Efficacy of Genetic Algorithm to Investigate Small Scale Aquitard Leakage

Abstract

The movement of water from a shallow aquifer to a semiconfined aquifer through a localized area of leakage in an aquitard, designated as an aquitard window, is an area of concern in certain gulf coastal plain aquifer settings. Locating these windows, or areas of leakage, has been historically arduous. The purpose of this study was to evaluate the use of a genetic algorithm (GA) technique as an inverse technique for locating highly probable areas of leakage based upon aquifer head data near the suspected area of leakage. The GA technique has previously been tested on synthetic data alone, and this research offers insight into the capabilities of the technique with actual field data from a site with a known aquitard window. Research sites with historical and ongoing data collection efforts are rare and offer the best opportunity for robust evaluation of new methodologies. Based upon the GA technique, the steady-state accretion flux to the Memphis aquifer at the Shelby Farms study site was determined to be 12,346 m3/day with a standard deviation of 821 m3/day and compared well with the range from 510 to 9,911 m3/day previously determined for the site.