A Soft Computing Approach for Estimating Natural Recharge Levels in the Semi-Arid Edwards Aquifer Region of Texas

Changes in land cover and land use, as well as shifts over time in ecological parameters such as precipitation and temperature, may influence the hydrologic response of an aquifer system. For water planning and management purposes, this is of unique interest in semi-arid regions where water availability is often low and thus, water balance sensitivity may be high. The semi-arid Edwards Aquifer region of Texas has undergone measurable increases in both population and impervious surface area over the last twenty years, particularly in the greater metropolitan areas of San Antonio and Austin, the eighth and nineteenth largest cities in the United States, respectively. Consequently, it is expected that the hydrologic response of the Edwards Aquifer has also undergone changes. Understanding historical land cover and land use, and other related changes within the various ecological signals of this region is advantageous when hydrologic modeling efforts are considered; however, quantifying these different types of changes in ecological signals for prediction is not a well understood task. The inherent non-stationary and non-linear nature of hydrologic signals makes a complete analysis by traditional methods very difficult or often impossible to successfully perform. Soft computing techniques such as artificial neural networks and fuzzy logic provide an alternative means of forecasting the response of a system when the inputs are not well understood. This work presents the development of a neuro-fuzzy model for estimating natural recharge levels in the semi-arid Edwards Aquifer region of Texas when land cover and land use, and other related system input changes are considered.