Table of Contents

• Introduction

• Project Overview

• GIS Data Preparation

• Non-Point Bacterial Loading Calculations

• Point Source Bacterial Loading Calculations

• Schematic Processor

• Water Quality Model in Model Builder

• Downloads

• Future Work

• Appendix A: Terrain Preprocessing

• Appendix B: WRAP Hydro Process

• Appendix C: Precipitation Rasters for Land Use Classifications

• Appendix D: Schematic Network

• Appendix E: Process Schematic

• Appendix F: Travel Time Calculations

• Works Cited

• Contact Information

Introduction

Section 303(d) of the 1972 Federal Clean Water Act (CWA) requires that each State identify water bodies that do not meet the State’s water quality standards and create a priority ranking of the impaired waters based on the severity of pollution and the water body’s intended use.

For the State of Texas, the Texas Commission of Environmental Quality (TCEQ) has identified water bodies that do not meet the Texas Surface Water Quality Standards.  Three water bodies that are identified on Texas’ 303(d) list of impaired waters are Copano/Port/Mission Bay, Mission River, and Aransas River, which are all located in the Copano Bay watershed near Texas’ southeastern coastline (shown in Figure 1).  All three water segments do not meet water quality standards due to high bacteria levels. 

Project Overview

In order to meet the State’s water quality (bacterial) standards, the major bacterial sources in the Copano Bay watershed need to be identified; the total bacterial loadings, Total Maximum Daily Loads (TMDLs), from these sources need to be calculated, and the amount of load reductions that is needed to attain water quality standards needs to be estimated.

Arc Hydro tools have been integrated with hydrologic processing engines to create a water quality model to calculate the total point and non-point source bacterial loadings to the Copano Bay watershed.  Using Model Builder in Arc Toolbox, an Arc GIS 9.0 model has been created to determine and accumulate the watershed bacterial loads for each of the impaired water body segments.  Processing engines have also been developed (using dynamic linked libraries or DLLs) to simulate bacterial load transport.

Geographic Information Systems (GIS) Data Preparation

Before calculating the bacterial loadings, watersheds need to be delineated because the bacterial loading per watershed is needed for the Arc GIS 9.0 model.  Watershed delineation requires a Digital Elevation Model (DEM), river network, and Critical Points, which is a feature class that contains points where the fecal coliform concentration can be examined.  Critical Points were determined to be USGS gauge stations, bacteria monitoring stations (so modeled values can be compared to existing monitoring data), and water segment endpoints. 

The DEM was obtained from the National Elevation Dataset (NED) from USGS, which provides seamless coverage of the United States, providing a 1:24,000-scale Digital Elevation Model (shown in Figure 2).  The river network was obtained from the National Hydrography Dataset (NHD), which provides digital spatial data and is based on the USGS Digital Line Graph (DLG) hydrograph data and on information from the Environmental Protection Agency (EPA) Reach File Version 3 (RF3), which is also shown in Figure 2.    The USGS gauge station locations were obtained from BASINS (Better Assessment Science Integrating Point and Non-point Sources), which is an environmental analysis system developed by the U.S. EPA that can be used to perform watershed and water quality studies (shown in Figure 2).  Annual statistics of the flow measured at the gauges found in this watershed were obtained from USGS.  The data and locations of the bacteria monitoring stations and water segment endpoints (shown in Figure 2) were provided by Sandra Alvarado, Natural Resource Specialist for TCEQ, who compiled bacteria data for the three water segments from the TCEQ Texas Regulatory and Compliance System (TRACS) database from the past five years, which includes Texas Department of Health (TDH) data. 

Terrain Preprocessing (found in the Arc Hydro toolbar) needs to be implemented on the DEM (Digital Elevation Model) to determine the flow patterns in the basin (flow direction), and subwatersheds need to be delineated for the Copano Bay watershed based on the Critical Points (USGS gauge stations, water segment endpoints, and bacteria monitoring stations).  After Terrain Preprocessing is complete (procedure found in Appendix A), Water Rights Analysis Package (WRAP) Hydro is used to delineate watersheds for the basin (procedure described in Appendix B).  See Figure 3 for the delineated watersheds.

