i CRWR Online Report 07-02 An Integrated Stream Classification System for Texas by Eric S. Hersh, M.S.E. Graduate Research Assistant and David R. Maidment, Ph.D. Principal Investigator October 2007 CENTER FOR RESEARCH IN WATER RESOURCES Bureau of Engineering Research • The University of Texas at Austin J.J. Pickle Research Campus • Austin, TX 78712-4497 This document is available online via World Wide Web at http://www.crwr.utexas.edu/online.shtml ii Acknowledgements The authors would like to thank the Texas Commission on Environmental Quality (TCEQ) for funding and supporting this research under Project Number UT-16-7-82924. In particular, the contributions of Dr. Wendy Gordon at TCEQ and of stakeholders and workgroup members too numerous to be named individually is appreciated. iii Abstract The recently-passed Senate Bill 3 tasks stakeholders and regulators with determining and reviewing environmental flow needs. A stream classification system was developed and tested for Texas to support analyses of environmental flows based on quantitative data for 18 distinguishing parameters encompassing watershed and stream channel processes from four disciplines: (1) Hydrology & Hydraulics, (2) Water Quality, (3) Geomorphology & Physical Processes, and (4) Climatology. The State of Texas was partitioned into five regions: East Texas, South-Central Texas, Lower Rio Grande Basin, West Texas, and North-Central Texas by 8-digit Hydrologic Unit Code (HUC) basins. This stream classification system might be used to: (1) discern likely similarities and differences between rivers and streams of the State, (2) remotely characterize stream segments for which resources are insufficient for detailed field studies, (3) recognize streams and watersheds of the State as having common identities, (4) allow conclusions drawn from an instream flow study from a particular river reach to have a wider applicability than the particular study site, and (5) assist in prioritization of rivers and reaches for future instream flow studies. iv Table of Contents List of Tables ........................................................................................................ vii List of Figures...................................................................................................... viii List of Acronyms .....................................................................................................x 1. STREAM CLASSIFICATION SYSTEM INTRODUCTION 12 1.1 Background......................................................................................................12 1.1.1 Project Purpose ....................................................................................12 1.1.2 Project Motivation ...............................................................................12 1.2 Legislative Framework ....................................................................................13 1.2.1 Senate bill 2 - Science..........................................................................13 1.2.2 Senate Bill 3 - Implementation ............................................................14 2. EXISTING STREAM CLASSIFICATION SCHEMES 16 2.1 Freshwater Ecosystem Classification ..............................................................16 2.2 River Environment Classification....................................................................19 2.3 River Styles Framework ..................................................................................20 2.4 Additional Classification Systems ...................................................................25 3. STREAM CLASSIFICATION SYSTEM FRAMEWORK 29 3.1 Conceptual Framework....................................................................................29 3.2 Physical Settings for Instream Flows in Texas................................................30 3.2.1 Generalized Districts............................................................................30 3.2.2 NRC Regions .......................................................................................31 3.3 Thematic Layers...............................................................................................34 3.3.1 Distinguishing Parameters ...................................................................34 v 4. STREAM CLASSIFICATION SYSTEM INTEGRATED DATA 35 4.1 Foundation: Hydrography................................................................................35 4.1.1 NHDPlus..............................................................................................35 4.1.2 NHDPlus Spatial Representation.........................................................36 4.1.3 Texas Hydrography..............................................................................40 4.2 Water Quality...................................................................................................42 4.2.1 TCEQ TRACS .....................................................................................42 4.2.2 Water Quality Variables ......................................................................43 4.3 Climatology......................................................................................................53 4.3.1 Data Sources ........................................................................................53 4.3.2 Climatology Variables .........................................................................54 4.4 Hydrology and Hydraulics...............................................................................59 4.4.1 Data Sources ........................................................................................59 4.4.2 Hydrologic and Hydraulic Variables ...................................................62 4.5 Geomorphology and Physical Processes .........................................................73 4.5.1 Data Sources ........................................................................................73 4.5.2 Geomorphology and Physical Processes Variables .............................75 4.6 Biology.............................................................................................................83 4.6.1 Data Availability Challenges ...............................................................83 4.6.2 Bio-aquatic Informatics for Texas Workgroup....................................84 4.6.3 Data Sources ........................................................................................84 5. A STREAM CLASSIFICATION SYSTEM FOR TEXAS 87 5.1 Data Integration ...............................................................................................87 5.1.1 Generalized Districts............................................................................87 5.1.2 Distinguishing Parameters ...................................................................89 5.1.3 Redundant Subbasins...........................................................................90 vi 5.2 Analysis of Original Generalized Districts......................................................92 5.3 Revision Methodology and Results .................................................................97 5.4 Revised Classes..............................................................................................102 5.5 Analysis of Revised Integrated Stream Classes.............................................104 5.5.1 Analysis of revised classes................................................................104 5.5.2 Ecoregion comparison ......................................................................113 5.6 Limitations .....................................................................................................114 6. CONCLUSIONS AND RECOMMENDATIONS 117 APPENDIX A – SCOPE OF WORK 119 APPENDIX B – SUPPORTING DATA 121 REFERENCES 127 vii List of Tables Table 1. TNC’s freshwater ecosystem classification levels and separation factors (Smith 2006). ............................................................................................................................................... 18 Table 2. River Styles Framework hierarchy (Brierley and Fryirs 2005, Phillips 2006).............. 22 Table 3. Summary of NRC regionalization variables.................................................................. 33 Table 4. Summary of quantitative variables. ............................................................................... 34 Table 5. USGS and NHDPlus Hydrologic Units......................................................................... 40 Table 6. Water quality parameters used in the stream classification system............................... 45 Table 7. Average monthly PET at selected Texas cities (from ITC 2005).................................. 57 Table 8. TCEQ TRACS and National Geochemical Survey dominant substrate type code key (STORET 89844).................................................................................................................. 81 Table 9. Groupings of STORET parameters developed for analysis of the TRACS SWQM database................................................................................................................................. 85 Table 10. Summary of biologic data in TRACS SWQM. ........................................................... 86 Table 11. Devils River subbasins switched from the Lower Rio Grande Basin to the South Central Texas Basin based on stakeholder consensus. ......................................................... 87 Table 12. Distinguishing parameters of the riverine environment incorporated into the stream classification system and their units. .................................................................................... 90 Table 13. Attributes of the six subbasins which lie at the state boundary and have been divided into multiple polygons. ......................................................................................................... 92 Table 14. Count and area of subbasins grouped by the original generalized districts in Texas. . 93 Table 15. Comparison of qualitative distinctions and quantitative data for East Texas.............. 93 Table 16. Comparison of qualitative distinctions and quantitative data for North Central Texas. ............................................................................................................................................... 94 Table 17. Comparison of qualitative distinctions and quantitative data for South Central Texas. ............................................................................................................................................... 94 Table 18. Comparison of qualitative distinctions and quantitative data for Lower Rio Grande Basin. .................................................................................................................................... 95 Table 19. Comparison of qualitative distinctions and quantitative data for West Texas. ........... 95 Table 20. Mean, standard deviation, and coefficient of variation statistics for the original generalized districts of Texas; blank cells indicate insufficient data.................................... 96 Table 21. Subbasins moved between regions during the revision process................................ 100 Table 22. Count and area of subbasins grouped by the revised stream classes......................... 105 Table 23. Mean, standard deviation, and coefficient of variation statistics for the revised stream classes of Texas................................................................................................................... 106 Table 24. Comparison (percent change) for the mean, standard deviation, and coefficient of variation following revisions. ............................................................................................. 112 viii List of Figures Figure 1. The Nature Conservancy’s freshwater ecosystem classification hierarchy (from Higgins et al. 2005)............................................................................................................... 17 Figure 2. River Environment Classification hierarchy (from Snelder and Biggs 2002). ............ 20 Figure 3. Stages and steps in the River Styles Framework (Brierley and Fryirs 2005)............... 23 Figure 4. Summary controls on the character and behavior of River Styles in Bega Catchment, New South Wales, Australia (Brierley and Fryirs 2000)...................................................... 24 Figure 5. Location and stream classification of the 420 gages of Olden and Poff (2003)........... 26 Figure 6. Classification rules employed by the USGS New Jersey Stream Classification Tool (Henriksen et al 2006)........................................................................................................... 27 Figure 7. US Forest Service spatio-temporal scaled patterns of (a) riverine systems; (b) physical features; (c) disturbance processes; and (d) biotic processes (from Maxwell et al 1995). ... 28 Figure 8. Data model thematic layers that organize the data by discipline. ................................ 29 Figure 9. Map-based interpretation of the NRC text-based qualitative regionalization.............. 31 Figure 10. Example representation of elevation data (brown to green color ramp) and streamflow data (blue lines of varying thickness). ............................................................... 36 Figure 11. National NHDPlus Production Units.......................................................................... 37 Figure 12. NHDPlus Production Units contributing flow to Texas waterways........................... 37 Figure 13. Major river basins of Texas........................................................................................ 38 Figure 14. USGS Hydrologic Regions......................................................................................... 39 Figure 15. Subbasins of Texas..................................................................................................... 41 Figure 16. NHDPlus subbasin attributes...................................................................................... 41 Figure 17. TRACS SWQM stations, 1968-2006. ........................................................................ 43 Figure 18. SWQM top fifteen parameters by result (Jantzen 2007)............................................ 44 Figure 19. Subbasins with any water quality data in TRACS, 1968-2006.................................. 46 Figure 20. Mean water temperature (in degrees C) by subbasin. ................................................ 47 Figure 21. Mean dissolved oxygen by subbasin. ......................................................................... 48 Figure 22. Mean pH by subbasin. ................................................................................................ 49 Figure 23. Mean specific conductance by subbasin..................................................................... 50 Figure 24. Mean total nonfiltrable residue (i.e., total suspended solids) by subbasin................. 51 Figure 25. Tests for correlation between water quality parameters, grouped by: DO (top four), temperature (three), TSS (two), and specific conductance (one). Note: scales and correlated variable are not important in this case; plots are simply meant to depict scatter in the data.53 Figure 26. Mean annual precipitation by subbasin. ..................................................................... 55 Figure 27. Mean annual temperature by subbasin. ...................................................................... 56 Figure 28. Interpolated mean annual PET. .................................................................................. 