Modeling the Effects of Consumption and Precipitation on the Water Table Oscillations (Case Study: Ajabshir Aquifer)

Prediction of groundwater table through specific models, provide a valuable information for water resources management and consumption planning. Among the different available methods, the multivariate nonlinear regression is of utmost importance for prediction of hydrological phenomena. The data bases of this research were the amounts of precipitation amounts, water table elevation and water consumptions in monthly time scale for the period of 2001- 2011. The cross correlation analysis indicated that the one lagged monthly precipitation as well as the two lagged monthly consumptions values had the highest impacts on water table elevation with the determination coefficients of 0.39 and 0.86, respectively. Then, the general relationship of these three variables obtained with 2=0.87 and root mean square error (RMSE) =0.35m through a multivariate nonlinear regression analysis. For prediction of the water table elevation in the coming years, initially, the consumptions and precipitation data extended up to 2014, using ANN and Thomas-firing methods, respectively. So, the outcome of putting them into regression equation gave the water table elevation. On the other hand, the artificial neural network was used to predict the water table elevation, for which, the resulted values of R2 and RMSE were 0.82 and 0.39m, respectively. The comparison of two methods showed that the multivariate nonlinear regression model represented more accurate results in predicting the elevation of water table, in the studied plain.