Non-Point Source Bacterial Loading Calculations

Non-point source bacterial loading calculations were made based on the basic relationship:

L = Q * C

Where:    L = Bacterial Loading (cfu/year), cfu is the number of microbiological organisms (coliform forming units)

               Q = Runoff (m³/year)

               C = Concentration (cfu/m³)

Thus, GIS data layers need to be compiled and prepared in order to calculate Runoff (Q) and Concentration (C). 

Runoff Calculations

Runoff calculations were made by using the empirical equations derived from Ann Quenzer’s thesis, A GIS Assessment of the Total Loads and Water Quality in the Corpus Christi Bay System [1].  These runoff equations were derived by using the Microsoft Excel 5.0 Regressions Tool, which was used to base the equations on a relationship between streamflow depth, precipitation depth and percent land use in each of the nine watersheds in the Corpus Christi Bay system, which includes the Copano Bay watershed.  The annual precipitation data was obtained from the National Resources Conservation Service (NRCS) and the Spatial Climate Analysis Service (SCAS) at Oregon State University (OSU), which is shown in Figure 4.  OSU developed PRISM (Parameter-elevation Regressions on Independent Slopes Model), which gives the average annual precipitation in inches from 1961-1990.  The land use land cover dataset was obtained from the United States Geologic Survey (USGS) and was developed from national land cover data from Multi-Resolution Land Characterization (MRLC) data called National Land Cover Data 1992 (NLCD 92) and is based on 30-meter Thematic Mapper data.  These are the following equations that were used to predict runoff based on land use classifications.

Agriculture:

Q = 0.008312 * exp ( 0.011415 * P )

Forest:

Q = 0.0053 * exp ( 0.010993 * P )

Urban:

Q = 0.24 * P

Open Water:

Q = 0

 Where:    Q = Runoff (mm/year)

                P = Precipitation (mm/year) – from PRISM

The precipitation data was obtained from PRISM in polygon feature class format.  Using the “Feature to Raster” tool in Arc Toolbox, the polygon feature class was converted to a raster based on the field, “RANGE”, which is the annual precipitation in inches.  The annual precipitation was then converted to millimeters by using Spatial Analyst's Raster Calculator: [Precipitation in in/year] * (25.4 mm/inch) = [Precipitation in mm/year] = P, where [] represents a raster.

Precipitation rasters were then created for each land use classification (Agriculture, Forest, Urban, and Open water).  The procedure for how this was done is described in Appendix C.  Once the Precipitation rasters, P, for each land use are created, Raster Calculator can be used to calculate the runoff for each land use.  Figure 5 shows an example of how the Runoff raster for Agriculture is calculated based on Ann Quenzer’s runoff equations: 

Once the four Runoff rasters are created for each land use, the “Mosaic” tool in Arc Toolbox was used to combine all four rasters into a single Runoff raster.

Once the total Runoff raster is created (Q_total in mm/year), the Raster Calculator needs to be used to convert the runoff into m³/year.  Because the raster contains 30m by 30m grid cells, a conversion factor of 0.9 needs to be used. [Runoff in mm/year] * 0.9 = [Runoff in m³/year].

Event Mean Concentratin (EMC) Calculations

GIS data layers that were used to calculate the bacteria concentration in the Copano Bay watershed.  The land use land cover dataset was obtained from USGS, and the EMC values (fecal coliform concentrations) can be approximated for each type of land use.  For this project, fecal coliform EMCs for each land use code were determined by Reem Jihan Zoun in her thesis, Estimation of Fecal Coliform Loadings to Galveston Bay (2003) [2], which are listed in Table 1 below.  However, fecal coliform EMC values for land use classifications 51 (Shrubland), 71 (Grasslands/Herbaceous), and 81 (Pasture/Hay) were modified to zero in order to not account twice for the bacterial waste produced from livestock animals.  Fecal coliform bacterial loadings from livestock are accounted for in Point Source Bacterial Loading Calculations.

                                                           Table 1. Fecal Coliform EMC values based on Land Use Classifications [1]

 The Source Code descriptions for EMC values in Table 1, also found in Zoun’s thesis [2], are given in Table 2.