58 Figure 29. Mean annual PET by HUC......................................................................................... 59 Figure 30. Mean annual streamflow, normalized by contributing drainage area. ....................... 63 Figure 31. Mean annual streamflow, normalized by contributing drainage area and grouped by HUC. ..................................................................................................................................... 64 Figure 32. Mean annual stream velocity, by HUC. ..................................................................... 65 Figure 33. Mean base flow index (BFI) by streamflow gage. ..................................................... 66 Figure 34. Mean BFI by subbasin................................................................................................ 67 Figure 35. Percent of zero flow days. .......................................................................................... 68 Figure 36. Percent of zero flow days, by HUC............................................................................ 69 ix Figure 37. Intermittent streams in Texas (in red), distinguished as having a median streamflow equal to 0 cfs. ........................................................................................................................ 70 Figure 38. Interquartile range of daily streamflow normalized by median streamflow. ............. 71 Figure 39. Interquartile range of daily streamflow normalized by median streamflow and grouped by HUC................................................................................................................... 72 Figure 40. Example NHDPlus map for the San Marcos basin, Texas, depicting channel slope by reach...................................................................................................................................... 74 Figure 41. Longitudinal profile of the Colorado River, Texas. Note the step-shaped reaches between kilometers 800 and 550, which are the Highland Lake system reservoirs and dams, with the largest vertical reach (approximately kilometer 590) being Mansfield Dam at Lake Travis. ................................................................................................................................... 75 Figure 42. Soil clay composition, in percent. .............................................................................. 76 Figure 43. Soil silt composition, in percent. ................................................................................ 76 Figure 44. Soil sand composition, in percent............................................................................... 76 Figure 45. Percent clay by subbasin. ........................................................................................... 77 Figure 46. Percent silt by subbasin. ............................................................................................. 77 Figure 47. Percent sand by subbasin............................................................................................ 77 Figure 48. Mean reach bed slope by subbasin. ............................................................................ 78 Figure 49. Tests for correlation between each water quality parameter and channel bed slope. Note: scales and correlated variable are not important in this case; plots are simply meant to depict scatter in the data........................................................................................................ 79 Figure 50. USGS National Geochemistry Survey database sample sites in Texas. .................... 80 Figure 51. Dominant stream substrate type. ................................................................................ 82 Figure 52. Dominant stream substrate type, by HUC.................................................................. 83 Figure 53. Devils River subbasins switched from the Lower Rio Grande Basin to the South Central Texas Basin based on stakeholder consensus. ......................................................... 88 Figure 54. Stakeholder consensus-based revision of generalized NRC districts......................... 89 Figure 55. The six subbasins at the state boundary and have been divided into multiple polygons................................................................................................................................ 91 Figure 56. Border subbasins that were tested during the revision process. ................................. 98 Figure 57. Subbasins moved between regions during the revision process: original NRC districts................................................................................................................................ 101 Figure 58. Subbasins moved between regions during the revision process: revised stream classes. ................................................................................................................................ 102 Figure 59. Results of the revision process: the integrated stream classification system for Texas. ............................................................................................................................................. 103 Figure 60. The integrated stream classification system for Texas, overlain with the major river basins of the state................................................................................................................ 104 Figure 61. Level III ecoregions of Texas overlain with the integrated stream classes.............. 114 x List of Acronyms BAIT Bio-aquatic Informatics for Texas BEG Bureau of Economic Geology BFI Base flow Index CFS Cubic feet per second CONUS Conterminous United States CRWR Center for Research in Water Resources DO Dissolved Oxygen EDU Ecological Drainage Unit ESRI Environmental Systems Research Institute GAT Geologic Atlas of Texas HAT Hydrologic Assessment Tool HCDN Hydro-Climatic Data Network HUC Hydrologic Unit Code IBWC International Boundary and Water Commission IQR Interquartile Range ITC Irrigation Technology Center MAF Mean annual flow MAV Mean annual velocity NAWQA National Water Quality Assessment NED National Elevation Dataset NHD National Hydrography Dataset NHDPlus National Hydrography Dataset Plus NLCD National Land Cover Dataset NRC National Research Council NRCS National Resource Conservation Service PET Potential Evapotranspiration PRISM Parameter-elevation Regressions on Independent Slopes Model RSF River Styles Framework SB2 Senate Bill 2 SB3 Senate Bill 3 SQL Structured Query Language xi STATSGO State Soil Geographic STORET Storage and Retrieval SWQM Surface Water Quality Monitoring SWQMIS Surface Water Quality Monitoring Information System TCEQ Texas Commission on Environmental Quality TIFP Texas Instream Flow Program TNC The Nature Conservancy TNRIS Texas Natural Resources Information System TPWD Texas Parks and Wildlife Department TRACS TCEQ Regulatory Activities Compliance Systems TSS Total Suspended Solids TWDB Texas Water Development Board UROM Unit Runoff Method USEPA United States Environmental Protection Agency USGS United States Geological Survey VBA Visual Basic for Applications WBD Watershed Boundary Dataset WWF World Wildlife Fund 12 1. STREAM CLASSIFICATION SYSTEM INTRODUCTION 1.1 Background 1.1.1 PROJECT PURPOSE The Texas Commission on Environmental Quality (TCEQ) has the statutory obligation to review water rights applications for their potential impacts on aquatic resources and set environmental flow requirements. The agency carries out this charge by relying on a hydrologic desktop method, publicly-available data, and site-specific information collected in the field to make environmental flow determinations. The TCEQ-designated water quality management segments with studies currently underway represent a small fraction of the total number of free-flowing rivers (i.e., not inland water bodies, tidal reaches or coastal segments) in Texas. Studies on the remaining segments will take more time and resources than are available, yet it is important to characterize the other segments to determine appropriate instream flow requirements. Thus, the purpose of this project was to use Geographic Information System (GIS) technology to organize existing information relevant to the understanding of Texas streams and rivers (i.e., water quality, geologic and geomorphic, hydrologic and hydraulic, and biologic data) and to develop a classification scheme such that particular classes or regions of streams and rivers could be recognized as having a common identity. 1.1.2 PROJECT MOTIVATION The value and need for stream classification was recognized in the May 2006 Texas Instream Flow Program (TIFP) Draft Texas Instream Flow Studies: Technical Overview: The TIFP has identified six priority river basins in which to initiate studies and implement recommendations. These priority basins represent a small subset of the total number of rivers and streams in the state. Ultimately, the program will need to be expanded to encompass these other rivers and streams. Expansion should be based on a priority-setting system and may involve additional studies. In addition, 13 it is anticipated that classification tools will be developed to aid in the application of instream flow standards to the state’s myriad rivers and streams. It would be a near-impossible task to individually study all the state’s 191,000 river miles. Derivation of hydrologically, ecologically, and geomorphologically similar aquatic ecosystem units would enable the establishment and application of streamlined methods for developing instream flow recommendations. Additionally, a recent paper by Arthington, Bunn, Poff, and Naiman (2006) put forth the concept of stream classification as a means to “bridge the gap between simple hydrological ‘rules of thumb’ and more comprehensive environmental flow assessments and experimental flow restoration projects”: Rather than attempting to manage for the ‘uniqueness’ of every individual river’s natural flow regime, we identify ‘classes’ of streams based on key attributes of flow variability, and then calibrate relationships between alterations in each flow attribute and measures of ecological condition for each stream class. The goal of this project was to develop a classification scheme such that particular classes or regions of streams and rivers could be recognized as having a common identity or sharing common attributes. Accordingly, conclusions drawn from instream flow studies in particular river reaches might be generalizable. 1.2 Legislative Framework 1.2.1 SENATE BILL 2 - SCIENCE The 77th session of the Texas Legislature passed Senate Bill 2 (SB2) in 2001, directing the Texas Commission on Environmental Quality (TCEQ), Texas Parks and Wildlife Department (TPWD), and Texas Water Development Board (TWDB) (hereinafter referred to as “the agencies”) to “…jointly establish and continuously maintain an instream flow data collection and evaluation program…” and to “…conduct studies and analyses to determine appropriate methodologies for determining flow conditions in the state rivers and streams necessary to support a sound ecological environment,” which was further defined by the agencies as “ a functioning ecosystem 14 characterized by intact, natural processes, resilience, and a balanced, integrated, and adaptive community of organisms comparable to that of the natural habitat of a region.” The agencies’ vehicle for implementing SB2 is the Texas Instream Flow Program (TIFP) (Senate Bill 2, TIFP 2006). Six subbasins were identified by the agencies for TIFP priority study based on potential water development projects, water rights permitting issues, and other factors. They are: the Lower Sabine, Middle Trinity, Middle and Lower Brazos, Lower Guadalupe, and Lower San Antonio River Basins. Specific instream flow studies were scheduled to be completed for each priority basin by December 31, 2010, but Senate Bill 3 (SB3), passed in 2007 by the 80 th Legislature, extended this deadline to December 31, 2016 (Texas Legislature 2001, TIFP 2002, Texas Legislature 2007). 1.2.2 SENATE BILL 3 - IMPLEMENTATION Passed in 2001, Senate Bill 2 established the TIFP to collect data and determine the flow regime protective of the ecological environment. Often referred to colloquially as “the science bill,” it did not address implementation nor did it include consideration of water users and uses external to the riverine ecosystem (Texas Legislature 2001, TIFP 2002). Senate Bill 3 (SB3), the implementation bill, was proposed in the 79th Legislative Session (2005) but was not passed until 2007 by the 80th Legislature. SB3 establishes the who, when, and how of environmental flow implementation in the State of Texas by creating: y an environmental flows advisory group, y an environmental flows advisory committee, y bay and basin stakeholder committees, and y bay and basin expert science teams. This hierarchy of rule makers, scientists and stakeholders is variously tasked with: y identifying environment flow needs for the bays and basins of the State, y reviewing the determinations, y building consensus through balanced representation by region, y considering the environmental flow needs in light of present and future water needs and other uses, 15 y establishing public and private market approaches for satisfying flow needs, and y developing a formal process of review and adaptive management (Texas Legislature 2007). SB3 mandate that the TCEQ shall adopt the environmental flow standards recommended by the basin and bay area stakeholders committee by September 1, 2010. SB3 does not explicitly address how the results of the SB2 priority studies to be completed by December 31, 2016 will be incorporated into the environmental flow standards to be promulgated by TCEQ. However, SB3 includes provisions for an adaptive management-based program of periodic reevaluation, validation, and refinement based on the best available science; this likely includes the SB2 study findings. Furthermore, SB3 alters the conditions under which water rights permits are issued by TCEQ such that “in its consideration of an application for a permit to store, take, or divert water…” the TCEQ must consider the environmental flow requirements of the bays and basins, and any new permits or new permit modifications must include provisions for adjustment to allow for adaptive management. With regards to the variability of hydrologic systems, SB3 mandates that “Environmental flow standards…must consist of a schedule of flow quantities, reflecting seasonal and yearly fluctuations that may vary geographically by specific location in a river basin and bay system” (Texas Legislature 2007). 16 2. EXISTING STREAM CLASSIFICATION SCHEMES 2.1 Freshwater Ecosystem Classification Numerous classification schemes have been developed for various regions of the world based on variables from a single discipline. By considering a limited subset of conditions representing the riverine environment, these classification schemes are inherently limited in application and value for holistic considerations such as the development of instream flow prescriptions. The preponderance of classifications have been based on hydrology and geomorphology. The World Wildlife Fund (WWF) developed freshwater ecoregions “which are derived by aggregating and subdividing watersheds based on the distribution patterns of aquatic species. With watersheds as their foundation, the freshwater ecoregions can be effective units for conservation planning” (Abell et al. 2000). The Nature Conservancy (TNC) subsequently developed a freshwater classification scheme for nationwide application and applied the scheme to much of Texas in support of the Conservancy’s internal ecoregional planning process (Fitzhugh 2005, Higgins et al. 2005) (Figure 1). The system is a four-tiered, hierarchical approach based on predominantly abiotic parameters, with the first tier being the WWF freshwater ecoregions. The original TNC scheme was crafted prior to the availability of multiple valuable electronic datasets (notably NHDPlus) and was managed as shapefiles in ESRI ArcView 3.2 software. TNC is currently in the process of ‘maturing’ the classification scheme in Texas and reclassifying certain regions of the State (Smith 2006). 