Estimation of groundwater depth based on precipitation data using geostatistical methods

Successive drought events as one of the most important environmental crises, along with population growth and uncontrolled water extraction, have led to an increase in the depth of groundwater, especially in arid and semi-arid regions. The purpose of this study is to investigate the geostatistical relationship between groundwater depth data and groundwater depth based on precipitation data in three observation wells located in Fars province. These wells were selected based on the PSO clustering technique. Thus, the three observation wells that were closest to the center of the calculated clusters were selected as the representative of the clusters. These wells are located in Karsia, Dolatabad, and Fatehabad regions for clusters 1 to 3, respectively. Monthly groundwater depth data has been used from 2003 to 2017. Kriging and cokriging methods were performed in the GS+ environment. In this research, precipitation data was used as an auxiliary variable. Furthermore, the models were selected based on the lowest RSS values ​​and the nearest R2 values, and the spatial structure ratio (C / C0 + C) to one. Accordingly, the selected models for the main variable (groundwater depth) in the first to third clusters are spherical, power, and linear, respectively, and for cross-variogram models (precipitation - groundwater depth) are all spherical. The results showed that in the validation and test stage, the Cokriging method has higher accuracy than the Kriging method. The test stage in kriging and cokriging methods for RMSE index are (0.92 and 0.41), (0.54 and 0.52), and (1.25 and 0.95) in 1, 2, and 3 clusters, respectively.