bayou watershed selected for investigation in this research encompasses the city of Brownwood, Texas (City) and lies directly downstream of Lake Brownwood, a water supply and flood control structure that discharges to Pecan Bayou as it proceeds to the Brazos River (Figure 1-2). 5 Figure 1-2 Location Map of Pecan Bayou Watershed A USGS flow gage site is located on Pecan Bayou near Mullin, Texas (downstream of the Brown County line), which records the flow in Pecan Bayou on 15-minute intervals. This watershed, as delineated from Lake Brownwood to the USGS gage station, is approximately 515.2 sq mi (1334.4 sq km), and consists of many small creeks, including Adams Branch, Willis Creek, and Delaware Creek, which all flow into Pecan Bayou above the main structure of interest, the bridge located on FM 2126, just south of the Atchison, Topeka, and Santa Fe rail line running southeast from the city. The watershed contributing flow to the bridge at FM2126 is significantly smaller at 168.4 sq mi (436 sq km). Lake Brownwood is an uncontrolled release reservoir, with a known stage-discharge curve for the reservoir spillway. Only one precipitation gauge, located approximately 26 mi (41 km) southwest of Mullin, TX has adequate historical data for use in this project. The portion of the Pecan Brown County 6 Bayou watershed upstream of FM 2126 is comprised primarily (70%) of rangeland, although in the vicinity of Brownwood, urban land use dominates the landscape. The bridge at FM 2126 has seen significant flooding, evidenced by flood records for severe precipitation events in 1991 and 1992. The City and TxDOT are interested in obtaining a reliable model of the floodplain based on existing river stage data from those storms. The City of Brownwood provided Microstation ? files for the majority of the city that supply two-foot contours of the terrain; the U.S. Army Corps of Engineers (USACOE) has also conducted flood studies on Pecan Bayou and made cross-section data available for use on the project. 1.3 Objectives The objectives of this project are three-fold: 1. To implement a seamless methodology for floodplain delineation in the digital domain using modified ArcView GIS scripts and software along with public-domain hydrologic and hydraulic modeling packages (HEC-HMS and HEC-RAS, respectively). 2. To evaluate the applicability of these tools on small-scale applications such as areas immediately adjacent to highway river crossings. 3. To evaluate the availability of digital and site-specific data available within, and external to, TxDOT at a resolution adequate for accurate floodplain delineation. The methodology developed to meet these objectives is comprised of six steps, summarized in Table 1-1. The synthesis of the tools presented above yields a 7 physical representation of the effects of flooding on areas immediately adjacent to the highway drainage structures being evaluated, and can be used to supplement floodplain planning and emergency response activities for TxDOT along with local and regional planning agencies. Table 1-1 Summary of Floodplain Delineation Methodology Methodology Software Description Terrain Data Development ArcView, CRWR- PrePro, CRWR- FloodMap Consists of terrain development for hydrologic analysis in raster and vector domains, and terrain development for hydraulic analysis using triangular irregular networks (TINs) GIS-Based Hydrologic Parameter Extraction ArcView, CRWR- PrePro Extracts spatially variable hydrologic parameters from GIS for export to HEC- HMS. GIS-Based Hydraulic Geometry Extraction ArcView, HEC- GeoRAS Extracts topographic information from a terrain TIN and provides data as input to HEC-RAS. Hydrologic Modeling HEC-HMS Lumped model using basin data imported from CRWR to produce discharge hydrograph Hydraulic Modeling HEC-RAS One-dimensional hydraulic model that generates water surface profile for design storm. Floodplain Delineation ArcView, HEC- GeoRAS Imports water surface profiles from HEC- RAS and displays the floodplain in GIS. 1.4 Organization The research presented in this thesis provides a realistic view of the applicability of integrating GIS analysis and display capabilities with hydrologic and hydraulic modeling tools for floodplain delineation and visualization. This chapter provides an introduction to the study areas and identifies the objectives of the research. Chapter 2 investigates historical work and other technical literature related 8 to GIS-based hydrologic and hydraulic modeling for floodplain delineation. Chapter 3 presents a discussion on the raw data used during the project, while Chapter 4 focuses on the methodology behind the application of each set of tools used in this research, including ArcView GIS, HEC-HMS, HEC-RAS, and HEC-GeoRAS. Chapter 5 presents the site-specific implementation procedures followed to generate a floodplain map for each project location. Chapter 6 contains a discussion of the results of the floodplain modeling and compares the results to existing data where available, and Chapter 7 provides conclusions and recommendations. An invaluable part of the modeling process is the necessity to have access to complete sets of data for each natural system that is to be modeled to fully understand and be able to apply the results obtained. Appendix A provides a data dictionary for both implementation sites that specifies the type and origin of each dataset used, while Appendix B presents a list of ArcView scripts utilized in the methodology. Appendices C and D provide existing hydrologic and hydraulic data and modeling results for Castleman Creek and Pecan Bayou, respectively. 9 2 LITERATURE REVIEW As digital terrain data becomes more readily available (with increasing accuracy and resolution) and computer processing become more efficient, the role of GIS in hydrologic and hydraulic modeling will continue to expand. At the present time, significant work has been accomplished to represent water surface elevations generated from hydrologic and hydraulic models in a three-dimensional terrain model, thereby providing the user with a representation of the spatial extent of the floodplain resulting from a particular precipitation event. This chapter investigates historical data and literature related to GIS-based terrain analyses, and has been divided into several sections to parallel subsequent discussions in the text. The first section addresses development of digital terrain models (DTMs) to represent the land surface in a GIS platform. GIS-based preprocessors used to represent the spatial variability of the hydrologic parameters of a watershed are then presented, followed by hydraulic modeling processes that are discussed in light of the use of GIS to extract channel properties from the land surface. Floodplain mapping in GIS is the final topic presented. 2.1 Digital Terrain Models Much emphasis has been placed on the development of distributed and lumped models to represent complex land-water interactions in last 30 years (Azagra, 1999). As currently practiced, the primary limitation in accurate floodplain mapping may not be found in representing these interactions, however, but rather in the existence of accurate DTMs that can be obtained cost-effectively. 10 GIS uses a DTM to describe the spatially distributed attributes of the terrain. DTMs describe the topography of the terrain by defining the elevation surface, while the geospatial data needed to define the ?connectivity? of each terrain element is represented by the topology of the DTM (i.e., points make up a line, lines make up an area). A DTM can be defined with the following data formats: !"Raster (grid) data; !"Vector (point, line, polygon) data; and !"Triangular irregular networks (TINs). The most common type of raster data used for DTM development is a digital elevation model (DEM). The popularity of DEM data is attributed in part to cost- effective access to the data, a complete coverage of the contiguous United States at various resolutions (i.e., 1 arc-second, 3 arc-second), and ever-increasing capabilities of GIS to process the data. The United States Geological Survey (USGS) provides DEMs for the United States and various countries around the world ? at this time, DEM selection for a particular application is generally driven by data availability as opposed to quality and resolution (Garbrecht and Starks, 1998). Currently, the automated extraction of topographic parameters from DEMs in GIS is recognized as a viable alternative to traditional surveys and manual evaluation of topographic maps, particularly as the quality and coverage of DEM data increases. Garbrecht and Martz (1999) provide commentary on the production, availability, quality, resolution, and capabilities of DEMs with respect to the derivation of topographic data in support of hydrologic and water resources investigations. Surface drainage, channel networks, and drainage divides can also be extracted from DEMs in the GIS domain (Jenson and Domingue, 1988; Martz and Garbrecht, 1992). Development of DEM 11 data is often necessary in very flat areas to create a topologically correct representation of the land surface. O?Callaghan and Mark (1984) and Jenson (1991) have demonstrated techniques for locating and removing depressions in gridded DEM data. Vector data consists of discrete spatial features, such as stream networks, elevation contour lines, or polygons representing areas with similar topographic or topologic properties. Digitized channel network data can be obtained from the USGS in the form of Digital Line Graphs (DLGs) or from the EPA in the form of RF1 and RF3 river reach files. This hydrographic coverage provides information on flowing water, standing water (lakes), and wetlands. More recently, the National Hydrography Dataset (NHD) has been developed by the USGS and EPA as a more complete vector representation of