17 Figure 1. The Nature Conservancy’s freshwater ecosystem classification hierarchy (from Higgins et al. 2005). 18 Table 1. TNC’s freshwater ecosystem classification levels and separation factors (Smith 2006). Level Examples Description Separation Factors Ecoregion 1) Edwards Plateau 2) Pineywoods Similar climate and physiography that corresponds to broad vegetation regions Forest type, shrubland vs. grassland Ecological Drainage Unit (EDU) 1) Guadalupe/San Antonio 2) Nueces 3) Lower Rio Grande/Devils Aggregates of watersheds that share ecological, biological, and aquatic zoogeographical characteristics, by 8- digit HUC. Within each EDU there is a regional subset of aquatic ecosystem types Physiography, zoogeography, watershed Aquatic Ecological System 1) Medium sized perennial prairie streams 2) Small Edwards Plateau rivers 3) Piney Woods bayous Hydrological subunits of EDU’s. Defined by landscape position of a stream size-class within 1 or 2 stream orders that represent a dynamic assemblage of aquatic communities Size, drainage network position, connectivity, hydrologic regime, geology Macrohabitat 1) Meandering, low gradient, riffle/pool plains stream 2) Medium gradient, foothills beaver-pond influenced stream Different valley segment types of stream reaches (think stream reach of 30km), within segments, relatively homogeneous. Finest scale classification unit on the maps. Surficial geology, drainage network position, connectivity, hydrologic regime, geology 19 2.2 River Environment Classification The New Zealand River Environment Classification (REC) is a classification scheme that has been considered and/or applied in other countries since its inception in 2002, including: Australia, Belgium, Chile, France, and the United States (Snelder and Biggs 2002, Snelder et al. 2004, Biggs 2007, Kilroy et al. 2007, Norris et al. 2007). REC is a physically-based system of nested hierarchical variables that each operate on different spatial scales. REC incorporates climate, topography, geology, and land cover, as defined by: • Climate: temperature, precipitation, potential evapotranspiration (PET) • Flow source: mountain, hill, low elevation, or lake • Geology: dominant rock type • Land cover: vegetation types at a 1-10 sq. km scale • Network position: Strahler stream order (or distance from river mouth, or average section elevation) • Valley landform: primarily slope, but also lateral (floodplain) and vertical (hyporheic) connectivity, hydraulic geometry, bankfull discharge, local stream power, sediment size range, and riparian conditions. In the order presented above, the variables are taken into consideration at diminishing spatial scales ranging from the order of 10 5 sq. km for climate down to 10 1 to 10 0 sq. km for land cover, network position, and valley landform. The REC was tested in 2005 via analyses of 13 streamflow variables from 335 gages across New Zealand to prove that inter-class differences were greater than intra-class differences and to quantify the “strength” of the classification compared to previously-developed geographic and ecoregion systems (Snelder and Biggs 2002, Snelder et al. 2004, Snelder and Hughey 2005, Snelder et al. 2004, Snelder et al. 2005). 20 Figure 2. River Environment Classification hierarchy (from Snelder and Biggs 2002). 2.3 River Styles Framework There are a host of geomorphic classifications, each with a differing approach and differing purpose, including those proposed by Rosgen (1994) and Kondolf et al. (2003). One such scheme generating much recent interest in the river science community is the River Styles Framework from the New South Wales Department of Land and Water Conservation in Australia (Brierley and Fryirs 2000, Thomson et al. 2001, Brierley et al. 2002, Fryirs 2003, Thomson et al. 2004, Brierley and Fryirs 2005, Chessman et al. 2006). Based on work conducted in 2005 by Dr. Jonathan Phillips of Copperhead Road 21 Geoscience and of the University of Kentucky, the River Styles Framework has been selected as the geomorphic classification system of choice for the TIFP. River Styles Framework “is not a classification system, per se, but a flexible, dynamic approach to river characterization” (Phillips 2006). River Styles Framework is designed to both assess current (static) and historical conditions and to forecast likely trajectories of change, thus moving beyond the traditional thinking on the subject of equilibrium and into an assessment of sensitivity and resiliency characterized by complex nonlinear dynamics. River Styles Framework differs from traditional categorical classification schemes such as the Rosgen Stream Classification System because it is “specifically intended to incorporate evolutionary pathways of the fluvial system, rather than static conditions that are presumed to be related to stable equilibrium states.” (Phillips 2006) The NRC (2005) review recognizes the importance of geomorphic classification for the TIFP and also the merit of evaluating both the current equilibrium status of a river system and also indicators of recent and historic change. Such an approach would tend to favor the strengths of a dynamic characterization system like the River Styles Framework over traditional, static categorization systems. Under the River Styles Framework, the geomorphology of a river system is examined first and a classification system is then developed based on the geomorphic findings. This a posteriori classification is set within a nested hierarchical framework where various physically-based components are used to distinguish between classes at each hierarchical level (Table 2). 22 Table 2. River Styles Framework hierarchy (Brierley and Fryirs 2005, Phillips 2006). Hierarchical Level Determining Characteristics Watershed Drainage divides, hydrologic units Landscape unit Geology, elevation, relief, slope, morphology River style Length of channel (and valley) with a characteristic assemblage of geomorphic units Geomorphic unit Instream and floodplain landforms reflecting distinct form-process associations Hydraulic unit Uniform patch of flow and substrate Microhabitat Individual elements, such as logs, boulders, and scour holes. The recommended methodology to conduct a River Styles Framework assessment is organized into stages and steps (Figure 3), and an example of a completed assessment for the Bega Catchment in New South Wales, Australia is presented in Figure 4. 23 Figure 3. Stages and steps in the River Styles Framework (Brierley and Fryirs 2005). 24 Figure 4. Summary controls on the character and behavior of River Styles in Bega Catchment, New South Wales, Australia (Brierley and Fryirs 2000) 25 The successful implementation of River Styles Framework requires extensive field work and a considerable understanding of geomorphic principles. River Styles Framework utilizes common descriptors but has no a priori styles, so the value of this system for understanding and classifying (grouping) the rivers of Texas for the TIFP or other purposes is unclear. 2.4 Additional Classification Systems Poff (1996) put forth a hydrogeographic regionalization of unregulated streams in the contiguous United States based on a study of flow regime characteristics and flow sources at 420 gaged sites; the ten classes were pared down into six in Olden and Poff (2003). The results of the national study likely bear little fruit for application in Texas, however, due to the lack of study locations within the State or other hydrologically- similar regions (Figure 5). 26 Figure 5. Location and stream classification of the 420 gages of Olden and Poff (2003). Olden and Poff’s (2003) regionalization forms the basis for the USGS Hydroecological Integrity Assessment Process. Under the process, the State of New Jersey has developed a state-specific classification tool to partition that state’s gaged streams into four stream classes, termed A, B, C, and D, by their relative degree of skewness of daily flows (high versus low) and by the relative frequency of low flow events per year (high versus low) (Figure 6). Group B streams have stable, groundwater- supported streamflow with a high base flow index; Group D streams are small and flashy with little base flow; Groups A and C are intermediate streams. The two measures employed provide an indication of the relative degree of flashiness of gaged streams across the State as well as the relative degree of base flow influence, as indicated by the MA5 and FL3 statistics within the USGS’s Hydrologic Assessment Tool (HAT): 27 • MA5 – The skewness of the entire flow record is computed as the mean for the entire flow record (MA1) divided by the median (MA2) for the entire flow record (dimensionless - spatial). • FL3 – Frequency of low pulse spells. Compute the average number of flow events with flows below a threshold equal to 5 percent of the mean flow value for the entire flow record. FL3 is the average (or median - Use Preference option) number of events (number of events/year – temporal). Figure 6. Classification rules employed by the USGS New Jersey Stream Classification Tool (Henriksen et al 2006). Another hierarchical system, based on the spatial and temporal scale of various processes and forcings is the U.S. Forest Service Aquatic Ecological Units in North America (Nearctic Zone) Classification (Figure 7). This hierarchy was developed in recognition of the varying and overlapping scales of influence, both in time and in space, of various forcing functions present in a riverine ecosystem (Maxwell et al 1995). 28 Figure 7. US Forest Service spatio-temporal scaled patterns of (a) riverine systems; (b) physical features; (c) disturbance processes; and (d) biotic processes (from Maxwell et al 1995). 29 3. STREAM CLASSIFICATION SYSTEM FRAMEWORK 3.1 Conceptual Framework To develop and apply an integrated stream classification system, the appropriate data must first be obtained and then assembled in a logical, systematic framework, otherwise known as a data model. This data aggregation and mediation is essential and must be accomplished before any classification schemes are considered or applied. Conceptually, data integrated into the data model are organized into themes by discipline, much like in the ArcHydro I data model for surface water (Figure 8) (Zeiler 1999, Maidment 2002). Figure 8. Data model thematic layers that organize the data by discipline. 30 3.2 Physical Settings for Instream Flows in Texas 3.2.1 GENERALIZED DISTRICTS A qualitative regionalization of Texas streams and rivers is presented in the National Research Council Committee (2005) review of the Texas Instream Flow Program that familiarizes readers with the “physical settings for instream flows in Texas.” In the current project, this qualitative regionalization and its boundaries were examined using quantified criteria. The State of Texas was partitioned into five regions: East Texas, South-Central Texas, Lower Rio Grande Basin, West Texas, and North-Central Texas via a series of qualitative parameters. Here, a mapping of the NRC (2005) text was interpreted by 8- digit Hydrologic Unit Code (HUC) basin (Figure 9). 31 East Texas North-Central Texas West Texas South-Central Texas Lower Rio Grande Basin Figure 9. Map-based interpretation of the NRC text-based qualitative regionalization. 3.2.2 NRC REGIONS The NRC (2005) regionalization is based on a series of qualitative distinguishing parameters, described below. East Texas consists of the Lower Red, Lower Trinity, Lower Brazos, Navasota, Sabine and Neches river basins and is characterized by: y 30 to 50 inches average precipitation y Flat landscapes y Clay-rich or sandy soils (sandy: Sabine, Neches) y Flood pulses y Dominant land use is agriculture 32 y High flow variations y High turbidity (esp. Trinity, Brazos, Red) y Soft, shifting substrate, large woody debris South-Central Texas consists of the Blanco, Comal, Frio, Guadalupe, Lower Colorado, Nueces, Sabinal, San Antonio, and San Marcos river basins, including the Hill Country, and is characterized by: y 10 to 40 inches average precipitation y Infrequent flash floods y Rocky, Edwards Plateau y Clear and cool water y Dominant land use is livestock grazing y High base flow index The Lower Rio Grande Basin consists of the Lower Rio Grande, Devils, Las Moras Creek, and San Felipe Creek river basins and is characterized by: y 11 to 26 inches average precipitation y Rio Grande occasionally reduced to series of isolated pools y Rio Grande occasionally fails to reach Gulf of Mexico y Dominant land use is irrigated row cropping (Lower Rio Grande) and livestock grazing (elsewhere) y Stressed aquatic biota West Texas consists of the Middle Rio Grande and Pecos river basins and is characterized by: y 8 to 16 inches average precipitation y High salinity in Pecos y Dominant land use is livestock grazing 33 North-Central Texas consists of the Canadian, Upper Brazos, Upper Colorado, Upper Red, and Upper Trinity river basins and is characterized by: y 15 to 28 inches average precipitation y Occasional severe droughts y Clay-rich soils y Flood pulses y Dominant land use is agriculture y High flow variations (drought/flood) (NRC 2005) A summary of the qualitative variables used in the NRC regionalization, grouped by discipline, can be found in Table 3. Table 3. Summary of NRC regionalization variables. 34 3.3 Thematic Layers 3.3.1 DISTINGUISHING PARAMETERS Based on a review of stream classification literature, discussions with stakeholders and peers, and a review of available data, a series of quantitative parameters were selected for evaluation in the stream classification system (Table 4). These parameters were chosen as broad indicators of the river environment encompassing multiple disciplines. Their selection was based on data availability and perceived relevance to the ecological environment. Table 4. Summary of quantitative variables. 1. Biologic data were not incorporated into the current version of the stream classification system; refer to the Biology section below for discussion. 35 4. STREAM CLASSIFICATION SYSTEM INTEGRATED DATA 4.1 Foundation: Hydrography 4.1.1 NHDPLUS The foundation of the proposed classification system is the NHDPlus hydrography dataset. The hydrography dataset is the mapped surface water system, often thought of as the ‘blue lines’ on a map (Maidment 2002). NHDPlus is an improved version of the USGS’s National Hydrography Dataset (NHD) Medium Resolution (1:100,000 scale) and had been jointly developed by USGS, the U.S. Environmental Protection Agency (EPA), and Horizon Systems, Inc. as a contractor to EPA. NHDPlus is “an integrated suite of application-ready geospatial data sets that incorporate many of the best features of the National Hydrography Dataset (NHD), the National Elevation Dataset (NED), the National Land Cover Dataset (NLDC), and the Watershed Boundary Dataset (WBD). The NHDPlus consists of nine components: y Greatly improved 1:100K National Hydrography Dataset (NHD) y A set of value added attributes to enhance stream network navigation, analysis and display y An elevation-based catchment for each flowline in the stream network y Catchment characteristics y Headwater Node Areas y Cumulative drainage area characteristics y Flow direction, flow accumulation and elevation grids y Flowline min/max elevations and slopes y Flow volume & velocity estimates for each flowline in the stream network” (Horizon Systems 2007) The integration of watershed and land surface data in NHDPlus represents a leap forward in the potential for the analysis of freshwater systems. Also, NHDPlus allows 36 the user to map streams by flow size, thus enabling an at-a-glance understanding of the hydrologic flow pattern of the landscape (Figure 10). Figure 10. Example representation of elevation data (brown to green color ramp) and streamflow data (blue lines of varying thickness). 