Impact of Climate on Water Availability for the Edwards Aquifer
San Antonio, Texas
Progress Report--October 20, 1997

by Kris Martinez

This report outlines the work currently being done in conjunction with the study to evaluate the impacts of climate on water resources in the Edwards Aquifer. The study aims to develop better relationships for predicting water levels and springflows based on climate conditions. After reviewing three different programs that have been used to simulate aquifer dynamics, the groundwater simulation model developed by Watkins (1996) has been chosen for use in the project. The model uses empirically derived functions to predict recharge to the aquifer based on streamflows and initial water levels. These recharge functions will be modified to account for climatological factors like precipitation. Historical rainfall and temperature data have been obtained. In addition, a Geographic Information Systems (GIS) representation of the Edwards Aquifer is being developed to illustrate climate effects on water availability in the region.

Model Review

The primary focus of my work to date has been reviewing models developed by other researchers to simulate the behavior of the Edwards Aquifer. These models have been calibrated so that predicted water levels and springflows match historical values over an extended period of time. At this stage of the project, the goal was to develop an understanding of the concepts that formed the basis of the FORTRAN code for each of the models. The three models reviewed were as follows:

• Lumped Parameter Model (Wanakule and Anaya)
• Groundwater Simulation Model (Watkins)
• Decision Support System (Hammond et al.)

where R is recharge, P is pumpage, S is storativity, T is transmissivity, and h is the water level in a given tank.

Piezometric heads are given for each of the tanks at the start of the simulation. Pumpage was estimated using monthly pumping data from 1978-1989. A distinctive feature of this model is that recharge is not constant. Recharge functions were developed from flow loss analysis of the drainage basins. The recharge is dependent on streamflow and water levels. Another major difference is that the lumped parameter model uses linear algebra to perform the integration and solve the equation analytically. Paynter’s algorithm is used to estimate the exponential terms.

Groundwater Simulation Model. The groundwater model developed by David Watkins is adapted from the lumped parameter model. It operates under the same fundamental relationships. However, this model solves the aquifer flow equations using a form of finite-difference method based on Gauss-Seidel iteration and Successive Over Relaxation (SOR). In order to better understand the modeling techniques that were employed in this program, I read through sections of Introduction to Groundwater Modeling (Wang and Anderson 1982). SOR is an effective tool for extrapolating water levels and arriving at a solution more rapidly. A graph of observed water levels for the Nueces River recharge basin and water levels predicted by the groundwater simulation model is provided as Attachment 1.

Decision Support System. The University of Texas at San Antonio (UTSA) Decision Support System uses a finite-difference model called GWSIM-IV and includes pre- and post-processors which allow the user to select different input scenarios and display the output graphically. The model has 300 cells and requires approximately one hour to complete a run for a 20-year simulation period. When a water conservation function was introduced into the model, run times increased tenfold. The program is currently used by the San Antonio Water System and Guadalupe Blanco River Authority.

Model Selection

After reviewing the three programs, the groundwater simulation model by David Watkins appears to be the most appropriate for this project. As previously mentioned, the focus of this study is to gain a better understanding of how changes in climate effect the availability of water from the Edwards Aquifer. Rainfall-runoff relationships will be developed to evaluate the connection between precipitation and recharge. These relationships are not easily incorporated into finite-difference models with a large number of cells, like GWSIM-IV. Additional functions increase run times significantly. Reprogramming the lumped parameter model to modify the recharge functions would be a difficult task. The model uses matrix algebra and the elements are referenced with pointers instead of arrays in MacFORTRAN. Consequently, relationships are not readily apparent and would require a great deal of time to elucidate. The groundwater simulation model simulates the Edwards Aquifer with only nine elements and the recharge functions can be easily modified. This model will be used for the study because it provides the most efficient means of including the necessary rainfall-runoff relationships.

Climatological Effects

Edwards Aquifer simulation models have typically used fixed recharge values as an input to a program. However, the groundwater simulation model uses recharge functions that are based on flow loss analyses performed by Wanakule and Anaya. The functions determine recharge to a tank as a function of upper gage flows and, in some cases, water level in the tank. Climatological factors, such as rainfall, are not considered. In order to evaluate the impact of climate changes on the aquifer, the functions will need to be modified or replaced to account for the climate.

The USGS uses another method to calculate recharge to the Edwards Aquifer. It presumably accounts for rainfall by calculating runoff based partially on individual storms in the area above a drainage basin’s upper gage. The relationships are described in the report, Method of Estimating Natural Recharge to the Edwards Aquifer in the San Antonio Area, Texas (Puente 1978). The general form of the equation is,

where Q_U is the volume of flow at a basin’s upper gage, Q_L is the volume of flow at the lower gage, and SI is given by,

where delta A is the drainage area between a basin’s upper and lower gage, A_u is the drainage area above the upper gage, and Q_tu is the volume of water contributed by a storm above the upper gage. Storm water contributions are obtained by analyzing the streamflow hydrograph for the upper gage.

For the purposes of this research, it may be beneficial to develop rainfall-runoff relationships to predict the value of SI (and possibly Q_U and Q_L) under a given set of climate conditions. The development of such equations would invariably lead to a better understanding of climatological impacts on water availability. HDR Engineering, Inc. (Choffel and Vaugh 1993) has developed a modified version of the Soil Conservation Service (SCS) runoff curve number method. In this method, Thiessen polygons are used to account for spatial variation in precipitation. This method warrants further investigation, although it has not been thoroughly reviewed at this time.

Climatological Data

A gridded data set of historical precipitation and temperature in the Edwards Aquifer region has been obtained. The data were provided by Nan Rosenblum and were compiled as part of the Vegetation/Ecosystem Modeling and Analysis Project (VEMAP). These data correspond to points on a 0.5 x 0.5 degree latitude and longitude grid. The grid covers an area from 29 to 32 degrees north latitude and 97 to 101 degrees west longitude. Climate data covering the years 1895-1993 are available for 63 grid points. The precipitation and temperature records are still being checked for accuracy and are considered preliminary. However, they provide a good starting point for developing methods to process geographically referenced climate data for use in rainfall-runoff analysis. A graph of precipitation over time at a location within the Nueces River recharge basin is provided as Attachment 2.

Geographic Information Systems (GIS) Component

Using a digital elevation model (DEM), the watersheds in the Edwards Aquifer region were delineated to develop a dynamic base map that will interface with the groundwater simulation model. The watersheds conform very closely to the sub-basin boundaries identified by Puente and also by Wanakule and Anaya. These sub-basins are conceptualized as tanks in the model. My current plan is to run the FORTRAN program separately and use Avenue scripts to retrieve the predicted water levels and springflows into attribute tables. The user will be stepped through time in order to illustrate the behavior of the aquifer in response to climate conditions. The base map for the study, which is still under development, is included at the top of the report.

Attachment 1

Attachment 2