these waterways. The NHD is based upon the content of DLG hydrography data integrated with reach-related information from RF3 data (USGS, 1999). The United States Department of Agriculture (USDA) Natural Resource Conservation Service (NRCS) provides information on land use and soil properties in the form of polygon coverages for the entire country. Ragan (1991) developed a personal computer-based GIS named GIS-HYDRO to assemble predetermined land use, soil, and slope data clipped within a user-defined boundary. Look up tables relating land use types (classified by the Anderson system) to establish curve numbers for the USDA Soil Conservation Service (SCS) have been proposed by Maidment (1993) to further develop vector-based data for hydrologic analysis purposes. Triangular irregular networks (TINs) are a collection of irregularly spaced points connected by lines. Delauney triangulation is most often used to generate 12 TINs, and is based on the principal of maximizing the minimum angle of all triangles produced by connector lines to nearest neighbor points (Lee and Schacter, 1980). Breaklines are used to control the smoothness and continuity of the surface ? these lines can represent such features as ridgelines, riverbanks, or roads. TINs can be generated from raster data, vector data, or a combination of both. Long (1999) describes techniques that can be used to develop quality TINs for automated floodplain delineation with HEC-RAS using ArcInfo methodology based on cross- section data in HEC-2 format. Elevation contour data generated from orthophotography has been successfully used to generate accurate elevation TINs in ArcView as part of a flood inundation study at Vandenberg Air Force Base (Buntz, 1998), and in a similar study, elevation contour data was successfully imported into GIS from a computer aided drafting (CAD) platform in .dxf file format at Uni?o da Vit?ria City, Paran?, Brazil (de Camargo, 2000). The use of coarse DEM surfaces is generally not suitable for the large-scale terrain representation required for hydraulic analysis of river channels (Tate, 1999) ? because they cannot vary in spatial resolution, DEMs may poorly define stream channels in areas of complex relief (Carter, 1988). For this reason, the hydraulic modeling of river channels may best be accomplished using TINs. TINs allow for a dense network of points where the land surface is complex and detailed, such as river channels, and for a lower point density in flat or gently sloping areas. GIS provides the links between the discrete data formats presented above by geo-referencing the spatial data, thus creating a DTM that can be used to facilitate spatially variable land-water interactions. Speight (1980) provides a complete list of spatially variable land surface attributes that can be derived from a DTM. 13 2.2 GIS in Hydrologic Modeling ?The use of computers in hydrologic analysis has become so widespread that it provides the primary source of data for decision making for many hydrologic engineers. Since so much of hydrology is linked to processes at the earth?s surface, the connection to the topographic, computer-based methodology of GIS is a predictable step in the evolution of hydrologic engineering? (DeVantier and Feldman, 1993). Once an acceptable DTM has been adequately defined, the spatial variability of the terrain and corresponding hydrologic parameters can be evaluated in a GIS for use in event-based and continuous hydrologic models. DeVantier and Feldman (1993) present a summary of past efforts in using DTMs and GIS to perform hydrologic analyses using grid, vector, and TIN data. The first application of GIS in hydrologic modeling utilized grid cell storage of information (Pentland and Cuthbert, 1971). Since then, spatial analysis capabilities in GIS have increased tremendously. In 1989, Cline et al developed an AutoCAD ? based watershed information system to extract and calculate the data necessary to create HEC-1 input files for a sample watershed in Idaho. Jensen and Domingue (1988) and Jensen (1991) present a methodology to delineate watershed boundaries and stream networks based on gridded elevation data to defined outfalls. The scheme uses the eight-direction pour-point model to define surface water flow from each cell in a grid to one (and only one) of its eight neighboring cells according to the path of steepest decent. The cells contributing flow to the outfall can be counted to represent drainage area, and cells with no contributing flow define the watershed boundaries. Tarboton (1997) developed a similar procedure that represents flow direction as a single angle taken as the steepest downwards slope on 14 the eight triangular facets centered at each grid point. Martz and Garbrecht (1992) present a set of ten algorithms to automate the determination of drainage network and subcatchment areas from DEMs. These algorithms perform such tasks as: DEM aggregation; depression identification and treatment; relief incrementation of flat areas; flow vector determination; watershed boundary delineation; drainage network and subcatchment area definition and systematic indexing; tabulation of channel and subcatchment area properties; and evaluation of drainage network composition. Procedures for delineating streams and watersheds from DEMs can be found in Maidment (1997), Meijerink et al (1994) and ESRI (1992). The advantage of using GIS in hydrologic modeling is to provide spatially derived hydrologic parameters (such as watershed area, curve number, gridded precipitation, flow length in each watershed, and slope) for input into more powerful hydrologic models. Some of the earliest work by HEC related to GIS hydrology involved the development of a systematic methodology for automating the data preparation process in grid format (Davis, 1978). Grid-based GIS is a very suitable tool for hydrologic modeling, mainly because ?raster systems have been used for digital image processing for decades and a mature understanding and technology has been created for that task? (Maidment, 1992). Stuebe and Johnston (1990) present a comparison of rainfall-runoff relationships calculated in GRASS (a USACOE grid- based hydrologic analysis system) to GIS-based watershed delineation and runoff routing procedures using the SCS Curve Number method, with results indicating that GIS is an acceptable alternative to the conventional rainfall-runoff method for watersheds lacking relatively flat terrain. This work has led to the development of several procedures for calculating spatially variable hydrologic parameters from 15 DEMs. Maidment (1993) outlines a conceptual grid model that incorporates flow direction and a runoff velocity field to develop unit hydrographs from isochrones. The Watershed Delineator extension to ArcView GIS (Djokic et al 1997, ESRI 1997) was developed to delineate watersheds to a point, line segment, or polygon, selected interactively by the user from a map. Similarly, Hellweger and Maidment (1997) present a procedure called HEC-PREPRO that automates the translation of data from a GIS to a hydrologic data structure used by lumped parameter hydrologic modeling programs. Olivera et. al. (1998) developed CRWR-PrePro (a more recent version of which is used in this thesis) that combines the terrain analysis capabilities of the Watershed Delineator with hydrologic parameter calculation capabilities and the topologic capabilities of HECPREPRO to conform a hydrologic modeling tool that prepares ? from readily available digital spatial data ? the input file for the HEC- HMS basin component (Olivera, 1999). Doan (1999) demonstrated that the development of a hydrologic model in HMS is practical with the aid of GIS software and spatial data by implementing CRWR-PrePro on the Buffalo Bayou watershed near Houston, Texas. HEC is currently undergoing the development of public- domain software to integrate GIS capabilities with the HEC-HMS rainfall-runoff model based on the tools discussed previously. 1 This software, known as HEC- GeoHMS is scheduled to be released in the latter part of 2000 or early in 2001. 2 There have been several applications of GIS-based hydrologic preprocessors developed primarily to support the design of highway drainage structures, the focus 1 Personal communication with David Maidment, Center for Research in Water Resources, University of Texas at Austin on August 20, 2000. 2 Personal communication with David Maidment, Center for Research in Water Resources, University of Texas at Austin on October 18, 2000. 16 of this thesis. GISHYDRO was developed and installed in the Maryland State Highway Administration?s (MSHA) Division of Bridge Design in 1991 (Ragan, 1991). This GIS permitted the user to assemble the land use, soil, and slope for any watershed in the state and interface with the SCS TR-20 rainfall-runoff model. The Hydrologic Data Development System (HDDS) was developed by Smith (1995) as a set of integrated ARC/INFO programs that utilize readily available spatial data to define drainage basin boundaries, areas, maximum flowpath length, estimated travel time, slope, soil group, rainfall, and runoff coefficients at catchments defined by a highway river crossing (Olivera, 1999). 2.3 GIS in Hydraulic Modeling The advantage of using GIS in hydraulic modeling is the potential for extracting topographically correct cross-section data from a DTM that can be used to determine river stage and floodplain extent as calculated in hydraulic modeling software packages. Beavers (1994) commenced the initial work to link hydraulic modeling data and GIS. ARC/HEC2, an interface between the HEC-2 hydraulic model and the ArcInfo GIS system, extracts channel geometry from elevation contour data and utilizes user-supplied information such as Manning?s roughness values and channel contraction/expansion coefficients for export to HEC-2. Upon completion of the water surface elevation calculations in HEC-2, an ArcInfo TIN coverage of the floodplain is produced. In 1997, data exchange modules were developed for HEC-RAS (by Thomas Evans) to permit the transfer of physical element descriptions to GIS software 17 (HEC, 1997). These modules enable a user to import cross-section locations as three-dimensional coordinates (XYZ) from DTMs to develop channel and reach geometry. This work was related to Beavers? 