Optimization of sampling wells by a spatial-temporal approach in the groundwater level monitoring network (Case study: Sarab plain)

Optimizing the piezometric wells in the groundwater monitoring network is important in terms of reducing maintenance costs and improving efficiency and increasing the speed of data updating. Statistical methods or PCA are commonly used to identify significant wells. In geostatistical methods, optimization is performed according to the location of the samples but the temporal information of the wells is not taken into account. In the PCA method, indicator wells are determined in terms of temporal information of wells. In this research, an approach based on a combination of these two methods is presented to consider the spatiotemporal information of wells. The present study was conducted by obtaining data from 47 wells related to the first aquifer of Sarab plain in 1397 in three main stages: 1- Exploratory spatial data analysis (ESDA), 2- Determining the priority of wells based on temporal information of the wells in the neighborhood by PCA method, 3- Investigating the surface accuracy changes by kriging method assuming the removal of wells with low temporal priority. The results showed that 9 wells of Sarab plain (19%) have relative importance less than 0.3. By removing these wells and evaluating the RMSE error in the deleted points, the value of these wells can be calculated with 46cm error through spatial autocorrelation information. Therefore, the removed wells do not enter much spatial information, and by removing them, it is possible to increase the accuracy of measuring the water level in other wells and save time and cost with the same accuracy.