A Comparison of Interpolation Methods for Depth of Groundwater and Its Temporal and Spatial Variations: A Case Study of Kouhdasht Plain, Iran

In some aquifers such as Kouhdasht alluvial aquifer, lack of sufficient data for groundwater modeling by numerical models constrains us to use the various interpolation methods to determine the required data. Understanding temporal and spatial variations in the depth of groundwater in this region is important for developing management strategies. Among numerous interpolation methods, no method is uniquely optimal, and so the best interpolation method for a specific situation can only beobtained by comparing their results. The records of depth of groundwater for 15 peizometers in Kouhdasht plain were available for three years from 2006 to 2009, allowing us to compare five different interpolation methods based on three selected years (2006-2007, 2007-2008, 2008-2009) as starting points. The five methods were IDW, GP, LP, RBF, Kriging (including OK, SP and UK) and Cokriging (including OK, SP and UK). Cross-validation was applied to evaluate the accuracy of various methods, and two indices- R2 and RMSE- were used to compare the interpolation methods. A Comparison of interpolated values with observed values indicated that Cokriging was the optimal method for interpolating depth of groundwater in this region: it had the lowest value of RMSE and the highest value of R2.