1994 work, but permitted data exchange between ArcView GIS and HEC-RAS, the successor to HEC-2, with improved graphical user interface (GUI) capabilities. In 1998, ESRI translated and improved Evans? AML code and added GUIs in the GIS environment, resulting in an ArcView extension called AVRAS. Azagra (1999) utilized AVRAS on the Waller Creek watershed in Austin, Texas (encompassing approximately 5.8 square miles) using a TIN provided by the local municipality. Topographic information was extracted from the TIN and imported as channel geometry for use in a HEC-RAS hydraulic model, resulting in an adequate representation of the floodplain for the 100-year storm event. De Camargo (2000) also utilized the AVRAS methodology to compare modeled floodplain results to actual flood events in Uni?o da Vit?ria City, Paran?, Brazil. In 1999, AVRAS was released as a commercial product by Dodson & Associates, Inc. under the trade name of GIS StreamPro. Kraus (1999) presents a methodology to extract channel geometry data from a TIN using GIS StreamPro on a watershed covering approximately 3.42 square miles. Also in 1999, HEC released HEC-GeoRAS as the public-domain version of GIS StreamPro. Ackerman et al (1999) present the development of HEC-GeoRAS as an interface between ArcInfo and HEC-RAS. This specific version of GeoRAS uses ArcInfo to develop geometric data for import into HEC-RAS using a TIN as the basis of a DTM, and allows the user to view exported water surface profile data. 18 A similar methodology for linking GIS DTM capabilities with hydraulic modeling software was presented when the Danish Hydraulic Institute (DHI) released MIKE 11 GIS in 1999. This software provides an interface between the world?s most widely applied dynamic modeling tool for rivers in channels (MIKE 11) and ArcView GIS. To develop a MIKE 11 GIS application, essential information comprising a MIKE 11 river network, a MIKE 11 hydraulic simulation, and a DEM is required. The MIKE 11 river network is geo-referenced in MIKE 11 GIS, and when combined with water surface elevation data, can produce several types of flood maps. 3 Because of the reliability of these methodologies on the existence of TINs or high-resolution DEMS to provide accurate channel geometry, further work has been undertaken to integrate readily available field-surveyed cross-section data and lower- resolution 30-meter DEM data defining the surrounding terrain. Tate (1999) developed Avenue scripts for ArcView GIS called CRWR-FloodMap to integrate field-surveyed stream geometry within a floodplain from a HEC-RAS model into a GIS-based DTM, generated from digital orthophotography, on Waller Creek in Austin, TX. Andrysiak (2000) applied Tate?s scripts to evaluate a 165 square mile watershed along Beargrass Creek near Cincinnati, Ohio using a DEM with 30-meter accuracy. 3 Danish Hydraulic Institute web site: http://www.dhisoftware.com/mike11/Description/MIKE_11_GIS.htm. Accessed: August 24, 2000. 19 2.4 Synthesis of GIS-Based Hydrologic and Hydraulic Modeling for Floodplain Delineation The majority of the work completed to date focuses on the use of GIS-based pre- and post-processing methodologies applied to either hydrologic or hydraulic models to reproduce actual conditions. However, few models have been developed that investigate the effects of spatially variable hydrologic and hydraulic parameters on flow hydrographs, water surface profiles, and the floodplain extent resulting from recorded storm events. The Pecan Bayou watershed model developed in this thesis addresses this issue. In addition, the GIS-based hydraulic modeling performed to date has evaluated floodplain delineation on watersheds in excess of 3 square miles. The floodplain model developed for the Castleman Creek watershed focuses on the feasibility of the HEC-GeoRAS methodology at highway river crossings, where detailed terrain data in the form of highly accurate TINs or DEMs may not be readily available. The extent of the modeled area upstream and downstream of the river crossing (approximately 0.5 river miles) encompasses approximately 4000 acres (0.1 sq mi). 20 3 DATA The data used during the development of the floodplain delineation models at each implementation site was obtained from a variety of public agencies. Because of the multitude of sources, the data was also provided in several different projections. In order to utilize this information within GIS, all spatial data was converted to a common map projection. The projection used throughout this project (for both Castleman Creek and Pecan Bayou) was defined by the Texas State Mapping System parameters: PROJECTION ALBERS UNITS METERS PARAMETERS 1 ST STANDARD PARALLEL: 27 25 0.00 2 ND STATNDARD PARALLEL: 34 55 0.00 CENTRAL MERIDIAN: -100 0 0.00 LATITUDE OF PROJECTION?S ORIGIN: 31 10 0.00 FALSE EASTING (METERS): 1000000.00 FALSE NORTHING (METERS): 1000000.00 This chapter is organized by location, with each subsection identifying the raw data available at each site, and the data development activities necessary to ensure the homogeneity, spatial connectivity, and completeness of each dataset. A data dictionary that documents the properties of each dataset is included in Appendix A. 21 3.1 Castleman Creek The data available for use in the Castleman Creek watershed consisted of raw data sources and several datasets that required further development to become usable inputs to modeling activities. 3.1.1 RAW DATA The raw data used as inputs for model development at the Castleman Creek site consisted of terrain data (DEM and high-resolution photogrammetric survey data along US 77), stream network data, regional regression peak flow data, existing HEC-HMS and HEC-RAS project information, and infrastructure data (TxDOT road coverages). 3.1.1.1 Terrain data A digital representation of the terrain at the Castleman Creek site was developed primarily from the USGS National Elevation Dataset (NED). This dataset provided seamless raster elevation data at a scale of 1:24,000 in one-degree blocks. The data was originally digitized from existing contour information and provided elevation data in one arc-second (approximately 30 meter) resolution. The NED was provided in a geographic projection (with units of decimal degrees) according to the NAD83 horizontal datum, and yielded elevation data in units of decimal-meters. At this site, the NED ID 9832 was sufficient to cover the entire watershed ? therefore, merging adjacent grids was not necessary. Aerial photogrammetric survey data describing US 77 and the areas immediately adjacent to it was also available (Figure 3-1). This elevation data was provided by TxDOT as a point elevation theme in the Texas State Plane ? Zone 14 projection 22 using the NAD83 horizontal datum and the NGD29 vertical datum. The source of the data was unknown. Figure 3-1 TIN Elevation Data for US 77 in Castleman Creek Watershed 3.1.1.2 Stream Network The stream network at the site was defined by the NHD, a feature-based dataset that interconnects and uniquely identifies the stream segments or "reaches" that make up the Nation's surface water drainage system. It is based upon the content of USGS Digital Line Graph (DLG) hydrography data integrated with reach- related information from the EPA Reach File Version 3 (RF3) 4 , and provides not only river reach data, but water body coverages as well. The NHD is currently based on the content of the USGS 1:100,000-scale data, giving it accuracy consistent with those data. Data for this project was provided in geographic coordinates (with units 4 National Hydrography Dataset web site: http://nhd.usgs.gov. Accessed: August 20, 2000. 23 of decimal degrees) on the North American Datum of 1983. 5 The Castleman Creek network includes approximately 51 km. of waterways within the 52.1 sq. km. watershed, but is only comprised of three waterways and as such is well defined by the NHD and required no modifications. Figure 3-2 provides a representation of the Castleman Creek Stream Network shown as projected into the Texas State Mapping System at a scale of approximately 1:12,000. Figure 3-2 Castleman Creek Stream Network 3.1.1.3 Precipitation Data Due to the size and the rural location of the Castleman Creek watershed, no flow gages were available for rainfall-runoff calibration and gauged precipitation data 5 National Hydrography Dataset web site: http://mapping.usgs.gov/mac/isb/pubs/factsheets/fs10699.html. Accessed: August 20, 2000. 24 was not necessary for development of the model at this site. However, SCS Type 2 synthetic storms were applied to the watershed to model rainfall-runoff relationships. This data was obtained from existing HEC-1 models previously developed by the TxDOT Bridge Hydraulics Division in Austin, TX. For use in this research, these synthetic storms were extracted from the HEC-1 model for use in modeling the watershed using CRWR-PrePro and HEC-HMS. This data consists of cumulative 15-minute interval precipitation measured to 0.001 inches over a 24-hour period for storm return periods of 2, 5, 10, 25, 50, and 100 years. Appendix C.1 presents the original synthetic storm data developed for the Castleman Creek site. 3.1.1.4 Flood Control Structure Data The Castleman Creek site includes three SCS flood control structures ? all of which lie upstream of the US 77 river crossing. SCS-1 is found on Castleman Creek, while SCS-2 and SCS-3 are found on Crow Creek. Figure 3-3 depicts the locations of the flood control structures as shown on the USGS DOQQ. Figure 3-3 Locations of SCS Flood Control Structures in Castleman Creek Watershed 25 The elevation-storage-discharge relationships for all of these structures were incorporated into the original HEC-1 model developed by TxDOT; this data was extracted for use in HEC-HMS to simulate rainfall-runoff relationships. Appendix C.1 presents the original elevation-storage-discharge data used by TxDOT personnel to develop the HEC-1 model. 