Abstract

Groundwater is one of the most important alternative water resources for civil and agricultural uses in arid and semi-arid regions, due to the increasing demand for domestic and agricultural uses. Therefore, groundwater forecasting is useful for managing the size, level, planning and maintenance of agricultural development. In this study, three geostatistics methods were compared. A dataset of the measured water level of 75 selected wells was selected in the fourth month of 2013 in Tal Afar for an area of 4596 km2, to choose the best method for producing the most accurate numerical map of groundwater (GWT). .Spatial Interpolation Techniques have been applied in three ways (kriging, IDW, and RBF) for spatial interpolation of groundwater. Several models of semivariograms (Spherical, Gaussian, Invers Multiquadric, Exponential) were performed and after testing for several models, the final model was selected to estimate groundwater levels in the area and the amount of variance (which expresses the accuracy of the water level estimation) Groundwater), and the best method of prediction was determined based on Cross-Validation technique. RMSE is the best model, and RBF is best used to create an Interpolation Map because it has a lower error ratio compared to the results of kriging (IDW). The above methods were implemented using ArcGIS 10.3.Keywords: spatial statistics, groundwater levels, cross-validation (IDW), (RBF), Kriging, ArcGIS 10.3