Predicting the Groundwater Salinity under Drain Pipes Using Artificial Neural Network

Awareness of salinity of soil layers under drains, particularly in areas with shallow saline groundwater such as Khozestan leads to design the best depth and spacing drain. In this study the application of artificial neural network modeling to predicting of changes in groundwater salinity under drain pipes have been tested. In order to calibrate and validate the model results, data collected from experimental model with 1.8 m long, 1 m wide and 1.2 m high were used. In the model, drains were installed at 20, 40 and 60 cm depths and spacing of 60, 90 and 180 cm. In the method of artificial neural network, LevenbergMarquardt learning algorithm with SigmoidAxon transfer function was used. After statistical analysis and calculation of RMSE, the standard error and correlation coefficient, adjustment between measured and simulated values of changes in groundwater salinity was calculated. The value of these product indexes 5.27 ds/m, 0.12 and 0.96 was estimated respectively. Changes in drains salinity in different depths and spaces over time with discharges of 0.07, 0.11 and 0.14 lit/s are 0.34 dS/m, 0.09 and 0.99, respectively. The results showed that artificial neural network method on simulating of changes in groundwater salinity under drain pipes and also changes in drain water salinity in difference depths and spaces of drains have reasonable accuracy.