3.1.1.5 Flow As stated previously, no historical flows were recorded at the Castleman Creek site, so calibration of the model was not possible with recorded flow data. Standard TxDOT practice is to model such watersheds without any flow control structures and compare the resulting data to regression equations developed to estimate peak flows based on historical data. 6 Regional regression equations have been developed by the USGS in 1993 to estimate the magnitude and frequency of floods for ungaged sites in six separate regions with Texas based on the area and slope of the watershed of interest. 7 Additional regression equations were developed specifically for Texas in 1997 by the USGS (in cooperation with TxDOT) that considered regionally variant conditions not considered in the 1993 equations. 8 These equations considered watershed shape factor in addition to the two parameters mentioned previously, and differentiated expected peak flows based on watershed size. Figure 3-4 presents a comparison of the 1993 and 1997 regional regression equations for Castleman Creek. 6 Personal communication with David Stolpa, TxDOT, on May 12, 2000. 7 Jennings et. al., Nationwide Summary of U.S. Geological Survey Regional Regression Equations for Estimating Magnitude and Frequency of Floods for Ungaged Sites, 1993 ? Water-Resources Investigations Report 94- 4002. 8 Asquith, W.H. and R.M. Slade, Regional Equations for Estimation of Peak-Streamflow Frequency for Natural Basins in Texas, 1997 ? Water-Resources Investigations Report 96-4307. 26 Figure 3-4 Comparison of Regional Regression Equations for Castleman Creek 3.1.1.6 Infrastructure A coverage of Texas roads, developed by TxDOT in the Texas Statewide Mapping System (TSMS) projection (in NAD27 format with units of feet), was clipped to the extents of the Castleman Creek watershed and utilized to identify the location of the US 77 main and relief bridges. 3.1.1.7 Existing Models Both HEC-1 and HEC-RAS models have been developed previously for the Castleman Creek site by TxDOT personnel. This section presents a summary of the data available for use in the development of floodplain delineation models at both sites. 27 3.1.1.7.1 HEC-1 The status of the bridge modification project at the Castleman Creek crossing is ongoing, and this project was undertaken to further understand the impact the modifications may have on the stage and spatial extent of flooding resulting from extreme precipitation events. Two HEC-1 models were developed for Castleman Creek in 1990 ? one evaluated peak flow as a result of SCS Type 2 synthetic storm events applied over the watershed without considering the effects of the flood retarding structures. This was performed to facilitate a comparison of the results to the regional regression equations. The other model considered the effects of the flood control structures on the expected peak flow at the US 77 Bridge. The input data for these models consisted of 15-minute cumulative precipitation data for storm return periods of 2, 5, 10, 25, 50, and 100 years (the discharges associated with these storms are presented in Chapter 6). Watershed areas were provided as inputs into the model based on delineation activities using USGS 7.5? quadrangle maps (quads). An SCS lag-time (in hours) and curve number were also provided for each watershed, along with Muskingum routing parameters (K, X) specified for each routing reach. These were developed based on the protocol presented in the TxDOT Bridge Division Hydraulic Manual, with revisions dated June 6, 1986. When the flood control structures were considered, the storage- elevation-discharge relationships were provided for each of the three SCS flood control structures. Finally, a schematic of the stream network was provided which yielded the connectivity of the network as understood from the USGS 7.5? quads. Appendix C.1 provides the input and output files for both HEC-1 runs. 28 3.1.1.7.2 HEC-RAS A one-dimensional HEC-RAS hydraulic model for the Castleman Creek site was developed by TxDOT personnel prior to the implementation of this project. Several different scenarios were employed to estimate the peak stage at the bridge for the six storm return periods noted previously. Models were developed without consideration of the flood control structures, with consideration of the flood control structures at the existing bridge, and with consideration of the flood control structures at the proposed bridge upgrade. This study focuses on the results of the HEC-RAS model that accounted for the effects of the flood control structures on the proposed bridge modifications. Five cross-sections were surveyed upstream of the bridge and four cross- sections were surveyed downstream of the bridge. Figure 3-5 presents a schematic of the original cross-sections. Figure 3-5 HEC-RAS Schematic of Surveyed Cross-Sections on Castleman Creek 29 The most-upstream cross-section (Station 2000) is located approximately 402 meters upstream of the bridge, and the most-downstream cross-section (Station 1100) is located approximately 332 meters downstream of the bridge. It is evident that the distance between the most upstream and downstream cross-section stations (900 m) does not match the reach lengths detailed in the existing HEC-RAS model (734 meters); this research assumes that the reach lengths provided in the HEC-RAS model were more accurate. Each cross-section provided data on the downstream reach lengths at the channel and left- and right-overbanks. Similarly, Manning?s Roughness coefficients were provided for the channel, and left- and right-overbanks. Finally, expansion and contraction coefficients were provided for each cross-section. The original sketches of the cross-section orientations were also made available. The bridge crossing was modeled as a ?multiple opening? due to the presence of a flood relief bridge south of the primary bridge structure. Geometric data was provided for each bridge in the HEC-RAS model that detailed the deck width, distance to upstream cross-section, and weir coefficient in the event that the peak flow stage overtopped the structure. Data was also provided on the high and low chord elevations for each bridge deck, along with the upstream and downstream embankment side slopes. The bridge was modeled using the energy approach for both low and high flows. Steady-flow boundary conditions were also assumed in the HEC-RAS model ? normal depths were assumed at both the upstream and downstream cross-sections, with corresponding water surface slopes of 0.00084 and 0.00239, respectively. Appendix C.2 presents a summary of this data. 30 3.1.2 DATA DEVELOPMENT While the data discussed in Section 3.1.1 encompasses the raw data that was used for the Castleman Creek model, modification was required for a number of data sets to facilitate accurate floodplain delineation models. This section addresses modifications to the terrain and stream networks, and the development of a curve number grid necessary to employ the SCS curve number method to determine rainfall-runoff relationships. Site-specific data development, necessary to populate the HEC hydrologic and hydraulic models with spatially derived terrain data, is addressed in Chapter 4. 3.1.2.1 Terrain The raw DEM covering the Castleman Creek site did not adequately represent the rise in elevation due to the presence of US 77, and therefore would produce erroneous results during watershed delineation activities. The solution to this problem was to create an artificial wall along the highway, forcing any runoff to be routed parallel to the highway until it reached the main and relief bridge structures. Unfortunately, the TxDOT road coverage did not match the more accurate coordinates provided by the aerial photogrammetric survey along the highway, nor did it match the representation of the roads provided in the Digital Orthophoto Quarter Quadrangles (DOQs) for McClennan County. Figure 3-6 depicts the discrepancies between the three data sets. 31 Figure 3-6 Discrepancies between DOQ, Photogrammetric Coordinates, and Texas Road Coverage The green points present the results of the photogrammetric survey overlain on the DOQ and the black line the TxDOT road coverage supplied by TxDOT. It is evident from this figure that the DOQ and the photogrammetric data correlate well with each other. The light blue line was manually added to match US 77 as shown in the DOQ, and was converted to a grid (with an elevation attribute of 10,000 meters) and merged with the surrounding DEM to ensure the no runoff would overtop the road during watershed delineation calculations (Figure 3-7). 