4.1.2 NHDPLUS SPATIAL REPRESENTATION NHDPlus Regions are subdivided into Production Units to allow for easier extraction, file storage, and manipulation of the data (Figure 11 and Figure 12). The Production Units differ slightly from the major river basins of Texas (Figure 13). 37 Figure 11. National NHDPlus Production Units. 13c 12e 13a 13b 11b 12a 12b 11e 12f 12c 11a 12d 13d ± 0 100 20050 Miles Figure 12. NHDPlus Production Units contributing flow to Texas waterways. 38 34 21 69 14 7 22 0 12 15 10 17 16 11 18 20 19 2 1 5 8 13 ± 0 100 20050 Miles Texas River Basins 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Figure 13. Major river basins of Texas. 39 NHDPlus has a hierarchy of nested drainage areas based on the United States Geologic Survey’s Hydrologic Unit system (Figure 14) (Seaber et al 1987). Figure 14. USGS Hydrologic Regions. The top-most level of NHDPlus classification, NHDPlus Regions, are analogous to USGS Hydrologic Regions; there are 18 in the contiguous U.S. and one each for Alaska, Hawaii, and Puerto Rico for a total of 21 (Table 5). 40 Table 5. USGS and NHDPlus Hydrologic Units. USGS NHDPlus Hydrologic Units Count Example Hydrologic Units Count Mean Area (km 2 ) Hydrologic Region 21 12 – Texas Gulf Region 21* 415,000 - - - Production Unit 61* 130,000 Subregion 222 1210 – Central Texas Coastal Subregion N/A** - Accounting Unit 352 121002 – Guadalupe Basin Watershed N/A** - - - - Subwatershed N/A** - - - - Basin N/A** - Cataloging Unit 2150 12100203 – San Marcos, Texas Subbasin 2117* 3,700 - - - Catchment 2,614,642 3.0 *NHDPlus for Alaska and Puerto Rico is currently in progress and thus not included in the counts. **Empty, but maintained within the NHDPlus schema as placeholders. 4.1.3 TEXAS HYDROGRAPHY There are 211 NHDPlus subbasins (i.e., 8-digit HUCs) which lie wholly or partly within the boundaries of Texas, with an average area of approximately 3,300 square kilometers (Figure 15 and Figure 16); these 211 subbasins were used as the base unit for the stream classification system. 41 Figure 15. Subbasins of Texas. Figure 16. NHDPlus subbasin attributes. 42 4.2 Water Quality 4.2.1 TCEQ TRACS An understanding of water quality (here synonymously used with the term “water chemistry”) is important to determine the suitability and quality of riverine habitat as well as to assess the level of impairment of a water body. Thus, the entire surface water quality database for Texas was acquired at CRWR for incorporation into this stream classification system. The TCEQ Regulatory Activities and Compliance Systems (TRACS) Surface Water Quality Monitoring (SWQM) system includes 7.5 million records from 733,000 sampling events at 7,138 stations measuring 1,072 parameters from 1968 through August 14, 2006 (Figure 17) (TCEQ 2007). 1 Water quality parameters included here (and all parameters in general) were selected to provide a synoptic picture of stream type and to minimize, as much as possible, the consideration of anthropogenic effects. For water quality, the intent is not to measure pollution or human impact explicitly. However, given the current and historic use of Texas waterways, it is not feasible to separate out altered conditions from natural conditions for water quality without sophisticated modeling and data reconstruction techniques. 1 Data are still being collected at present. However, the TCEQ is in the process of transitioning the TRACS SWQM database into a new format, SWQMIS. Data collected since August 2006 are only being stored in the new system. 43 Figure 17. TRACS SWQM stations, 1968-2006. 4.2.2 WATER QUALITY VARIABLES The water quality parameters considered here include: y water temperature, y dissolved oxygen, y pH, y specific conductance, and y total suspended solids (i.e., total nonfiltrable residue) and rank as the first, second, third, fourth, and seventh most frequently sampled parameters in the SWQM database (Figure 18 and Table 6). Water temperature is 44 measured and recorded in both degrees Celsius and Fahrenheit (ranked #1 and #10 with EPA Storage and Retrieval (STORET) codes 00010 and 00011, respectively). An examination of the data has revealed that the majority of the Fahrenheit data was collected prior to 1986 and is recorded redundantly with Celsius data; that is, both records for water temperature from a given sampling event carry the same TagID identification. 0 100 200 300 400 500 600 700 W a te r T e m p ( D eg. C) Di ss ol ve d Ox y g e n pH F i el d S pec i f i c C o nduc t anc e Ch l o ri d e Su l fa te To t a l N o n f ilt e r a b l e R e s i d u e Amm o n i a P hos phor us W at er T e m p ( D eg. F ) F ec al C ol if o rm O rt hp hos phat e S a lin i t y A l k a l in it y T ot al O r ga ni c Ca r b on Thou s a n d s N u m b er o f R eco r d s Figure 18. SWQM top fifteen parameters by result (Jantzen 2007). 45 Table 6. Water quality parameters used in the stream classification system. WQ Parameter STORET Code Water Temperature 00010 Specific Conductance 1 00094 Dissolved Oxygen 00300 pH 00400 Total Suspended Solids 2 00530 1. Specific conductance considered in lieu of salinity (STORET 00480), eliminated from consideration due to poor spatial and temporal coverage. 2. TSS considered in lieu of turbidity (as measured by STORET 82079, field measure; 82078, lab measure; or 61028 unfiltered) due to poor spatial and temporal coverage. The SWQM database at CRWR is stored in both Microsoft Access and Structured Query Language (SQL) Server formats; the former is only 920 megabytes in size while the latter is 7.22 gigabytes. A series of SQL queries were written in Access to extract water quality data by parameter and then aggregate it by subbasin. That is, the 8-digit HUC code for each of the 7,138 stations was appended to the sample attribute (Event) table in ESRI ArcMap software via the spatial join tool, the data were extracted from the Results table by station, and the data for the entire period of record from all stations within a given subbasin were averaged together to create a single value for each subbasin for each parameter. Of the 211 subbasins in Texas, water quality data for each of the five parameters of interest was available for 156 subbasins for DO and water temperature, 155 for pH and specific conductance, and 148 for total suspended solids (Figure 19, Figure 20, Figure 21, Figure 22, Figure 23, and Figure 24). Data was also available for each of the five parameters of interest in the Gulf of Mexico. Dissolved oxygen records were filtered to include only samples taken at a depth less than 1.6 meters (5.25 feet) to eliminate redundant records from depth profile sampling at the same site and time. 46 Subbasins with Water Quality Data Other Subbasins Figure 19. Subbasins with any water quality data in TRACS, 1968-2006. 47 Mean Water Temperature by HUC (degC) 12 - 15 15 - 18 18 - 21 21 - 24 24 - 27 27 - 30 Figure 20. Mean water temperature (in degrees C) by subbasin. 48 Mean Dissolved Oxygen by HUC (mg/L) 4 - 5 5 - 6 6 - 7 7 - 8 8 - 9 9 - 10 10 - 11 Figure 21. Mean dissolved oxygen by subbasin. 49 6.0 - 6.5 6.5 - 7.0 7.0 - 7.5 7.5 - 8.0 8.0 - 8.5 8.5 - 9.0 Mean pH by HUC Figure 22. Mean pH by subbasin. 50 Mean Specific Conductance by HUC (umhos/cm) 0 - 500 500 - 1000 1000 - 2000 2000 - 5000 5000 - 10,000 10,000 - 50,000 Figure 23. Mean specific conductance by subbasin. 51 < 50 50 - 100 100 - 500 500 - 1000 > 1000 Mean Total Nonfiltrable Residue (aka Total Suspended Solids) by HUC (mg/L) Figure 24. Mean total nonfiltrable residue (i.e., total suspended solids) by subbasin. As can be seen in the above figures, Texas streams exhibit, on average: y Warmer temperatures to the south and along the coast y Lower dissolved oxygen at the coast and in east Texas y Higher acidity in east Texas y Higher specific conductance (~salinity) along the coast (includes tidal systems), in the Pecos and Red River basins y Higher total suspended solids (TSS) in the Red River and Brazos River basins. 52 Water quality data were examined for redundancy and correlation using the square of the Pearson product-moment correlation method (R 2 ), also known as the coefficient of determination. R-squared ranges from 0 to ±1 and describes the percent of variation in Y that can be explained by variation in X. Each water quality parameter was tested against every other for a total of 10 tests. In general, one parameter could explain only 0 to 11 percent of variation in each other parameter except for dissolved oxygen, where variations in DO were able to explain 28% of the variation in pH (Figure 25). Results of this analysis indicate that the level of redundancy between water quality variables chosen is low. This implies that it is appropriate to include all these variables in the classification system, as each describes different components of stream type and represent different sources of control and variation. 53 Figure 25. Tests for correlation between water quality parameters, grouped by: DO (top four), temperature (three), TSS (two), and specific conductance (one). Note: scales and correlated variable are not important in this case; plots are simply meant to depict scatter in the data. 4.3 Climatology 4.3.1 DATA SOURCES Climatology is a driver of habitat and hydrology on a macro-scale. In this project, mean annual temperature, mean annual precipitation, and mean annual potential evapotranspiration (PET) were variables considered for their classification potential. Data for average annual precipitation and average annual temperature are included in the CatchmentAttributesTempPrecip table within NHDPlus and are derived from the Parameter-elevation Regressions on Independent Slopes Model (PRISM) (PRISM Group 54 2006). PET data were obtained from the Texas Evapotranspiration Network from the Irrigation Technology Center at the Texas Water Resources Institute of the Texas A&M University System (ITC 2005). 4.3.2 CLIMATOLOGY VARIABLES Precipitation exhibits a strong east-west gradient across Texas, with eastern regions being much wetter than arid western regions (Figure 26). Temperature exhibits a strong gradient as well, with south Texas and the lower elevation coastal plain being warmer than north Texas and the higher elevation areas of the panhandle and west Texas (Figure 27). 55 Average Annual Precipitation by HUC (in) 10 - 15 15 - 20 20 - 25 25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 Figure 26. Mean annual precipitation by subbasin. 56 Average Annual Temperature by HUC (degC) 12 - 13 13 - 14 14 - 15 15 - 16 16 - 17 17 - 18 18 - 19 19 - 20 20 - 21 21 - 22 22 - 23 Figure 27. Mean annual temperature by subbasin. PET data consist of monthly average values for 19 selected Texas cities calculated from National Weather Service data with periods of record ranging from 26 to 99 years and an average of 56 years (Table 7). Mean annual PET data were interpolated across the State using the inverse distance weighting method and averaged by HUC; for the northernmost HUCs in the panhandle of Texas (the ‘white space’ at the top of Figure 28), PET values were extrapolated by assuming similarity to the PET values calculated for the City of Amarillo (Figure 28 and Figure 29). In general, PET is higher in arid west Texas and lower along the more humid coast and east Texas regions. 57 Table 7. Average monthly PET at selected Texas cities (from ITC 2005). 58 Figure 28. Interpolated mean annual PET. 59 51 - 54 54 - 57 57 - 60 60 - 63 63 - 66 66 - 69 69 - 72 72 - 75 75 - 78 Potential Evapotranspiration by HUC (inches/year) Figure 29. Mean annual PET by HUC. 4.4 Hydrology and Hydraulics 4.4.1 DATA SOURCES Hydrology is the ultimate controlling factor on riverine ecosystems and hydraulics is the physical manifestation of the movement of water within a river channel and its floodplain. The hydrology and hydraulics variables considered in this stream classification system include: 60 y Mean annual streamflow y Mean annual stream velocity y Base flow Index y Percent of zero flow days y Flow Variability, as expressed by the Interquartile Range (IQR). The incorporation of mean annual streamflow (MAF) data into the NHD was one of the main motivating factors behind the creation of NHDPlus (USEPA and USGS 2006). In NHDPlus, three different methods are used to estimate average streamflow and the results from all three are included in the dataset. The Unit Runoff Method (UROM) (Research Triangle Institute 2001) and the Vogel Method (Vogel et al. 1999) are both used to calculate mean annual streamflow at the bottom of a flowline and the estimates generated by these two methods can be found in the FlowlineAttributesFlow attribute table. Similarly, one attribute of the StreamGageEvent layer is AVE, the average daily flow for the period of record at every USGS stream gage in NHDPlus. In contrast to the UROM and Vogel estimates which are calculated and reported at each flowline in NHDPlus, the average daily flow is only calculated and represented at stream gage point features. In addition, while the UROM and Vogel estimates are modeled streamflows for every stream reach in the United States, the average daily flow is calculated from actual data (USEPA and USGS 2006). The UROM and Vogel methods both rely on the 1,338 national Hydro-Climatic Data Network (HCDN) gages. The HCDN subset of gages are selected from the USGS NWIS stream gage network because they are believed to be less affected by human activities and thus flow conditions recorded at these gages are likely more representative of natural conditions. The UROM method estimates a unit discharge for an ungaged site from a distance-weighted average of unit discharge from up to five HCDN gages within a 200-mile search radius. The unit discharge is then multiplied by the catchment area at each point of interest to generate an incremental streamflow, and the mean annual flow for each flowline in NHDPlus is calculated by summing these incremental flows. The Vogel method uses a log-log regression incorporating drainage area, precipitation, temperature, and multiple region-specific coefficients derived for 18 hydrologic regions 61 of the country. Estimates of mean annual streamflow in NHDPlus are individually derived using the Vogel method at the bottom point of each flowline (Vogel et al. 1999, Research Triangle Institute 2001, USEPA and USGS 2006). The average daily flow values stored in the StreamGageEvent layer were calculated from the USGS NWIS database. Approximately 23,000 stream gages nationwide were snapped to the NHD medium-resolution flowlines for use in NHDPlus. The flow statistics included in NHDPlus were calculated for the period of record for each streamflow gage from the date of first measurement through June 15, 2005 (USEPA and USGS 2006). The mean annual flow values included in this dataset were