Groundwater level simulation using artificial neural network: a case study from Aghili plain, urban area of Gotvand, south-west Iran

Abstract

In this paper, the Artificial Neural Network (ANN) approach is applied for forecasting groundwater level fluctuation in Aghili plain,<br />southwest Iran. An optimal design is completed for the two hidden layers with four different algorithms: gradient descent with<br />momentum (GDM), levenberg marquardt (LM), resilient back propagation (RP), and scaled conjugate gradient (SCG). Rain,<br />evaporation, relative humidity, temperature (maximum and minimum), discharge of irrigation canal, and groundwater recharge from<br />the plain boundary were used in input layer while future groundwater level was used as output layer. Before training, the available data<br />were divided into three groups, according to hydrogeological characteristics of different parts of the plain surrounding, each<br />piezometer. Therefore, FFN-LM algorithm has shown best result in the present study for all three hydrogeological groups. At last, to<br />evaluate applied division, a unit network with all data and using LM algorithm was trained. Validation of the network shows that<br />dividing the piezometers into different groups of data and designing distinct networks gives more focus on simulating groundwater<br />level in the plain. The degree of accuracy of the ANN model in prediction is acceptable. Thus, it can be determined that ANN provides<br />a feasible method in predicting groundwater level in Aghili plain.