32 Figure 3-7 Artificial Wall Integrated in the DEM along US 77 3.1.2.2 SCS Curve Number The development of a curve number grid at the Castleman Creek site was necessary to estimate the spatial variability of runoff resulting from a precipitation event that would be subsequently used in HEC-HMS to calculate the discharge hydrograph at US 77. The SCS curve number method calculates the quantity of precipitation falling onto the land surface that is converted to runoff based on the depth of precipitation, the potential maximum soil moisture storage after runoff begins, and an estimate of the initial quantity of infiltration at the beginning of the storm event. The soil storage and initial abstractions can be considered a function of soil type and land use and land cover (LULC) characteristics. The curve number grid for this project was taken from the Blacklands Research Center in Temple, TX. This statewide grid was produced by combining the 33 USDA/NRCS STATSGO soil coverage with the USGS LULC coverage. A lookup table was used to translate the combinations of soil and land use into curve numbers using the 1972 SCS Engineering Hydrology Handbook as a reference. The LULC and STATSGO files are both 1:250,000 scale map products, so the resulting curve number grid was relatively coarse compared to the DEMs and stream networks at both sites. Figure 3-8 presents the curve number grid used at the Castleman Creek site. Figure 3-8 SCS Curve Number Grid Coverage at the Castleman Creek Watershed The light blue cells in this curve number grid represent a curve number of 90 ? a surface that generates significant runoff ? and correspond to the presence of Interstate Highway 35 in the central portion of the watershed. The majority of the western side of the watershed has a curve number of 85 (the dark green cells), while the eastern portion was dominated by curve numbers of 70 (the brown cells). The average curve number for the Castleman Creek watershed was calculated to be 82.6. 34 3.2 Pecan Bayou The data available for use in the Pecan Bayou watershed consisted of raw data sources and several datasets that required further development to become usable inputs to modeling activities. 3.2.1 RAW DATA The raw data used to define the Pecan Bayou watershed consisted of a DEM, contour information extracted from Microstation ? drawings provided by the City of Brownwood, stream network data, precipitation data, reservoir storage and discharge data, USGS stream gage data, and existing HEC-RAS project information. 3.2.1.1 Terrain Data At the Pecan Bayou site, the majority of the terrain was defined by 2-foot contour data provided on CD-ROM by the City of Brownwood as 152 Microstation ? drawings (Figure 3-9). The contours were derived from aerial mapping activities conducted by United Aerial Mapping on February 21, 1995. The scale of each drawing was 1 inch = 100 feet. The contours were provided in the Texas State Plane ? Zone 14 projection using the NAD83 horizontal datum (with units of meters) and the NGD29 vertical datum (with units of feet). 35 Figure 3-9 Example Microstation ? Drawing for Pecan Bayou Watershed Each drawing also contained the following additional information in the form of lines, polygons, points, and text annotations: !"Improved and unimproved roads; !"Cultural features, such as buildings, property lines, and natural landscape features; !"Existing and abandoned railroads and their associated infrastructure; !"Water features, such as wetlands, creeks, dams, and water supply and treatment infrastructure; !"Utility features; !"Vegetation features; and !"Drawing information such as latitude/longitude labels, sheet outlines, and title blocks. 36 Building footprint information was extracted from each drawing to supplement the definition of the terrain in urban areas along Pecan Bayou. Water features were also extracted and compared to stream network data to verify the orientation and connectivity of the stream network. The portion of Pecan Bayou to be modeled for floodplain delineation, between bridges located on US 183 and FM 2126, was located almost entirely within the limits of the terrain defined by the contour data. However, in the overbank areas east of the bayou, no contour data was available; in this area, elevation data derived from the NED coverage for Brown County was used. 3.2.1.2 Stream Network The Pecan Bayou stream network was defined by the NHD, but required modifications to remove pipelines, drainage ditches, and other water supply appurtenances. The Pecan Bayou flow network was defined for this study to include only the portions of Pecan Bayou and its tributaries found south of Lake Brownwood and north of USGS Gage 08143600 near Mullin, TX (Figure 3-10). The network includes 367 records representing a total stream length of over 952 km in a 1334 sq km study area. Although the NHD was adequate for modeling the hydrologic response of the system in GIS, it was evident from the detailed contour data that a more detailed stream centerline would be necessary for accurate floodplain modeling. A line theme was digitized from US 183 to FM 2126 to reflect the detailed channel centerline in this area. 37 Figure 3-10 Pecan Bayou Stream Network 3.2.1.3 Precipitation Data At the Pecan Bayou site, peak flows and stages were recorded at USGS Gage 08143600 for high flow conditions in December 1991; this data was used to calibrate the rainfall-runoff relationships and hydraulic characteristics of the Pecan Bayou channel. Unfortunately, while there were several gage stations within Brownwood and the surrounding vicinity, there was only one rainfall gage with a significant, continuous, period of record. This hourly precipitation data (recorded in 0.1-inch increments) was obtained from NOAA Cooperative Station 419817 at Winchell, TX 38 through the National Climatic Data Center (NCDC) web site. 9 Figure 3-11 presents a summary of this data. Figure 3-11 Precipitation Data Recorded at NOAA Cooperative Station 419817 at Winchell, TX in Pecan Bayou Watershed The NOAA station has an elevation of 445 meters above mean sea level (msl) and lies approximately 41 km. west-southwest of USGS Gage 08143600 (Figure 3-12). 9 National Climatic Data Center web site: http://www.ncdc.noaa.gov/ol/climate/stationlocator.html. Accessed June 15, 2000. 39 Figure 3-12 Location of NOAA Cooperative Station 419817 in Pecan Bayou Watershed 3.2.1.4 Reservoir Data Lake Brownwood was defined as the upstream flow source for the Pecan Bayou model. Lake elevation data was provided by the Brown County Water Improvement District No. 1 (BCWID) for December 1991 and January 1992. Lake elevation data (accurate to 0.1 ft.) was recorded sporadically throughout the months of interest, and therefore required interpolation to estimate hourly elevation levels. Figure 3-13 presents a summary of the recorded lake elevations during the Christmas 1991 storm event. Unfortunately, continuous lake elevation data was not available between December 21 and January 6, 1991; this data was estimated via linear interpolation between known data points. 40 Figure 3-13 Recorded Lake Brownwood Elevation Data during December 1991 Storm Event The BCWID also provided elevation-discharge relationships for flow above the spillway at Lake Brownwood, which was used to develop the inflow hydrograph for the December 1991 storm event. Appendix C.1 presents the spillway-rating curve supplied by the BCWID. 3.2.1.5 Flow Data At the Pecan Bayou site, stream flow and stage data was available at USGS Gage 08143600. This gage is located approximately 43.5 km downstream of the FM 2126 Bridge, so the watersheds contributing flow between the bridge and the gage were evaluated as part of the Pecan Bayou hydrologic model. Hourly flow and stage data recorded between November 19, 1991 and January 2, 1992 was provided by the Abilene, TX USGS office (acquired from the USGS Federal Records archives in 41 Denver, CO) in paper format. The discharge rating curve for the USGS Gage was also provided to allow interpolation of flow and stage at time intervals during which the gage was offline. The flow data was accurate to 1 cfs, while the stage data was accurate to 0.01 ft. Appendix D.1 presents the recorded flow data for December 1991 and January 1992 at USGS Gage 08143600. 3.2.1.6 Infrastructure Several additional datasets were also utilized during the development of the floodplain delineation models in the Pecan Bayou watershed. The same TxDOT road coverage used at the Castleman Creek was clipped to provide roadway data within the extents of the Pecan Bayou watershed. A point coverage of hydrologic and hydraulic gauging sites was also generated from geographical coordinates supplied by web sites designated for each gage; these are presented in Table 3-1. Table 3-1 Point Coverage of Hydrologic and Hydraulic Gauging Sites Latitude Longitude Gage Degrees Minutes Degrees Minutes USGS Gage 08143600 31 31 -98 44 NOAA Cooperative Station 419817 31 27 - 99 10 3.2.1.7 Existing Models A HEC-RAS model was developed for ongoing flood mitigation studies in the Pecan Bayou watershed by the U.S. Army Corps of Engineers (USACOE) Ft. Worth District, which provided critical channel cross-sections for Pecan Bayou from 42 Lake Brownwood to its confluence with the Colorado River. The survey dates for these cross-sections is unknown, but estimated to be in the early 1990s. 10 For the purposes of this study, only the cross-sections between US Highway 183 and FM 2126 were utilized. Figure 3-14 presents a schematic of the USACOE cross-sections on Pecan Bayou in the area of interest. Pecan Bayou flows under three bridges in the area of interest for this project. The first is at US 183, approximately 18.5 km downstream of Lake Brownwood; the second is at FM 2525 (also known as Hawkins St.), 375 meters south of the US 183 Bridge; the third is at FM 2126, approximately 8.1 km south of the FM 2525 bridge. Figure 3-15 presents the location of the three bridges of interest in the Pecan Bayou watershed (although FM 2525 does not appear to continue southwest over Pecan Bayou, there is a structure present at that location). Figure 3-14 HEC-RAS Schematic of Surveyed Cross-Sections on Pecan Bayou 10 Personal communication with Craig Lofton, USACOE, Ft. Worth District on August 1, 2000. 