used in classification system here as they are believed to be more representative of actual streamflow conditions across the State and are calculated from measured data (as opposed to the modeled data from the UROM and Vogel methods). Mean annual velocity (MAV) is another attribute in the NHDPlus FlowlineAttributesFlow table and is calculated from both MAF methods. MAV is estimated from regression analyses performed on hydraulic variables (drainage area, flowline slope, mean annual discharge, and discharge at the time of the measurement) from 980 time-of-travel studies representing 90 rivers in the United States. The resulting set of regression equations relates stream velocity to actual drainage area, dimensionless drainage area, streambed slope, actual discharge, and dimensionless relative discharge (Jobson 1996, USEPA and USGS 2006). Base flow is another important control on habitat availability. Base flow is the portion of stream discharge that is not attributable to direct runoff from precipitation or snowmelt and is usually sustained by throughflow and groundwater flow; base flow can be thought of as the typical flow condition of a river in the absence of a rain event and ranges from 0 to 1 as the proportion of streamflow derived from base flow. Base flow data from the United States Bureau of Reclamation’s BFI program are included in NHDPlus in the StreamGage layer, including: y BFIyrs: number of years of flow data used in the base flow index (BFI) calculation y BFI_Ave: average annual base flow index y BFI_Stdev: standard deviation of the annual base flow index 62 y GotBFI: flag indicating the presence/absence of BFI data. Zero flow days are important in their role as a stressor on the aquatic ecosystem, both in their role in the lifecycles of native species and in controlling and managing non- native, invasive species. The percentage of zero flow days was calculated from the NHDPlus StreamGage layer by subtracting the total number of non-zero flow days (NDaysGT0) from the total number of days of flow data (NDays). Similarly, variability of a flow regime is another important control in a riverine environment, here represented by the daily streamflow IQR. The IQR was calculated from the NHDPlus StreamGage layer by subtracting the 25 th percentile of daily flow (P25) for the period of record from the 75 th percentile (P75). Using these particular flows provides an understanding of the spread of daily flow data without being disproportionately affected by the extreme hydrologic events, either flood or drought, which often control the upper and lower flow quartiles. The NHDPlus StreamGage layer contains 918 gages in Texas. Of these: y 730 gages have greater than or equal to 1 year of daily flow data; y 558 gages have greater than or equal to 10 years of daily flow data; and y 427 gages have greater than or equal to 20 year of daily flow data. The subset of 427 gages with at least 20 years of record was chosen for analysis. Twenty years is believed to be the minimum daily time-series record to sufficiently represent long-term hydrologic conditions for the purposes of the study and to minimize the importance of extended hydroclimactic aberrations. 4.4.2 HYDROLOGIC AND HYDRAULIC VARIABLES The mean annual flow was divided by the contributing drainage area to permit direct comparison between stream gages. As can be expected given the precipitation gradient in the State, MAF also exhibits a strong east-west pattern with much higher normalized streamflows in east Texas than west Texas and the panhandle (Figure 30 and Figure 31). 63 Mean Annual Streamflow Normalized by Drainage Area (cfs/sq mi) 0 - 0.09 0.09- 0.21 0.21 - 0.37 0.37 - 0.60 0.60 - 0.92 0.92 - 1.42 1.42 - 2.31 2.31 - 4.71 Figure 30. Mean annual streamflow, normalized by contributing drainage area. 64 Mean Annual Streamflow Normalized by Drainage Area, by HUC (cfs/sq mi) 0.25 - 0.50 0.50 - 0.75 0.75 - 1.00 1.00 - 2.00 No Data 0 - 0.05 0.05 - 0.15 0.15 - 0.25 Figure 31. Mean annual streamflow, normalized by contributing drainage area and grouped by HUC. Mean annual velocity appears to be patterned along the lines of major river basin, with subbasins in the Brazos and Colorado Rivers exhibiting, on average, higher stream velocities than streams in other major basins of the state (Figure 32). 65 Mean Annual Velocity by HUC (ft/s) No Data 0 - 1.00 1.00 - 1.16 1.16 - 1.32 1.32 - 1.48 1.48 - 2.00 Figure 32. Mean annual stream velocity, by HUC. In a similar spatial averaging manner as described above, the BFI from the stream gage points within each subbasin was averaged to obtain a mean subbasin BFI (Figure 33 and Figure 34). The presence of isolated springs in the Comal, Frio, and Devils Rivers is strongly evident in the data via higher average base flow indices, as is the artesian zone of the Edwards Aquifer and the generally wetter systems of east Texas (as compared to west Texas). 66 Average Base Flow Index (BFI) 0 - 0.10 0.10 - 0.25 0.25 - 0.50 0.50 - 1.00 Figure 33. Mean base flow index (BFI) by streamflow gage. 67 Average Baseflow Index (BFI) by HUC 0.0 - 0.2 0.2 - 0.40 0.40 - 0.60 0.60 - 0.80 Figure 34. Mean BFI by subbasin. Just as with the normalized MAF, the percent of zero flow days exhibits an east- west gradient, where very few streams in east Texas are absent of streamflow for any significant number of days (i.e., a greater prevalence of perennial streams), whereas streams in south and west Texas are more likely to experience a majority of days without flow (i.e., a greater prevalence of intermittent streams) (Figure 35 and Figure 36). 68 Percent of Zero-Flow Days 0 - 1 1 - 5 5 - 10 10 - 25 25 - 50 50 - 75 75 - 100 Figure 35. Percent of zero flow days. 69 No Data 0 - 1 1 - 5 5 - 10 10 - 25 25 - 50 50 - 75 75 - 100 Percent of Zero-Flow Days by HUC Figure 36. Percent of zero flow days, by HUC. As with MAF, the IQR was normalized to allow for statewide comparison, in this case by median streamflow. Of the 427 stream gages analyzed having 20 or more years of streamflow data, 33 of the gages had a median streamflow equal to zero (Figure 37), that is, intermittent streams where the majority of days have no streamflow. This subset of 33 gages was excluded from the analysis of flow variability, leaving a sample population of 394 gages. Thus, gages were studied individually and aggregated by subbasin. From this analysis it can be seen that subbasins with a high base flow component and/or occupying a more downstream location in the same river network (i.e., closer to the coast within a river basin that drains to such) exhibit a lesser flow variability (Figure 38 and Figure 39). 70 Gages where the Median Flow is Zero (n=33) Gages where the Median Flow is Greater Than Zero (n=394) Figure 37. Intermittent streams in Texas (in red), distinguished as having a median streamflow equal to 0 cfs. 71 Interquartile Range (IQR) Normalized by Median Streamflow (cfs/cfs) 0 - 0.5 0.5 - 1 1 - 2 2 - 5 5 - 10 10 - 25 25 - 100 100 - 500 Figure 38. Interquartile range of daily streamflow normalized by median streamflow. 72 Interquartile Range (IQR) Normalized by Median Streamflow, by HUC (cfs/cfs) 10 - 25 25 - 100 100 - 200 1 - 2 2 - 5 5 - 10 No Data 0 - 0.5 0 - 1 Figure 39. Interquartile range of daily streamflow normalized by median streamflow and grouped by HUC. 73 4.5 Geomorphology and Physical Processes 4.5.1 DATA SOURCES “In combination with the hydrologic flow regime, [the physical features of a channel and floodplain] form the habitats to which all biological elements in the river ecosystem have adapted and become dependent” (TIFP 2006). Thus, an understanding of both the physical habitat and the processes and controls that act upon such habitat is of immense importance to developing an understanding of stream ecosystems. In the proposed classification system, the processes of sediment formation, transport, and deposition are respectively represented by consideration of: y watershed soils composition, y channel bed slope, and y stream substrate composition. Soils and geology are important drivers of water quality, hydrology, geochemistry, substrate composition, channel and floodplain shape and valley confinement (cross-sectional), and planform geomorphology (i.e., sinuosity). They act as a control on the riverine system both at the site of interest through local channel and substrate conditions and through upstream conditions such as infiltration and runoff rates, geochemistry, and sediment load. Soil composition data have been obtained for Texas from the Conterminous United States Multilayer Soil Characteristics Dataset (CONUS-SOIL) from the Earth System Science Center of Pennsylvania State University (Miller and White 1998). CONUS-SOIL is derived from the U.S. Department of Agriculture (USDA) Natural Resource Conservation Service (NRCS) State Soil Geographic Database (STATSGO). The data include information on soil texture, including: percent clay, percent silt, and percent sand. Channel bed slope is an important driver of habitat availability as it is a primary factor in determining flow velocity and mesohabitat (pool, riffle, run, etc). Bed slope for each linear stream reach is included in the FlowlineAttributesFlow table in NHDPlus and is calculated by taking the difference in the upstream and downstream node elevations 74 from the National Elevation Dataset divided by the reach length from the National Hydrography Dataset (Figure 40). Also, the marriage of NHD and NED in NHDPlus allows for the examination of the longitudinal stream bed profiles of rivers and streams (Figure 41). These profiles describe the way a stream’s elevation (vertical axis) changes over its distance downstream (horizontal axis). ± 010205 Miles Reach Slope (ft/ft) 0 0.000001 - 0.005 0.005 - 0.01 0.01 - 0.02 0.02 - 15 Figure 40. Example NHDPlus map for the San Marcos basin, Texas, depicting channel slope by reach. 75 Figure 41. Longitudinal profile of the Colorado River, Texas. Note the step-shaped reaches between kilometers 800 and 550, which are the Highland Lake system reservoirs and dams, with the largest vertical reach (approximately kilometer 590) being Mansfield Dam at Lake Travis. 4.5.2 GEOMORPHOLOGY AND PHYSICAL PROCESSES VARIABLES Soil texture of the contributing watersheds of the State, as represented by percent clay, percent silt, and percent sand (Figure 42, Figure 43, and Figure 44), was assessed and averaged by subbasin (Figure 45, Figure 46, and Figure 47, respectively). It can be seen in these figures that the Brazos and Trinity basins, west Texas, and the coastal basins have high clay content, the Devils River basin has high silt content, and the high plains, south Texas, and east Texas regions have high sand content. 76 Percent Clay 0 - 25 25 - 50 50 - 75 Figure 42. Soil clay composition, in percent. Percent Silt 0 - 25 25 - 50 50 - 75 Figure 43. Soil silt composition, in percent. Percent Sand 0 - 25 25 - 50 50 - 75 75 - 100 Figure 44. Soil sand composition, in percent. 77 Percent Clay by HUC 10 - 20 20 - 30 30 - 40 40 - 50 50 - 60 Figure 45. Percent clay by subbasin. Percent Silt by HUC 10 - 20 20 - 30 30 - 40 40 - 50 50 - 60 Figure 46. Percent silt by subbasin. 0- 20 20 - 40 40 - 60 60 - 80 Percent Sand by HUC Figure 47. Percent sand by subbasin. 78 The reach slope within each subbasin was averaged to obtain a mean subbasin bed slope (Figure 48). It is evident that streams are steeper up on the high plains, in west Texas, along the lower Rio Grande, and coming down off the Edwards Plateau than elsewhere in the State. Mean Bed Slope by HUC (ft/ft) 0.0000 - 0.0006 0.0006 - 0.0014 0.0014 - 0.0022 0.0022 - 0.0035 0.0035 - 0.0055 0.0055 - 0.0103 0.0103 - 0.0158 0.0158 - 0.0258 0.0258 - 0.1109 0.1109 - 0.2500 Figure 48. Mean reach bed slope by subbasin. Channel bed slope was tested for redundancy and correlation with water quality parameters using the same method described for the water quality parameters. As was the case with those parameters, variations in bed slope by subbasin could only explain 1 79 to 8% of the variation in the water quality parameters (Figure 49). It is possible that stronger linkages between slope and water quality and among water quality parameters may exist on a reach-scale (not tested) than indicated on a subbasin scale (tested). Variation between headwater streams and mainstem rivers and or other physical categories may exist within subbasins, but was not explicitly tested. Figure 49. Tests for correlation between each water quality parameter and channel bed slope. Note: scales and correlated variable are not important in this case; plots are simply meant to depict scatter in the data. The USGS has been compiling and expanding a National Geochemistry Survey to “produce a body of geochemical data for the United States based primarily on stream sediments, analyzed using a consistent set of methods” in order to “enable construction of geochemical maps, refine estimates of baseline concentrations of chemical elements in the sampled media, and provide context for a wide variety of studies in the geological and environmental sciences” (USGS 2004). Accessible at http://tin.er.usgs.gov/geochem/, the database contains records for 2,710 sites in Texas, including 2,379 stream sediment samples (88%) and the remainder being soil samples (Figure 50); only the stream sediment samples were included in the analyses herein. 80 Figure 50. USGS National Geochemistry Survey database sample sites in Texas. The geochemistry database contains a wealth of information on: y sample y geography y source y in situ sample y substrate y channel y streamflow y velocity 81 y possible sample contamination and sources y water quality y vegetation y metals by spectrometry and neutron activation (as percent weight or concentration) y gasses by atomic absorption y total and organic carbon y fertilizers. TCEQ TRACS also contains information on dominant substrate type (STORET code 89844) from stream samples taken across the State since the 1960s. These data are available at 347 stations with a total of 775 coded records, following the same codes as the National Geochemistry Survey (Table 8). Table 8. TCEQ TRACS and National Geochemical Survey dominant substrate type code key (STORET 89844). Dominant Substrate Type Code Clay 1 Silt 2 Sand 3 Gravel 4 Cobble 5 Boulder 6 Bedrock 7 Other 8 The 1785 stream substrate data records from the National Geochemistry Survey were combined with the corresponding data from TCEQ TRACS for a total of 2560 samples statewide (Figure 51). These data were counted by 8-digit HUC and the most frequently-occurring substrate type within each HUC was reported for each HUC (Figure 82 52). Although the relative frequency of dominant substrate types within each subbasin is not necessarily an unbiased estimator of the most common substrate types within an area, it is nonetheless a worthwhile means of assessing the relative substrate types and sizes between watersheds. As can be seen, silt and particularly sand are the dominant substrate types in the rivers and streams of Texas, with some regions, particularly the Edwards Plateau and central Texas, having larger dominant substrate types such as gravel, cobble, and even bedrock. When the subbasins are considered in this counted fashion: none had boulders as the dominant substrate type; two had bedrock; and some, particularly along the coast and in the Sulphur and Nueces Basins,, had clay dominating. Substrate Type Clay Silt Sand Gravel Cobble Boulder Bedrock Figure 51. Dominant stream substrate type. 83 Clay Silt Sand Gravel Cobble Bedrock Substrate Type by HUC No Data Figure 52. Dominant stream substrate type, by HUC. 4.6 Biology 4.6.1 DATA AVAILABILITY CHALLENGES Discussions and data exploration exercises conducted as part of the initial phase of this project revealed that biological data are not as well developed or as accessible as other data relevant to stream classification, and the timeframe of various ongoing and future biological data organization endeavors in the State was not conducive for their systematic inclusion in this classification scheme. For these reasons, it was decided that biologic information would not be explicitly incorporated into the initial development 84 phase of the stream classification system. However, the system has been designed to be robust enough to accommodate the future addition of biologic information. Due in part to the complexity of environmental systems and the sizeable resources required for sample collection, biological data have: y comparatively limited spatial and temporal coverage, y greater complexity in the data structure, y greater flexibility required in the data storage framework, and y the ability to predict presence of species with confidence, but generally not the ability to predict the absence. Conceptually, biological data can be viewed as a response variable to the determining factors of physical, chemical, and hydrologic variables. It then follows that driving factors can first be assembled in a predictive classification which can then be subsequently calibrated and validated to the biological data via a multiple-regression style analysis. This is the proposed methodology for future stream classification work. 4.6.2 BIO-AQUATIC INFORMATICS FOR TEXAS WORKGROUP A biological data workgroup was formed in 2006 with the mission to: discover, deliver, and publish biological data in Texas using a common technology and format. The group has been actively meeting monthly to discuss the issues of biological data discovery, organization, and access. The Bio-Aquatic Informatics for Texas (BAIT) Workgroup has discussed various current and proposed datasets and structures (including USGS, TCEQ, and TPWD scientific collection permit data) and is currently working on a benthics data discovery web portal. 