                                                           Table 2. Description of Source Code for EMC values

The Event Mean Concentration (fecal coliform concentration) raster is created by using the relationship between fecal coliform concentrations and land use found in Table 1.  Once the EMC table (Table 1) is joined to the land use land cover feature class based on the land use codes (found in both EMC Table and the Land Use Land Cover Polygon Feature Class), a raster can be created based on the EMC field (now in the land use land cover feature class) using the “Feature to Raster” tool in Arc Toolbox.  Raster Calculator then needs to be used to convert cfu/100mL to cfu/m³. [cfu/100mL] * 10,000 = [cfu/m³].

Non-Point Source Bacterial Loading Calculations

The bacteria load per grid cell can now be calculated by using Q and C rasters created from Runoff Calculations and EMC Calculations: L = Q * C, using Raster Calculator.

Using Zonal Statistics and the delineated watersheds (Figure 3), the cumulative runoff and bacterial loadings can be calculated for each watershed.  Figure 6 shows the results. 

Point Source Bacterial Loading Calculations

The point sources that may contribute to fecal coliform bacteria contamination to the Copano Bay watershed are livestock, wastewater treatment plants (WWTPs), colonial waterbirds, leaking septic systems, confined animal feedlot operations (CAFOs), and so forth.

Concentrated Animal Feedlot Operations (CAFOs)

A shapefile that contains the two permitted CAFOs within the Copano Bay watershed was obtained from the TCEQ.  However, one of the permitted facilities is actually located within the Nueces River Basin and drains to Lake Corpus Christi and not Copano Bay.  The other permitted facility is in the Copano Bay watershed, but the permit was not recently renewed because the company is no longer operating (Personal Communication with CAFO Permitting Team Leader, Beth Fraser.)  Thus, there are no CAFOs within the Copano Bay watershed at this time.

Livestock

Livestock data (annual count per county) were obtained from the 2002 Census of Agriculture, National Agricultural Statistics Service (NASS), and the 2004 Texas Livestock Inventory and Production, United States Department of Agriculture (USDA), NASS, Texas Statistical Office.  The animals that were considered in the calculations were cattle, goats, horses, sheep, hen, hogs, and chickens.  For example, in 2002, there were 22,253 cattle in San Patricio County.

The density of livestock per county (count/m²) were calculated for each animal by using the following equation: [Total annual count of each animal] / [Area in m² where the animals would be located within county].  The area where animals would be located was assumed to be from the land use land cover values 51 (Shrubland), 71 (Grasslands/Herbaceous), and 81 (Pasture/Hay).  To find the area, the land use land cover dataset needs to be masked by each county, and the corresponding grid cells need to be summed.  For example, in San Patricio County (calculation shown in Figure 7), the total area where animals are located is 177,403,500 m².  Thus, the density of cattle in San Patricio county is 22,253 cattle/177,403,500 m² = 0.000125 cattle/m².

The area (m²) of each county within each delineated watershed (Figure 3) was determined and then multiplied by each livestock's density in each corresponding county.  For example, Watershed JunctionID 45422 has two counties overlapping it, so there are two different areas and cattle densities to account for.  The calculation of how many cows are in Watershed JunctionID 45422 is shown in Figure 8.  Thus, there are approximately 14,366 cattle in Watershed JunctionID 45422.

After determining the count of each animal within each watershed, the count is multiplied by the cfu/year produced by each animal.  Table 3 shows the cfu/year produced by each animal that is considered in this bacterial loading model.

Table 3. Annual Fecal Coliform Bacterial Loading (cfu/year) for Livestock Animals

For example, in Watershed JunctionID 45422: 14,366 cows * (1.97 x 10¹² cfu/year cow) = 2.78 x 10¹⁶ cfu/year from cattle.  The cfu/year needs to be summed for all species within each watershed.  The total cfu/year for each watershed then needs to be added to the non-point bacteria loading calculations to obtain the total cfu/year in each watershed as shown in Figure 9.

Wastewater Treatment Plants (WWTPs)

The locations of WWTPs were obtained from the Permitted Wastewater Outfalls shapefile provided by the TCEQ.  Descriptions of the permitted facilities were obtained from Sandra Alvarado from the TCEQ TRACS database.  The average cfu/year from the WWTP monitoring data were determined and compared to cfu/year values from literature.  However, WWTPs must be in compliance with surface water quality standards and disinfect the bacteria in the wastewater effluent.  It is assumed that the fecal coliform bacteria in the effluent is disinfected before being discharged into the surface waters.

