A new approach in using the GRACE satellite data and artificial intelligence models for modeling and predicting the groundwater level (case study: Aspas aquifer in Southern Iran)

Industrial and agricultural development, population increase, limitations in water resources renewability, lack of timely management of water resources, and the recent years' droughts have caused pressure on groundwater. One of the aquifers that have faced a sharp drop in water level in recent years is the Aspas aquifer in Fars province. In this study, the condition of the groundwater level (GWL) in this aquifer was analyzed using the data of the gravity recovery and climate experiment (GRACE) Satellite. In addition, pre-processing tools, such as complementary ensemble empirical mode and decomposition (CEEMD) and wavelet transform (WT), were utilized. The support vector regression (SVR) and artificial neural networks (ANN) models were used in two simple and hybrid ways with pre-processing tools. According to the results, combining the models with pre-processing tools has improved their efficiency. As a result, the coefficient of determination (R2) has been improved from 0.927 in ANN to 0.938 in W-ANN and 0.998 in CEEMD-ANN. The R2 has reached from 0.918 in the SVR to 0.949 in the W-SVR and 0.948 in the CEEMD-SVR. The comparison between the results of processing algorithms of GRACE satellite in the test phase determined that the GFZ processing algorithm shows the best performance. CEEMD-ANN performance was compared to GFZ algorithm. In addition, a new approach was utilized to forecast the GWL shifts. The results indicated that the new approach provides a suitable estimate of the groundwater in the shortest time with the lowest cost. Therefore, this approach can be used to predict the GWL in other aquifers.