4.6.3 DATA SOURCES An exploration was made of the coverage and type of biological data within the TRACS SWQM database. Prior to this analysis it was generally believed that the bio- data coverage within SWQM was limited; these quantitative analyses have confirmed this suspicion. STORET parameter codes within TRACS were divided into groupings of biologic and ecologic significance (Table 9). SQL Queries were then performed in MS Access to 85 extract appropriate data and statistics, and the results of these queries were summarized (Table 10). From these analyses, it was confirmed that TRACS contains relatively little data (number of records) specific to biology, but a large proportion (over 55%) of the codes in TRACS are dedicated to biologic data. Table 9. Groupings of STORET parameters developed for analysis of the TRACS SWQM database. 86 Table 10. Summary of biologic data in TRACS SWQM. Additional biological/ecological data sources actively being explored and considered for possible future inclusion: y A sizeable (on the order of 16,000 unique records from 31 museum collections worldwide) georeferenced database of fishes of Texas from the Texas Natural Science Center Texas Natural History Ichthyology Collection y Fishes and habitat dataset from Professor Timothy Bonner of Texas State University from sampling and analyses conducted on the Blanco River y USGS National Water Quality Assessment (NAWQA) Program data for fishes and benthics. 87 5. A STREAM CLASSIFICATION SYSTEM FOR TEXAS 5.1 Data Integration 5.1.1 GENERALIZED DISTRICTS A meeting of stakeholders and users of the integrated stream classification system was held on April 6, 2007 to share ideas and solicit feedback and suggestions; attendees included researchers, state agency officials, and environmental organization representatives. One goal was to evaluate the generalized districts presented in the NRC report and make consensus-based modifications based on a collective wealth of experience with the issues, conditions, and waterways of the State. The stakeholder panel was largely in agreement with the districts as originally presented, with the one exception being that the conditions typical of the Devils River basin and its tributaries are more commonly observed in South-Central Texas streams than West Texas streams. Thus, four subbasins were switched from the Lower Rio Grande Basin district to the South Central Texas district (Table 11 and Figure 53) and the generalized districts were revised accordingly to reflect the consensus-based regionalization (Figure 54). Table 11. Devils River subbasins switched from the Lower Rio Grande Basin to the South Central Texas Basin based on stakeholder consensus. HUC Name 13040301 Upper Devils 13040302 Lower Devils 13040303 Dry Devils 13070011 Howard Draw 88 East Texas North-Central Texas West Texas South-Central Texas Lower Rio Grande Basin switched subbasins Figure 53. Devils River subbasins switched from the Lower Rio Grande Basin to the South Central Texas Basin based on stakeholder consensus. 89 East Texas North-Central Texas West Texas South-Central Texas Lower Rio Grande Basin Figure 54. Stakeholder consensus-based revision of generalized NRC districts. These five districts were used as the baseline case for data integration and evaluation and served as the point for departure in the development of the integrated stream classification system for Texas. 5.1.2 DISTINGUISHING PARAMETERS Eighteen distinguishing parameters from four disciplines were incorporated into the stream classification system (Table 12). 90 Table 12. Distinguishing parameters of the riverine environment incorporated into the stream classification system and their units. 5.1.3 REDUNDANT SUBBASINS Of the 211 subbasins (8-digit HUCs) contained in Texas, 199 are unique and contiguous, meaning that those straddling the state boundary are diminished in size from their original representation in NHDPlus, but the entire contributing area is contained within one cohesive polygon. Additionally, six subbasins at the state boundary have been divided such that two different polygons with the same hydrologic unit code are included within the total count of 211 (Figure 55 and Table 13). Thus, there are a total of 205 unique HUCs in Texas but 211 total subbasins. 91 Figure 55. The six subbasins at the state boundary and have been divided into multiple polygons. 92 Table 13. Attributes of the six subbasins which lie at the state boundary and have been divided into multiple polygons. 5.2 Analysis of Original Generalized Districts An analysis was made of the eighteen distinguishing parameters based on grouping by the original NRC Generalized Districts from the “Physical Settings for Instream Flows” description (NRC 2005) (Table 14, Table 20, and Appendix B – Supporting Data). In addition, the qualitative distinctions highlighted in the NRC Report were tested with the results of the quantitative analysis based on the data types examined here for: East Texas (Table 15), North Central Texas (Table 16), South Central Texas (Table 17), the Lower Rio Grande Basin (Table 18), and West Texas (Table 19). 93 Table 14. Count and area of subbasins grouped by the original generalized districts in Texas. Table 15. Comparison of qualitative distinctions and quantitative data for East Texas. NRC Qualitative Distinction Quantitative Result Assessment 30-50 inches average precipitation 36-56 inches average precipitation Accurate Flat landscapes Lowest average stream slope (0.0008 ft/ft) Accurate Clay-rich or sandy soils Median percentage of clay and sand content of all regions; more clay and sand that silt Inconclusive distinction High flow variations Median normalized IQR Inconclusive distinction High turbidity Lowest TSS (and specific conductance) of all regions Not supported by the TSS data, although turbidity is not necessarily directly related to TSS Soft, shifting substrate Smallest substrate class (silt) of any region Accurate 94 Table 16. Comparison of qualitative distinctions and quantitative data for North Central Texas. NRC Qualitative Distinction Quantitative Result Assessment 15-28 inches average precipitation 15-39 inches average precipitation Generally accurate Clay-rich soils Second-highest clay content of all regions Accurate Flood pulses & high flow variations (drought/flood) Second-lowest IQR of all regions Inaccurate on a daily time step; inconclusive over longer timeframes Table 17. Comparison of qualitative distinctions and quantitative data for South Central Texas. NRC Qualitative Distinction Quantitative Result Assessment 10-40 inches average precipitation 18-46 inches average precipitation Accurate Clear water Second-lowest TSS of all regions Accurate Cool water Second-warmest water temperature of all regions Not supported by the data High base flow index Highest BFI of all regions Accurate 95 Table 18. Comparison of qualitative distinctions and quantitative data for Lower Rio Grande Basin. NRC Qualitative Distinction Quantitative Result Assessment 11-26 inches average precipitation 20-27 inches average precipitation Generally accurate Rio Grande occasionally reduced to series of isolated pools Lowest mean stream velocity, second-highest percentage of zero flow days of all regions Inconclusive based on available data Rio Grande occasionally fails to reach Gulf of Mexico Second-highest percentage of zero flow days of all regions Inconclusive based on available data, but observed Stressed aquatic biota Highest water temperature, highest pH, highest specific conductance, highest air temperature, lowest stream velocity, second-highest percentage of zero flow days, lowest IQR of all regions Inconclusive but probable based on available data Table 19. Comparison of qualitative distinctions and quantitative data for West Texas. NRC Qualitative Distinction Quantitative Result Assessment 8-16 inches average precipitation 11-19 inches average precipitation Accurate High salinity in Pecos Median specific conductance of all regions (for entire West Texas region) but high in Pecos reaches Accurate 96 Table 20. Mean, standard deviation, and coefficient of variation statistics for the original generalized districts of Texas; blank cells indicate insufficient data. 97 5.3 Revision Methodology and Results A methodology was devised to test the strength of the grouping as determined by the generalized districts from the NRC Report (2005) and to revise the groupings in a manner that would result in an improved stream classification. First, the subbasins which lie on the border between two or more groups were identified; there are 50 border subbasins of which 10 border two different neighboring regions (Figure 56 and Appendix B – Supporting Data). A program was written using pivot tables in conjunction with Visual Basic for Applications (VBA) macros in Microsoft Excel to test the value of switching each bordering subbasin into the neighboring district to see if the strength of the classification system improves. Since there are 50 bordering HUCs and 10 have dual neighbors, 60 such trials were conducted. During each trial: 1. one subbasin was switched to its neighboring generalized district, 2. the mean, standard deviation, and coefficient of variation for each of the 18 distinguishing parameters for each district were calculated, 3. the median of these 18 values was calculated for each district, 4. the 5 median values were compared with the 5 median values calculated using the same methodology on the original districts, 5. the change in medians (i.e., level of improvement or lack thereof) were calculated for each district, 6. the changes of all 5 districts were summed, and 7. the switch was determined to be beneficial if it resulted in a decrease in variability within the classes. Once a trial had determined if a particular switch was beneficial or not, the generalized districts were reset to their original groupings (i.e., a switched subbasin was flagged but returned to its original district), thus enabling the merit of every trial to be tested individually. 98 East Texas North-Central Texas West Texas South-Central Texas Lower Rio Grande Basin Figure 56. Border subbasins that were tested during the revision process. From this exercise, it was established that 35 of the 60 trials resulted in improvements to the generalized districts. These 35 successful trials encompassed 31 unique subbasins. In the four HUCs displaying redundancy, the trial which resulted in the greater improvement was retained and the other trial discarded. The 31 successful trials were next ranked in decreasing order of value; that is, trials that provided the greatest improvement to the system were the most highly ranked. The same testing procedure and program described above was then applied again, this time: 1. one beneficial subbasin was switched to the neighboring generalized district, 99 2. the same calculations and comparisons were made, 3. the switch was determined to be beneficial or not as above, 4. if beneficial, the switch was made permanent, and the next trial begun; 5. if not determined to be beneficial, the previous switch was discarded and the next trial begun. In this manner, each successive switch determined to be beneficial resulted in a cumulative improvement in the strength of the classification system. As such, 21 trials resulted in improvements and 10 were discarded, resulting in a 44 percent improvement in the sum of the coefficients of variation for the 5 revised districts. This set of trials can be thought of as the intermediate revised stream classification. The 21 beneficial switches were viewed in a geographic context and tested according to the following rule: no subbasin may be an island; i.e., a subbasin of one class must border a subbasin of the same class. In addition, subjective judgment was applied to determine if a subbasin shared a sufficient amount of border with neighbors in its own class. This was undertaken as an acknowledgement of the geographic scales of influence of the distinguishing parameters that dictate that close neighbors are more likely than not to share similar characteristics. Trade-off trials were evaluated to see whether the recommend switch provided enough improvement when a neighbor was also switched or if the original trial should be discarded. Ultimately, 25 switches were determined to be beneficial to the classification system, resulting in a 43% improvement in the sum of the coefficients of variation for the 5 revised districts (Table 21, Figure 57, and Figure 58). 100 Table 21. Subbasins moved between regions during the revision process. 101 North-Central Texas East Texas South-Central Texas Lower Rio Grande Basin Figure 57. Subbasins moved between regions during the revision process: original NRC districts. changed subbasins: original districts 102 West Texas East Texas South-Central Texas Lower Rio Grande Basin Figure 58. Subbasins moved between regions during the revision process: revised stream classes. 