43 Figure 3-15 Locations of Bridges of Interest in Pecan Bayou Watershed 3.2.2 DATA DEVELOPMENT While the data discussed in Section 3.2.1 represents the raw data that was used to initially define the Pecan Bayou watershed, further development was required for a number of datasets to facilitate accurate floodplain delineation models. This section addresses data development for the Pecan Bayou stream network, development of a curve number grid, adjustment of precipitation data based on areal-reduction factors, interpolation of reservoir discharge data, and interpolation of recorded flow and stage data at USGS Gage 08143600. 3.2.2.1 Development of Stream Network The Pecan Bayou stream network was developed from the NHD and, due to the completeness of the dataset, was clipped to be more manageable in size. Figure 44 3-16 presents the full extent of the Pecan Bayou NHD route.rch file, while Figure 3-17 depicts the clipped route.rch coverage, highlighted in purple. Figure 3-16 Extent of Pecan Bayou NHD route.rch Coverage Figure 3-17 Clipped NHD route.rch Coverage 45 The structure of the NHD coverage is much more detailed than previous river network files, such as the RF3 files developed by the EPA. As shown in Figure 3-18 in the attribute table associated with the route.drain file (a sister file to route.rch), stream segments are identified by type, and some types are not appropriate for inclusion in modeling the hydrologic or hydraulic routing relationships within a watershed. Figure 3-18 Attribute Table for NHD route.drain Coverage The attribute table of the route.drain coverage was linked to the route.rch attribute table to facilitate selection of inappropriate stream reaches for inclusion in the Pecan Bayou watershed model. These reaches are highlighted in yellow in Figure 3-17 above, and were deleted from the final Pecan Bayou stream network. Once these modifications were completed, the network was overlain on a digital raster graphic (DRG) image of the watershed to ensure that the NHD coverage matched the network as portrayed on the appropriate USGS 7.5? quads. 46 DRGs are scanned images of USGS topographic maps, provided in this case at 1:24,000 scale by the USGS in UTM ? Zone 14 projection. DRGs may be used as a source or background layer in GIS as a means to perform quality assurance on other digital data. 11 Figure 3-19 presents an example of manually editing of the river network for connectivity. The blue segment inside the purple circle was not originally included in the NHD coverage and was added manually. Figure 3-19 Manual Editing of NHD route.rch Coverage with DRG 11 USGS web site: http://edcwww.cr.usgs.gov/glis/hyper/guide/drg. Accessed: August 20, 2000. 47 3.2.2.2 SCS Curve Number The same curve number grid was used on the Pecan Bayou watershed that was used at the Castleman Creek site. The original grid coverage is statewide, so the grid was simply clipped to cover the extent of the Pecan Bayou watershed. 3.2.2.3 Precipitation The precipitation data available for the December 1991 storm event at Pecan Bayou was recorded at NOAA Cooperative Station 419817 near Winchell, TX and yielded greater-than-expected flows within the Pecan Bayou channel when modeled hydrologically in HEC-HMS. Further development of this precipitation data was warranted, and the concept of areal-reduction factors was investigated. The reduction of precipitation depths from a given storm to an effective (mean) depth over a watershed often is important for cost-effective design of hydraulic structures by reducing the volume of precipitation. An effective depth can be calculated by multiplying the precipitation depth at a point by an areal-reduction factor (ARF). ARFs range from 0 to 1, vary with the recurrence interval of the storm, and are a function of watershed characteristics such as size and shape, geographic location, and time of year that the design storm occurs. 12 ARFs for Austin, Dallas, and Houston have been derived from several precipitation-station monitoring networks in the vicinity of each city, with varying periods of record. The large daily precipitation databases available in Texas allowed an approach that considered the distribution of precipitation concurrent with, and surrounding, an annual precipitation maxima. 12 Asquith, W.H., Areal-Reduction Factors for the Precipitation of the 1-Day Design Storm in Texas, 1999 ? Water Resources Investigations Report 99-4267. 48 Because NOAA Coop Station 419817 did not correspond directly with the location of the ARFs currently available for Texas, a conservative approach was used to estimate an ARF applicable to the Pecan Bayou watershed. Figure 3-20 presents the equations used to estimate the ARF for the Dallas area, which were also implemented at the Pecan Bayou site. The watershed was estimated to be circular, with a radius of approximately 13.6 miles. According to the equations provided, the ARF was estimated conservatively to be 0.67. Figure 3-20 Areal-Reduction Factor Equations for Dallas, TX 12 The incremental precipitation depths recorded in December 1991 and January 1992 were multiplied by this ARF to produce a 33% reduction in precipitation depth. Figure 3-21 presents a comparison of the original incremental precipitation depths to the adjusted values. 49 Figure 3-21 Comparison of Recorded Incremental Precipitation to Precipitation Adjusted with ARF 3.2.2.4 Reservoirs Interpolation of lake elevation and discharge data was necessary at Lake Brownwood to facilitate hourly hydrologic modeling in HEC-HMS. Unfortunately, only sporadic data was recorded during the Christmas 1991 storm. Adequate data was recorded at the beginning of the storm, but few data points were recorded after the peak elevation, assumed to occur at 9:00 am on December 21. A linear relationship was assumed to describe the lake elevation as it fell to normal elevation levels as the effects of the storm dissipated. These lake elevations were then cross- referenced with the spillway-rating curve to determine the hourly discharge from the reservoir. Figure 3-22 depicts the linear tendencies of the discharge hydrograph from Lake Brownwood. 50 Lake Brownwood Discharge Hydrograph 0.00 5000.00 10000.00 15000.00 20000.00 25000.00 30000.00 35000.00 10/29 11/8 11/18 11/28 12/8 12/18 12/28 1/7 1/17 Date Fl ow ( c f s ) Linear Interpolation Figure 3-22 Lake Brownwood Discharge Hydrograph 3.2.2.5 Flow The flow and stage data recorded at USGS Gage 08143600 near Mullin, TX was available in hourly increments for the majority of the December 1991 storm. This data was used to develop a flow-discharge relationship for the gage site (Figure 3-23). 51 Discharge Rating Curve for USGS Gage near Mullin, TX 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 0 5000 10000 15000 20000 25000 30000 35000 Discharge (cfs) Ga g e Hei g h t ( f t Figure 3-23 Rating Curve at USGS Gage 08143600 near Mullin, TX During the highest flow conditions (December 21-24), the gage was damaged and flow was recorded manually at intervals greater than one hour. The rating curve was then used to determine stage elevation and flow in the hourly increments not recorded. 52 4 METHODOLOGY The methodology presented in this chapter results in a seamless procedure for generating a floodplain at highway river crossings given digital terrain, hydrologic, and hydraulic data. Although the data sources available for each site differed, both the Castleman Creek and Pecan Bayou models were developed using the following methodology: 1. Site Specific Terrain Data Development a. Terrain Development for Hydrologic Analysis b. Terrain Development for Floodplain Delineation 2. GIS-based Hydrologic Parameter Extraction 3. GIS-based Hydraulic Geometry Extraction 4. Hydrologic Modeling 5. Hydraulic Modeling 6. Floodplain Delineation Figure 4-1 presents a schematic of this methodology. It is evident from the figure the importance of an accurate DTM, as it affects both hydrologic and hydraulic modeling activities. While CRWR-PrePro contains adequate tools for the development of the terrain for hydrologic purposes, the steps shown on the bottom of the figure (HEC-RAS ? CRWR-Floodmap ? HEC-GeoRAS) represent the detailed terrain development necessary for accurate floodplain delineation. As shown, detailed terrain development activities can occur simultaneously to hydrologic and hydraulic modeling activities ? and should ? to optimize the efficiency of the process. 53 Figure 4-1 Schematic of Floodplain Delineation Methodology The theory behind each step of the methodology is presented in this chapter; a systematic implementation procedure ? presented subsequently in Chapter 5 ? highlights the applicability of the methodology to each of the two sites selected as part of this research project. 4.1 Site Specific Terrain Data Development Although general data development activities were presented in Chapter 3 that addressed the homogeneity, spatial connectivity, and completeness of each dataset, site specific data development activities are necessary to preprocess the terrain data for use in the HEC programs for hydrologic and hydraulic modeling (HEC-HMS and HEC-RAS, respectively). 54 4.1.1 TERRAIN DEVELOPMENT FOR HYDROLOGIC ANALYSIS The procedure for processing raw raster terrain data is comprised of three conceptual modules: 1) raster-based terrain analysis, 2) raster-based subbasin and stream network delineation, and 3) vectorization of subbasins and reach segments. These activities were carried out in the GIS domain using CRWR-PrePro, a system of ArcView scripts and associated controls developed at the CRWR to extract topographic, topologic, and hydrologic information from the digital spatial data of a hydrologic system for eventual export to HEC-HMS. 13 The procedure implemented at both sites was identical, and followed the steps presented in the ArcView pull- down menu displayed in Figure 4-2. Figure 4-2 CRWR-PrePro Implementation Procedures Olivera and Maidment (1999) present an excellent discussion of DEM-based terrain analysis using CRWR-PrePro. Beginning with a DEM and a stream network 13 CRWR Pre Pro website: http://civil.ce.utexas.edu/prof/olivera/esri99/p801.htm. Accessed: August 1, 2000. 