Use of Back-propagation Artificial Neural Networks for Groundwater Level Simulation

Abstract

This article presents simulation of groundwater level fluctuation based on an artificial neural network modelling. The prediction used multi-layer back-propagation neural networks (BPANN). The case of study area was Jakarta, Indonesia, that has high population density and several purposes of groundwater resource usage. Input variables were using delay five-daily groundwater level fluctuation (GLF) of observation well interest to predict current GLF. The applicability of BPANN for GLF prediction was verified in three sets of input variables. The result showed that application of BPANN to simulate GLF gives satisfied prediction results.