5.4 Revised Classes Based on the methodology presented herein, results of the testing and revision processes were incorporated into the NRC (2005) generalized districts with consensus- based stakeholder revisions to produce an integrated stream classification system for Texas based on 18 distinguishing parameters encompassing watershed and stream channel processes and functions from four disciplines (Figure 59 and Figure 60). This integrated stream classification system might be used to: (1) discern likely similarities changed subbasins: final districts 103 and differences between rivers and streams of the State, (2) remotely characterize stream segments for which resources are insufficient for detailed field studies, (3) recognize streams and watersheds of the State as having common identities, (4) allow conclusions drawn from an instream flow study from a particular river reach to have a wider applicability than the particular study site, and (5) assist in prioritization of rivers and reaches for future instream flow studies. Moreover, the increased range of depth of data resources collected and incorporated into the integrated stream classification system could be of value to stakeholders and regulators. East Texas North-Central Texas West Texas South-Central Texas Lower Rio Grande Basin Figure 59. Results of the revision process: the integrated stream classification system for Texas. 104 Brazos Red Colorado Rio Grande Trinity Nueces Canadian Neches Sabine Sulphur Cypress Lavaca Guadalupe San Antonio San Jacinto East Texas North-Central Texas West Texas South-Central Texas Lower Rio Grande Basin Figure 60. The integrated stream classification system for Texas, overlain with the major river basins of the state. 5.5 Analysis of Revised Integrated Stream Classes 5.5.1 ANALYSIS OF REVISED CLASSES Similar to that conducted on the original generalized districts, an analysis was made of the eighteen distinguishing parameters based on grouping by the revised stream classification (Table 22 and Table 23). From the data, some generalized patterns are evident. 105 Table 22. Count and area of subbasins grouped by the revised stream classes. 106 Table 23. Mean, standard deviation, and coefficient of variation statistics for the revised stream classes of Texas. 107 With regard to water quality: y East Texas streams have lower DO and are more acidic, likely evidence of higher organic matter content in the water column and sediment and thus more decay and greater biochemical oxygen demand. y West Texas and North Central Texas streams have higher DO, possibly linked to their lower mean water temperatures and a lower organic matter content. y All districts had coefficients of variation for DO of approximately 10 percent. y Water temperatures are highest in the Lower Rio Grande Basin and South Central Texas where the mean atmospheric (air) temperatures are higher. y All regions had a water temperature standard deviation of approximately 1 to 2 degrees Celsius and coefficients of variation of approximately 10 percent. y The mean and standard deviation of TSS is considerably higher in West Texas and North Central Texas and lower in East Texas and South Central Texas. y With the exception of East Texas, the pH of streams across Texas is very consistent, with an average pH of 8.0 +/- 0.05. East Texas streams had a mean pH of 7.4 +/- 0.5. y The coefficients of variation for pH for every district were between 0 and 6 percent. y The mean specific conductance is considerably higher in South Central Texas and the Lower Rio Grande Basin than elsewhere, possibly evidence of increasing stream salinities towards the downstream end of agricultural basins but also of tidally-influenced coastal streams. y The specific conductance varied widely within a district, with coefficients of variation of 100 to over 200 percent. 108 With regard to climatology: y The mean annual air temperature was considerably cooler in the higher elevation regions of North Central Texas and West Texas. y All mean air temperatures were within 1 to 2 degrees Celsius with coefficients of variation of approximately 10 percent. y The mean annual precipitation is considerably higher in East Texas and considerably lower in West Texas. y The precipitation was more variable in South Central Texas and North Central Texas, each of which had coefficients of variation of approximately 20 percent versus 13 to 15 percent for other regions. y PET was more uniform across the State than precipitation and temperature, as it is a measure of the potential for water to evaporate and transpire irrespective of the availability of water (surficial and as soil moisture) for these processes. y PET was higher in West Texas than East Texas, but coefficients of variation ranged from 1 to 6 percent. With regard to hydrology and hydraulics: y The mean annual normalized streamflow (streamflow contributed per unit of drainage area) was considerably higher in East Texas than anywhere else in the State. y There was insufficient gaged streamflow data from the USGS in the Lower Rio Grande Basin to derive statistics in that district; these data could be obtained from the International Boundary and Water Commission (IBWC). y Mean annual streamflow was highly variable in every region of the State with coefficients of variation ranging from 50 to 110 percent. y Mean annual stream velocity was fairly uniform across all regions, albeit slightly lower in the Lower Rio Grande Basin. Mean velocities were 109 typically 1.1 +/- 0.2 feet per second with coefficients of variation of around 10 to 15 percent. y The proportion of streamflow derived from base flow (or BFI) was higher in the wetter areas of East Texas, in South Central Texas where spring-fed streams are more common, and in the Lower Rio Grande Basin, possibly due to irrigation water return flows. y BFI was variable, with coefficients of variation ranging from 60 to 80 percent. y The proportion of days experiencing no flow exhibited a strong east-west gradient as expected; more perennial streams are present in wetter East Texas than arid west Texas where intermittent (i.e., ephemeral) streams in are more common. y West Texas had a considerably higher percentage of zero flow days than any other region as well as a relatively lower coefficient of variation, 60 percent versus 80 to 110 percent for the other regions. y The variability and flashiness of the streamflow regime was much higher in West Texas and South Central Texas than the remaining regions. The 75 th percentile of flow minus the 25 th percentile of flow (IQR) in West Texas is almost 17 times the median daily streamflow. y IQR was highly variable, with coefficients of variation of approximately 90 to 100 percent in East Texas and North Central Texas and nearly 300 percent in the flashier systems of South Central Texas. With regard to geomorphology and physical processes: y There was a gradient across the State in average reach slope, with the lower elevation regions of East Texas and South Central Texas having gentler bed slopes than the higher elevation regions of West Texas. y Bed slope exhibited a range of coefficients of variation, from 60 to 70 percent in the Lower Rio Grande Basin and East Texas to 200 to over 400 percent in North Central Texas and South Central Texas; this is indicative 110 of the different types of terrain and elevations present within each of these regions. For example, South Central Texas includes portions of the Texas Hill Country but also portions of the very flat coastal plain. y Mean stream substrate type was generally silt (STORET code 2) to sand (code 3) for all regions except West Texas, which was sand to gravel (code 4). y Substrate type was fairly uniform, with coefficients of variation in all regions except South Central Texas of approximately 20 to 30 percent; South Central Texas was more variable at 50 percent. y Watershed soil texture data indicate that there is a greater proportion of sand than silt or clay in every region of the State. y Proportions of each soil classification were relatively uniform between regions. That is, similar proportions of sand, silt, and clay were present in each region; the sole exception to this result is the reduced proportion of sand in West Texas. y Although it is evident in the absence from this data and not the presence, this discrepancy is possibly due to elevated gravel content in the watershed soils of West Texas, as the total percentage of soil texture composition for each region does not total 100 percent. Nonetheless, soil texture data within the CONUS-SOILS database was limited to these three soil classes. y The coefficients of variation of the three soil classes generally ranged from 10 to 30 percent, with the sand content in the Lower Rio Grande Basin and South Central Texas being slightly more variable at 60 and 40 percent, respectively. When the eighteen multidisciplinary distinguishing parameters are considered in aggregate, the median coefficients of variation for each revised stream class range from 15 percent for the Lower Rio Grande Basin to 31 percent for South Central Texas with an average median coefficient of variation for the stream classes of 25 percent. 111 Certain variables are much more uniform within a region than others, indicating that their controlling processes operate over a geographic scale which is broader. For example, coefficients of variation for the climatology variables were in the range of 1 to 21 percent with an average (mean) of only 9%, whereas TSS, specific conductance, and stream slope had average coefficients of variation of approximately 100, 160, and 190 percent, respectively. Thus, it is likely that these three parameters are controlled by more localized conditions such as land use, land cover, and local geology or that the variable mean is close to zero. In contrast, DO, water temperature, pH, and stream velocity all exhibited very low variability, meaning that they are likely controlled by more regional forcings. The process of applying the revision methodology resulted in quantitative improvements in the strength of the stream classification system over the original generalized districts (Table 24). In the climatology variables, the greatest change was observed in East Texas and particularly in precipitation, which had a 40 percent improvement. The greatest changes in geomorphology and physical processes were in West Texas, which experienced improvements in stream slope and substrate of 200 and 140 percent, respectively, and also in East Texas, which had a 50 percent improvement in dominant substrate type. 112 Table 24. Comparison (percent change) for the mean, standard deviation, and coefficient of variation following revisions. 113 5.5.2 ECOREGION COMPARISON The integrated stream classes developed here were compared to the Level III Ecoregions of Texas. The ecoregions arose from a federal-level interagency effort to develop a spatial framework of ecological units in the United States “within which biotic, abiotic, terrestrial, and aquatic capacities and potentials are similar.” (McMahon et al. 2001) The ecoregions were further refined in Texas via a cooperative efforts between state and federal agencies (Griffith et al. 2004). The ecoregions are “…based on the premise that ecological regions are hierarchical and can be identified through the analysis of the spatial patterns and the composition of biotic and abiotic phenomena… [including:] geology, physiology, vegetation, climate, soils, land use, wildlife, and hydrology.” (Griffith et al. 2004) Texas is large in both size and ecological diversity, containing 12 level III and 56 level IV (the finest classification level) ecoregions. When the integrated stream classes are viewed in comparison to the level III ecoregions, differences in the areas subtended become evident (Figure 61). With a few exceptions where an ecoregion is wholly contained within a stream class, the stream classes generally cut across the ecoregion boundaries and vice versa. These differences likely reflect: (1) the conceptual framework under which each classification system was developed; (2) the type and source of data incorporated into each system; and (3) the scales of influence of the incorporated data. As such, a primary difference is rooted in the relative importance and incorporation of primarily-aquatic versus primarily-terrestrial indicators. 114 West Texas North Central Texas East Texas South Central TexasLower Rio Grande Basin High Plains Chihuahuan Deserts Edwards Plateau Cross Timbers South Central Plains South- western Tablelands Southern Texas Plains Central Great Plains East Central Texas Plains Western Gulf Coastal Plain Texas Blackland Prairies Figure 61. Level III ecoregions of Texas overlain with the integrated stream classes. 5.6 Limitations The stream classification system presented here incorporates multiple recently developed and published datasets in an integrated fashion to develop a synoptic characterization of streams and watersheds on a broad geographic scale in Texas without any additional field work. As such, it can be used to draw generalizations and answer broad questions about the physical setting for instream flows research and study and for other related riverine analyses. It is not intended to be the final authority on integrated 115 stream classification systems methodology nor of definitive stream classification in Texas, but nonetheless represents a step forward in both of these areas. The integrated stream classification system for Texas might benefit from future work on: 1. A logical interpretation and integration of geologic data. A project was recently completed to digitize the Bureau of Economic Geology (BEG) Geologic Atlas of Texas (GAT) into both raster and vector formats at a 1:250,000 scale; these data are available from TNRIS. A challenge will be to interpret this data in the context of riverine research and to quantify the impact and relevance of geologic controls and forcings. 2. An evaluation of flowline versus subbasin representation. The distinguishing parameters incorporated exert control on differing spatial scales. As such, some of the parameters might be best represented on a subbasin-scale (by polygon), whereas others may be better represented on a reach-specific scale by flowline (by line). Representing the data in this polygon and line fashion may reveal additional information about particular streams and stream types within a subbasin. For example, small tributaries of the lower Brazos River could be classified differently than the mainstem of the river, a distinction likely to be of importance when characterizing the riverine ecosystem and probable species occurrence. 3. Study of the importance of contributory drainage area to water quality, hydrologic and hydraulic parameters. For example, the percentage of zero flow days and the flow variability are both likely to be affected by the size of drainage area contributing to the streamflow gaging location. One analysis method to determine the effect of drainage area would be to study the parameters in relation to stream order and/or stream size, both of which are value-added attributes within NHDPlus. 4. Further distinction between predominantly naturally-controlled parameters and those which are more affected by anthropogenic influence. One such parameter where a human-influenced signature is likely to emerge is the base flow index at sites located downstream of wastewater return flow outfalls. 116 5. Investigation into the correlation between salinity and specific conductance and between total suspended solids and turbidity, respectively. The specifics of turbidity measurement and the physics of solids in suspension add sources of error when directly comparing these two indices of stream water quality. For example, the mean TSS in East Texas streams was the lowest of any stream class. Based on collective field experience, it is commonly held that East Texas streams are more opaque and thus more turbid than other streams across the State. It is likely that the prevalence of many very small particles in the water column in East Texas results in a smaller total mass of TSS (as captured on filter paper) than streams with larger particles yet causes very high scattering of light (and thus very high turbidity), whereas fewer larger particles elsewhere would result in higher TSS and lower turbidity. 