55 file (in this case, NHD files), the Burn Streams menu item is selected, which raises the land surface cells that are off the streams by an arbitrary elevation so that the streams delineated from the DEM exactly match those in NHD network file. The Fill Sinks menu item is then activated which ensures that there are no cells within the DEM that would adversely affect the flow direction of surface runoff applied over the watershed. In practice, DEM cells may contain errors that create artificially raised or depressed cells within the grid. The Fill Sinks algorithm (Figure 4-3) iteratively raises or lowers the cell to match the elevation of the lowest surrounding cell elevation. Figure 4-3 Fill Sinks Algorithm CRWR-PrePro then calculates the direction that any runoff would take on the DEM surface according to the eight-direction pour point model (Figure 4-4) and generates a flow direction grid covering the same extent as the original DEM. 56 Figure 4-4 Raster-Based Functions for Terrain Analysis The flow direction grid is determined by finding the direction of steepest descent from each cell, and is calculated as the change in elevation divided by the horizontal distance between the center of each cell. From this point, the number of cells contributing flow to one ? and only one ? downstream cell are calculated, and, if 57 multiplied by the cell area, equal the drainage area. This flow accumulation grid represents the amount of precipitation that would flow into each cell assuming that all precipitation becomes runoff (assuming no interception, evapotranspiration, or infiltration). A raster-based stream network can then be developed based on the flow accumulation grid and the definition of the minimum number of cells (and corresponding drainage area based on the grid cell size) that contribute flow to a certain point in the DEM, defined as the stream threshold. This stream threshold is user-defined, which permits the delineation of streams to match existing stream network files. CRWR-PrePro also permits the user to add streams manually to further ensure that the resolution of the resulting stream network meets the requirements of the project. The Stream Segmentation (Links) menu item is then activated, which allows the user to identify unique stream segments within the stream network. This is followed by the Outlets from Links command, which identifies the most-downstream cell on a stream segment as an outlet. This can be followed by the Add Outlets command, which permits the user to manually identify additional outlets, such as the presence of flow gage or water rights locations. Finally, with the outlets and stream network identified and the elevation in each cell known, CRWR-PrePro delineates the subbasins contributing flow to each outlet. Once the stream network and subbasin extents have been identified in the raster domain, CRWR-PrePro converts the raster data to vector format using the Vectorize Streams and Watersheds command. Subbasins can be merged as necessary using the Merge Sub-Watersheds command. This is the final step prior to the extraction of the spatially variable hydrologic parameters intrinsic to each subbasin, which is 58 addressed in Section 4.2 of this chapter. The methodology presented in this section, as it relates to the overall methodology developed in this thesis, is highlighted in Figure 4-5. Figure 4-5 Terrain Development for Hydrologic Analysis 4.1.2 TERRAIN DEVELOPMENT FOR FLOODPLAIN DELINEATION The development of terrain data at a resolution that facilitates an accurate representation of a floodplain is critical in obtaining the realistic extent of potentially impacted surface features resulting from an extreme precipitation event. The use of coarse DEM surfaces is generally not suitable for the large-scale terrain representation required for floodplain delineation activities because they cannot vary in spatial resolution (Carter, 1988). For this reason, the hydraulic modeling of river channels and the associated floodplain may best be accomplished using TINs. 59 Tate (1999) developed a system of ArcView GIS scripts called CRWR- FloodMap to import cross-sectional geometry into GIS and ultimately define the floodplain resulting from water surface profiles modeled hydraulically in HEC-RAS (Figure 4-6). Figure 4-6 CRWR-FloodMap Menu In this thesis, selected ArcView scripts created by Tate have been modified strictly to supplement existing terrain data (such as DEMs and photogrammetric survey data) and therefore provide a more accurate representation of the terrain in the channel and overbank areas (all scripts used in this methodology are presented in Appendix B.1). Thus, Tate?s floodplain mapping scripts are not used ? the floodplain mapping capabilities of HEC-GeoRAS are instead ultimately utilized (Figure 4-7) for that purpose. Parts of the following text documenting the CRWR-FloodMap methodology are excerpted from Tate?s 1999 thesis. Scripts modified for this thesis are noted as such. 60 Figure 4-7 Terrain Development for Floodplain Delineation The methodology presented subsequently assumes that cross-sectional geometry data is available in HEC-RAS (or HEC-2) format. The methodology developed by Tate, and used in HEC?s GeoRAS software, assumes that the stream centerline defined at each cross-section is connected linearly (with a straight line) to the subsequent cross-section. Therefore, there is a possibility that, for tortuous streams (streams that meander), it may be necessary to interpolate between surveyed cross-sections to adequately model the tortuous nature of the stream (Figure 4-8). As applicable to this work, this is especially important when considering small areas (less than 100 acres) prone to flooding at highway river crossings. This methodology assumes the user interpolates an adequate number of cross-sections to effectively represent the tortuous nature of the stream. Once an acceptable number of cross-sections have been defined in HEC- RAS, hydraulic model output information must be extracted and imported into the 61 GIS environment. After the HEC-RAS report has been generated, the user selects the first CRWR-FloodMap menu item Import HEC-RAS Data (shown in Figure 4-6). Figure 4-8 Surveyed and Interpolated Cross-Sections Upon activation of this menu item, the user is prompted to specify the units desired for analysis in GIS. The Import HEC-RAS Data script (modified in this thesis), called FloodRasRead, reads the HEC-RAS output file (that consists of cross-sectional geometries and reach lengths between each cross-section) and creates a table in ArcView that specifies: !"River station ID; !"A text description of the cross-section; !"Coordinates of the stream centerline, located at the point of minimum channel elevation; !"Bank station locations as measured from the stream centerline; and 62 !"Reach lengths. Figure 4-9 presents an example of the HEC-RAS output file and the corresponding table created in ArcView GIS. Figure 4-9 HEC-RAS Output File and Imported ArcView GIS Table The lateral and elevation coordinates of each surveyed cross-section point are read and stored as ArcView global variables. The coordinates of the point possessing the minimum channel elevation are also determined ? if there are multiple points with the same minimum elevation, the average lateral coordinate of all points with the 63 same elevation is used. Similarly, the distance from the centerline of a cross-section to the bank station is also identified and written to the table. The next step is to link the HEC-RAS stream representation to the digital representation of the stream in ArcView GIS. This is accomplished by the Format Digital Stream menu, which calls the FloodFormatStream script. Any vectorized representation of the stream can be used, but it must reflect the attributes of the surrounding terrain. Therefore, it may be necessary to digitize the stream from DOQs or obtain the stream centerline from surveyed information as opposed to using low-resolution stream network files. Georeferencing the surveyed cross-sections to known landmarks (such as bridges, culverts, or distinct terrain features) occurs next. The user selects the button, which calls the Addpnt script, and clicks on the upstream, intermediate, and downstream boundaries to tie the cross-section data imported from HEC-RAS to known landmarks in ArcView GIS by snapping to the closest point on the digital stream. Once this is accomplished, the Map HEC-RAS Cross-Sections menu is selected and the FloodTerrain3d script is called. This script requires the user to define the stream centerline theme, stream definition point theme, as well as the HEC-RAS import table with the surveyed cross-sections corresponding to the boundaries identified highlighted. Because there can be differences between the stream length represented in GIS and those surveyed or derived from stream network files, FloodTerrain3d calculates the ratio of the length of the RAS-modeled stream to that of the digital stream and places the georeferenced cross-sections at the boundaries, while adjusting the locations of the intermediate cross-sections accordingly. 