Septic Systems

The number of septic systems per county was obtained from the U.S. Census of Bureau 2000 for the Copano Bay watershed, but the exact locations were not determined.  However, this data does not include possible leakages in the septic systems.  

The Comprehensive Sanitary Survey of the Shellfish Producing Waters of Copano Bay [3] gives approximate locations of septic systems around the Copano Bay area and reports minimal leakage from the septic systems surrounding the Bay.

The work of Dr. Joanna Mott, Texas A&M - Corpus Christi, which shall be complete by the end of July, will identify the possible sources that the bacteria is coming from (using Bacteria DNA fingerprinting; also known as bacteria source tracking).  Pending on the results of Dr. Mott's study, the density of septic systems may be significant (if the bacteria contamination is primarily from human feces.)  This would indicate that there may be possible leakage in septic systems where fecal coliform bacteria contamination is the greatest.

Thus, septic system leakages are not going to be accounted for in the bacteria loading model (due to lack of leakage data), but the septic system concentration within each county may be critical if there are no other sources in the area that may account for the high fecal coliform concentrations.

Schematic Processor

Once the bacteria loadings are calculated per watershed (accounting for both point and non-point bacterial loadings), the transport of bacteria from the watersheds to Copano Bay needs to be modeled.  In order to simulate bacterial load transport, “Process Schematic” (a script tool that was developed by Jon Goodall and Tim Whiteaker in 2003) implements dynamic linked libraries (DLLs).  The two processing engines (DLLs) that were used in this bacteria watershed model were clsDecay.dll, which accounts for first-order decay of bacteria along water segments, and clsCFSTR.dll, which calculates the increase in bacteria concentration in Copano Bay due to bacteria loadings from the upstream watersheds.  Goodall and Whiteaker submitted a journal article describing the Schematic Processor and Schematic Network in more detail for publication in Transactions in GIS called  "Integrating Arc Hydro Features with a Schematic Network".  The journal article can be viewed here.

Schematic Network

Before the “Process Schematic” tool can be implemented, a Schematic Network of the Copano Bay watershed needs to be created, and the procedure is described in detail in Appendix D.  The Schematic Network is made up of two feature classes: SchemaNode and SchemaLink.  SchemaNode represents the nodes in the watershed (a watershed, drainage point, or Copano Bay.)  SchemaLinks connect the Schematic Nodes and is a way to model what happens to the bacteria as it travels to Copano Bay.  Figure 10 shows the Schematic Network that was created for the Copano Bay watershed and includes the necessary parameters (inputs) necessary to run the model that are explained in more detail in Appendix D and below in Dynamic Linked Libraries (DLLs).

Dynamic Linked Libraries (DLLs)

clsDecay.dll simulates decay of bacteria along stream segments (shown in Figure 11) and assumes first-order decay by the following equation:

 load_passed = load_received * e^-kt

 Where:    k = first-order decay coefficient (day^-1), which is stored as an attribute in SchemaLink

                t = travel time (residence time) along streams, t (days), which is stored as an attribute in SchemaLink

clsCFSTR.dll calculates the increase in concentration of a bay due to bacteria loadings.    The bay is assumed to be completely mixed and acts as Continuous Flow, Stirred Tank Reactor (CFSTR), and the inflow into the bay equals the outflow.  The following equation calculates the concentration in the bay:

 c = L / (Q + kV) 

Where: c = concentration in bay (cfu/m³)

            L = bacteria load entering bay (cfu/year)

            Q = total flow (m³/year), which is stored as an attribute in SchemaNode

            k = first-order decay coefficient (day^-1), which is stored as an attribute in SchemaNode

            V = Volume of bay (m³), which is stored as an attribute in SchemaNode                                            

Once the Schematic Network is created and the parameter values are input to the corresponding fields in the attribute tables of SchemaNode and SchemaLink, “Process Schematic” can be run (see Appendix E).

Water Quality Model in Model Builder

All the processes previously described were also modeled in Model Builder (non-point bacterial loading calculations) as seen below in Figure 12.