6. Further research into the evaluation of the level of redundancy of the variables incorporated here and the relative merit of each variable in distinguishing between and among stream classes. 7. Incorporation of biologic data, including information on: species presence/absence, relative abundance, and zoogeography. 117 6. CONCLUSIONS AND RECOMMENDATIONS A large number of recently developed and recently publicized data sources have made the present an exciting time for the study of riverine systems, and the near future looks to offer more of the same. Discussions with project participants and with peers near and far have focused on the need for a methodology to scale up the results of site-specific environmental flow studies (typically habitat-based) into a larger framework for management, regulation, and implementation; integrated stream classification systems are viewed by many to be a promising avenue to accomplish this task. To preserve ecological relevance and thus ensure the protection of a “sound ecological environment,” the development of stream classes must take into account the importance of multiple systems and processes typically grouped into the four disciplines of instream flows: (1) Hydrology & Hydraulics (including climatology); (2) Water Quality; (3) Geomorphology & Physical Processes; and (4) Biology. A qualitative regionalization of Texas streams and rivers is presented in the National Research Council Committee (2005) review of the Texas Instream Flow Program to familiarize readers with the “physical settings for instream flows in Texas.” In this project, this qualitative regionalization and its boundaries were examined using quantified criteria. The State of Texas was partitioned into five regions: East Texas, South-Central Texas, Lower Rio Grande Basin, West Texas, and North-Central Texas via a series of qualitative parameters by 8-digit Hydrologic Unit Code (HUC) basin An analysis was made of the eighteen distinguishing parameters based on grouping by the original NRC Generalized Districts from the “Physical Settings for Instream Flows” description, then a methodology was devised to test the strength of the grouping as determined by the generalized districts from the NRC Report (2005) and to revise the groupings in a manner that would result in an improved stream classification. The process of applying the revision methodology resulted in quantitative improvements in the strength of the stream classification system over the original generalized districts. The changes in the median of the coefficients of variation for each 118 region ranged from a 20 percent improvement to a 3 percent reduction with a mean change of an 8 percent improvement. Based on the methodology presented herein, the results of the testing and revision processes were incorporated to produce an integrated stream classification system for Texas based on 18 distinguishing parameters encompassing watershed and stream channel processes and functions from four disciplines. This integrated stream classification system might be used to: (1) discern likely similarities and differences between rivers and streams of the State, (2) remotely characterize stream segments for which resources are insufficient for detailed field studies, (3) recognize streams and watersheds of the State as having common identities, (4) allow conclusions drawn from an instream flow study from a particular river reach to have a wider applicability than the particular study site, and (5) assist in prioritization of rivers and reaches for future instream flow studies. Moreover, the increased range of depth of data resources collected and incorporated into the integrated stream classification system could be of value to stakeholders and regulators. 119 APPENDIX A – SCOPE OF WORK The Texas Commission on Environmental Quality (TCEQ) has the statutory obligation to review water rights applications for their potential impacts on aquatic resources and set environmental flow requirements. The agency carries out this charge by relying on a hydrologic desktop method, publicly-available data, and site-specific information collected in the field to make environmental flow determinations. In addition, Senate Bill 2 passed by the 2001 Texas Legislature established for the first time the principle that a “sound ecological environment” in Texas streams and rivers should be protected by the establishment of scientifically-determined instream flow requirements. SB2 provided for a 10-year study period. Studies are currently underway on six priority river segments. In total, however, the segments being studied constitute only 8 of the [189] TCEQ-designated water quality management segments in Texas that are free flowing rivers (i.e., not inland water bodies, tidal reaches or coastal segments). Studies on the remaining [181] segments will take more time and resources than are available, and yet, it is important to characterize the other segments to determine appropriate instream flow requirements. This project would use GIS technology to organize existing information relevant to the understanding of Texas streams and rivers (i.e., water quality, geologic and geomorphic, hydrologic, and biological data), and develop a classification scheme such that particular classes or regions of streams and rivers could be recognized as having a common identity. Accordingly, conclusions drawn from instream flow studies in particular river reaches might have a wider applicability than the particular study site. The increased range and depth of data resources at the agency’s fingertips as it conducts environmental reviews of water rights applications would enhance the quality of those reviews. Moreover, this project could also help prioritize areas for future instream flow studies. A qualitative regionalization of Texas streams and rivers is presented in the National Research Council (2005) report and TCEQ proposes to examine this type of regionalization more closely and attempt to draw its boundaries using quantified criteria that will yield a stream classification system for the state. 120 Data will be managed in a spatially-explicit fashion, using the geo-referenced databases currently available and under development for the Instream Flows Program, and also building upon current research being performed within the Program. Boundaries drawn will be consistent with the prior delineation of river basins for use in TCEQ’s water availability models. Data incorporated into the classification structure may include: y Hydrology & Hydraulics: USGS discharge and stage data, climate data. y Biology: Texas Parks and Wildlife Inland Fisheries database, Index of Biologic Integrity metrics. y Geomorphology & Physical Processes: geology, land use/land cover, soils, channel cross-sectional form and size, channel plan form, bed morphology, and bed slope. y Water Quality: TCEQ Regulatory Activities and Compliance System (TRACS) database, including: temperature, dissolved oxygen, total suspended solids, and nutrients. 121 APPENDIX B – SUPPORTING DATA 122 123 124 125 126 127 REFERENCES 77 th Texas Legislature. (2001). “Senate Bill 2.” 80 th Texas Legislature. (2007). “Senate Bill 3.” Arthington, A.H., S.E. Bunn, N.L. Poff, and R.J. Naiman. (2006). “The challenge of providing environmental flow rules to sustain river ecosystems” Ecological Applications: 16(4) 1311–1318. Biggs B. J. F. (2007). Personal communication. Brierley G.J. and Fryirs K.A. (2005) Geomorphology and River Management: Applications of the River Styles Framework. Blackwell Publishing: Malden, Massachusetts. Brierley, G. J., and Fryirs, K. (2000). "River styles, a geomorphic approach to catchment characterization: Implications for river rehabilitation in Bega catchment, New South Wales, Australia." Environmental Management, 25(6), 661-679. Brierley, G., Fryirs, K., Outhet, D., and Massey, C. (2002). "Application of the River Styles framework as a basis for river management in New South Wales, Australia." Applied Geography, 22(1), 91-122. Chessman, B. C., Fryirs, K. A., and Brierley, G. J. (2006). "Linking geomorphic character, behaviour and condition to fluvial biodiversity: implications for river management." Aquatic Conservation-Marine and Freshwater Ecosystems, 16(3), 267- 288. Fitzhugh, T.W. (2005). “GIS Tools for Freshwater Biodiversity Conservation Planning.” Transactions in GIS, 9(2): 247-263. Fryirs, K. (2003). "Guiding principles for assessing geomorphic river condition: application of a framework in the Bega catchment, South Coast, New South Wales, Australia." Catena, 53(1), 17-52. Griffith G.E., Bryce S.A., Omernik J.M., Comstock J.A., Rogers A.C., Harrison B., Hatch S.L., and Bezanson D. (2004). Ecoregions of Texas (color poster with map, descriptive text, and photographs). U.S. Geological Survey: Reston, Virginia. Henriksen J. A., Heasley J., Kennen J.G., and Niewsand S. (2006). Users’ manual for the hydroecological integrity assessment process software (including the New Jersey Assessment Tools). U.S. Geological Survey, Biological Resources Discipline, Open File Report 2006-1093. 128 Higgins, J.V., Bryer, M.T., Khoury, M.L, Fitzhugh, T.W. (2005). “A Freshwater Classification Approach for Biodiversity Conservation Planning.” Conservation Biology, 19(2): 432-445. Horizon Systems, Inc. (2007) “National Hydrography Dataset Plus.” http://www.horizon-systems.com/nhdplus/ Irrigation Technology Center, Texas Water Resources Institute, Texas A&M University. (2005). “Texas Evapotranspiration Network.” http://texaset.tamu.edu/pet.php Jantzen, T.L. (2007). “Implementation of a State Hydrologic Information System.” Thesis, The University of Texas at Austin. Jobson, H. E.. (1996). Prediction of Travel Time and Longitudinal Dispersion in Rivers and Streams. U.S. Geological Survey Water Resources Investigations Report 96-4013. Kilroy C, Biggs B.J.F., and Vyerman W. (2007). “Rules for macroorganisms applied to microorganisms: patterns of endemism in benthic freshwater diatoms.” Oikos, 116, 550- 564. Kolbe, C.M. (2005). A Guide to Freshwater Ecology. Texas Commission on Environmental Quality, Austin, Texas. Kondolf, G., Montgomery D., Piégay H., and Schmitt L.. (2003). Geomorphic classification of rivers and streams. In Tools in Fluvial Geomorphology, eds. G. Kondolf and H. Piégay. Chichester, England: John Wiley & Sons. Maidment D.R. (2002). Arc Hydro: GIS for Water Resources. ESRI Press, Redlands, California. Maxwell, J.R., Edwards, C.J., Jensen, M.E., Paustian, S.J., Parrott, H., and Hill, D. M. (1995). A hierarchical framework of aquatic ecological units in North America (Nearctic Zone). General Technical Report NC-176. St. Paul, MN: U.S. Dept. of Agriculture, Forest Service, North Central Forest Experiment Station. McMahon G., Gregonis S.M., Waltman S.W., Omernik J.M., Thorson T.D., Freeouf J.A., Rorick A.H., and Keys J.E. (2001). “Developing a spatial framework of common ecological regions for the conterminous United States.” Environmental Management, 28(3), 293-316. Miller, D.A. and R.A. White. (1998). “A Conterminous United States Multi-Layer Soil Characteristics Data Set for Regional Climate and Hydrology Modeling.” Earth Interactions, 2. http://www.soilinfo.psu.edu/index.cgi?soil_data&conus 129 National Research Council Committee (2005). The Science of Instream Flows: A Review of the Texas Instream Flow Program. Committee on Review of Methods for Establishing Instream Flows for Texas Rivers, National Research Council. The National Academies Press, Washington, D.C. Olden, J.D., and Poff, N.L. (2003). Redundancy and the choice of hydrologic indices for characterizing streamflow regimes. River Research and Applications, 19, 101-121. Phillips, J.D. (2006). “Geomorphic Context, Constraints, and Change in the lower Brazos and Navasota Rivers, Texas.” Texas Water Development Board #2005-483-564. Poff, N.L. (1996). A hydrogeography of unregulated streams in the United States and an examination of scale-dependence in some hydrological descriptors. Freshwater Biology, 36, 71-91. PRISM Group, Oregon State University. (2006). “Parameter-elevation Regressions on Independent Slopes Model.” http://www.prismclimate.org Research Triangle Institute. (2001). The National Water Pollution Control Assessment Model (NWPCAM) Version 2 Draft Report. Research Triangle Park, NC. Rosgen, D.L. (1994). A Classification of Natural Rivers. Catena 22:169-199. Seaber, P.R., Kapinos, F.P., and Knapp, G.L. (1987). Hydrologic Unit Maps. U.S. Geological Survey Water Supply Paper 2294: U.S. Department of the Interior, Denver. Smith, R. (2006). Personal communication. Snelder, T. H., and Biggs, B. J. F. (2002). "Multiscale River Environment Classification for water resources management." Journal Of The American Water Resources Association, 38(5), 1225-1239. Snelder, T. H., Biggs, B. J. F., and Weatherhead, M. (2004). New Zealand River Environment Classification User Guide. National Institute of Water and Atmospheric Research (NIWA): Christchurch, New Zealand. Snelder, T. H., and Hughey, K. F. D. (2005). "The use of an ecologic classification to improve water resource planning in New Zealand." Environmental Management, 36(5), 741-756. Snelder, T. H., Biggs, B. J. F., and Woods, R. A. (2005). "Improved eco-hydrological classification of rivers." River Research and Applications, 21(6), 609-628. Snelder, T. H., Cattaneo, F., Suren, A. M., and Biggs, B. J. F. (2004). "Is the River 130 Environment Classification an improved landscape-scale classification of rivers?" Journal of The North American Benthological Society, 23(3), 580-598. Texas Commission on Environmental Quality. (2007). “TCEQ Regulatory Activities and Compliance Systems Surface Water Quality Monitoring system.” http://www.tceq.state.tx.us/compliance/monitoring/crp/data/samplequery.html Texas Instream Flow Program. (2002). Texas Instream Flow Studies: Programmatic Work Plan. Texas Instream Flow Program. (2006). Draft Texas Instream Flow Studies: Technical Overview. Thomson, J. R., Taylor, M. P., and Brierley, G. J. (2004). "Are River Styles ecologically meaningful? A test of the ecological significance of a geomorphic river characterization scheme." Aquatic Conservation-Marine And Freshwater Ecosystems, 14(1), 25-48. Thomson, J. R., Taylor, M. P., Fryirs, K. A., and Brierley, G. J. (2001). "A geomorphological framework for river characterization and habitat assessment." Aquatic Conservation-Marine And Freshwater Ecosystems, 11(5), 373-389. United States Bureau of Reclamation. (2004). “BFI: A Computer Program for Determining an Index to Base Flow.” Department of the Interior. United States Environmental Protection Agency. (2006). Biological Indicators of Watershed Health. http://www.epa.gov/bioiweb1/html/indicator.html United States Environmental Protection Agency and United States Geological Survey. (2006). “NHDPlus User’s Guide.” http://www.horizon-systems.com/nhdplus/data/NHDPLUS_UserGuide.pdf United States Geological Survey. (2004). The National Geochemical Survey - database and documentation. U.S. Geological Survey Open-File Report 2004-1001, U.S. Geological Survey, Reston, Virginia. http://tin.er.usgs.gov/geochem/ Vogel, R.M., Wilson, I.W., and Daly, C. (1999). “Regional Regression Models of Annual Streamflow for the United States.” Journal of Irrigation and Drainage Engineering, 125(3):148-157. Zeiler, M. (1999). Modeling Our World: The ESRI Guide to Geodatabase Design. ESRI Press, Redlands, California.