64 The FloodTerrain3d script, as modified for this thesis, then prompts the user to define the orientation of each cross-section (Figure 4-10). Figure 4-10 Cross-Section Orientation Selection Menu The user has the option of manually inputting the angle of the cross-section (as measured from a horizontal line proceeding left to right across the screen that equals 0?) or allowing the FloodTerrain3d script to define the perpendicular orientation of each cross-section by calculating the bearing between two points located immediately upstream and downstream of the cross-section location (the locations of the upstream and downstream points are calculated as a percentage of the total stream length) and drawing a perpendicular line at that location. Figure 4-10 presents a modification to Tate?s 1999 work; this change was driven by the fact that intersecting cross-sections may be acceptable depending on the degree to which the water surface profile has migrated above the bank station elevations. If the floodplain to be modeled is inside of the limits of the intersection cross-sections, the fact that they intersect is not of concern. However, due to the linear nature of HEC-RAS cross- section interpolation algorithms, interpolated cross-sections that intersect within the limits of the floodplain do present unrealistic terrain features (Figure 4-11) and should be edited manually in HEC-RAS prior to generation of a terrain TIN if required. 65 Figure 4-11 Interpolated Cross-Section Discrepancies Although the interpolation of elevations at each point in the cross-section is a time- consuming process, the user is able to calculate the expected change in elevation for the interpolated cross-section because it is a linear interpolation. This amended methodology permits the user to duplicate field survey sketches and actual conditions more realistically in the GIS domain. Each cross-section is then attributed with river station ID, cross-section length, and the location of the stream centerline and bank stations as a function of the percentage of the length of the cross-section (measured from the outer-most cross-section lateral coordinate left of the main channel). The result of the cross-section georeferencing is that every vertex on each cross-section is assigned a series of three-dimensional map coordinates in GIS ? the easting and northing are derived from the mapping process in GIS, and the elevation coordinate from the global variable created in the data import step. Using these 66 three-dimensional points, in conjunction with surrounding terrain data, a TIN model of the stream channel and surrounding floodplain can be created. It is important to synthesize this detailed channel and overbank data with the surrounding terrain because many cross-sections surveyed in the field many not fully define the lateral extent of the overbanks. Figure 4-12 presents an example of a hydraulic model in which the water surface elevation extends beyond the limits of the cross-section. Figure 4-12 Cross-Sections Inadequately Defined for Expected Water Surface Profile In this figure, the black line represents the ground surface, the blue line the modeled water surface profile, and the green dashed line the energy grade line. By combining the detailed channel and overbank coordinates with surrounding terrain data, and re- cutting the cross-sections (using HEC-GeoRAS), the lateral extent of the resulting water surface elevations can be more realistically defined and re-analyzed in HEC- RAS. Tate (1999) has developed a script to resample the resulting georeferenced cross-sections with surrounding digital terrain data such as DEMs or photogrammetrical survey data. The FloodNewXSects script called by the Resample 67 Cross-Sections menu recalculates the elevation of every cross-section point outside of the main channel, and creates a smooth transition from the bank elevations to the surrounding elevations of the DEM or additional survey data. The user then calls the FloodBanklines script by selecting the Stream Centerline and Banklines menu. This script, as modified for this thesis, takes channel cross- sections and creates a three-dimensional theme of the stream centerline that will ultimately be used with the three-dimensional cross-section points to create the terrain TIN. Once the above methodology has been implemented, there is enough information to create a TIN. However, there may be additional data that the user may be able to take advantage of to further define the terrain. On many TxDOT projects, more reliable elevation data than that resulting from field surveys may be available immediately adjacent to the road or bridge being evaluated in the form of aerial photogrammetric survey information (this is true because of the difficulty of duplicating the orientation of the surveyed cross-sections within GIS). Many projects utilize software packages such as GeoPAK ? to calculate cut and fill based on detailed aerial survey information, which also have the capability to create an output XYZ file that can be used to supplement or replace cross-section information in the immediate vicinity of the drainage structure of interest. The user should use good engineering judgment in determining what data is most accurate, and can edit the cross-sections generated in CRWR-FloodMap as necessary to take advantage of this additional data. Many cities also have detailed elevation contour data in highly populated areas that may have a higher resolution than standard 30-meter DEMs and take into consideration buildings and other relevant structures. 68 A TIN can be created from the any of the following types of data using the 3D Analyst extension in ArcView GIS: !"3D cross-section points; !"3D DEM points (converted from grid format using the Convert Grid to Points menu and the FloodR2Vpoint script; !"3D points from photogrammetric surveys or any other point elevation data; !" Hard breaklines representing the stream centerline and banklines; !" Hard breaklines representing building footprints, or !"Soft Breaklines representing elevation contours. The more data used to create the TIN, the more accurate the representation of the terrain and, thus, the more realistic the resulting floodplain once hydraulic modeling is complete. 4.2 GIS-based Hydrologic Parameter Extraction The extraction of spatially variable hydrologic parameters can be accomplished using the bottom four menu items shown on the CRWR-PrePro menu in Figure 4-2. In this research, a curve number grid has already been defined, so only the bottom two menu items need to be activated. Olivera and Maidment (1999) present a detailed discussion of this methodology. 4.2.1 EXTRACTION OF HYDROLOGIC PARAMETERS FROM SUBBASINS CRWR-PrePro calculates the following parameters for each subbasin: !"Area; 69 !"Lag time; and !"Average curve number. The other parameters needed for estimating lag time, such as the length and slope of the longest flowpath, are also calculated and stored in the subbasin attribute table. The calculation of lag time might depend solely on spatial data (i.e., DEM, land use, soils), or it might require additional externally supplied input, depending on the algorithm. The subbasin area is calculated as a result of the vectorization procedure discussed previously. The lag time is calculated with the following formula: ()[] ? ? ? ? ? ? ? ? ? ? = t S CNL t w p 5.3, 67.31 9/1000 max 5.0 7.0 Equation 4.1 where t p (minutes) is the subbasin lag time measured from the centroid of the hyetograph to the peak time of the hydrograph, L w (feet) is the length of the longest flowpath, S (%) is the slope of the longest flowpath, CN is the average curve number in the subbasin, and t (min) is the analysis time-step. The first term in the parentheses corresponds to the lag time according to the SCS (1972), and the second term is the minimum lag time value required by HEC-HMS (HEC, 1990). The longest flowpath, as calculated in CRWR-PrePro, is the distance from the centroid of the furthest cell in the watershed to the outlet of the subbasin. This distance may not follow the main channel in all cases (Figure 4-13). 70 Figure 4-13 Longest Flowpath Calculation The light blue lines in the figure represent the main channel, while the dark blue lines depict the longest flowpath as determined by CRWR-PrePro algorithms. CN is calculated as the average of the curve number values within the subbasin polygon and is derived from the curve number grid developed from land use and land cover data, along with STATSGO soil data. The curve number grid used in this thesis was provided by the Blacklands Research Institute. 4.2.2 EXTRACTION OF HYDROLOGIC PARAMETERS FROM REACHES CRWR-PrePro calculates the following parameters for each reach: !"Reach length; !"Reach routing method (Muskingum or Lag); and !"Either the number of sub-reaches into which the reach is subdivided (when Muskingum routing is used), or !"The flow time (when pure lag routing is used). 71 Other reach parameters such as flow velocity and Muskingum X cannot be computed from spatial data and must be supplied by the user. The Muskingum flow routing method (the method used in this research) models the volume of water stored in a stream as the sum of a prism and a wedge, as presented in Figure 4-14. Figure 4-14 Prism and Wedge Storage in Muskingum Routing The prism represents storage across a constant cross-section along the length of the channel, while the wedge represents the surface ?wave? of water that enters the section with the inflow. Assuming a constant velocity, there is a constant ratio between the flow rate and the cross-sectional area. This means that flow is also directly dependent on the volume of prism storage, a function of reach length and cross-sectional area, by a factor of K (prism storage = K*Q). K, therefore, represents the time of travel of the flood wave through the modeled reach. The volume of the wedge of water is dependent on the difference between the inflow and outflow, such that storage can be calculated according to the following equation: ()[]QXXIKS ?+= 1 Equation 4.2 where X is a weighting factor ranging from 0 to 0.5 depending on the shape of the wedge. This method is used for routing in reaches long enough not to present 72 numerical instability problems. In short reaches in which the flow time is shorter than the time-step, the pure lag method of routing is used. In very long reaches, each reach is subdivided into shorter reaches to again avoid numerical instability such that the flow time satisfies the condition: ktXk