UC Santa Barbara

Accurate estimates of aquifer parameters are necessary for effective groundwater management and for geotechnical engineering applications. Pumping tests may be employed to estimate the hydraulic conductivity in leaky aquifer/aquitard systems. This work introduces a hybrid algorithm with global search capacity (the Genetic algorithm, GA) and local search capacity—the Levenberg-Marquardt (LM) algorithm—coupled with a modified Neuman-Witherspoon solution for leaky aquifers to estimate the aquifer’s hydraulic parameters from pumping-test data. The GA is employed to determine the initial guesses of the aquifer parameter values. The optimal parameter values are then obtained with the LM algorithm, yielding a mixed GA/LM algorithm, herein named GALMA. Results show that the drawdown trends based on the estimated parameters agree well with measured drawdown. The proposed estimation algorithm identifies aquifer parameters with greater reliability than previous approaches. Verification of the GALMA is carried out based on three pumping tests in a layered aquifer in Tianjin, China, and on four historical case studies involving diverse hydrogeological settings. The excellent match between observed drawdown and GALMA-estimated parameters demonstrates the estimation accuracy and superior performance relative to previously reported estimation methods.