Artificial Neural Network Technique For Predicting of Groundwater Level In Sarir Wellfield – GMMRP, SE Libya.

Abstract

In this study, an application of Artificial Neural Networks technique (ANN) was presented to predict the confined aquifer water level (SWL) located in Sarir Wellfield, SE-Libya. Subsequent to instating initializing the model with (SWL) observed, the developed artificial neural networks back propagation model (BP-ANN) should be able to reproduce (SWL)using input variables, including aquifer transmissivity (T m²/min),and wells location coordinate (E m,N m, Z m). The performance of ANN models was evaluated using the mean absolute percentage error (MAPE%) and the efficiency factor (E). The results indicated that, ANN technique was well suited for predicting the confined aquifer water levels at Sarirwellfield. According to the results observation, the (ANN3) model optimized with the Levenberg-Marquardt(LM) algorithms showed the most beneficial results with the minimum (MAPE%) value of (0.055) and maximum (E) value of (0.98), obtained for simulation of groundwater levels. The present research conclusively showed the capability of ANN to provide excellent estimation accuracy.

Key words: Confined Aquifer, BP-ANN, Groundwater level, Sarir Wellfield.