Assessment of Geostatistical Methods for Determining Distribution Patterns of Groundwater Resources in Sari-Neka Coastal Plain, Northern Iran

This study aimed to evaluate the temporal change and accuracy of interpolation techniquesused for spatial zonation of two groundwater quantity parameters including water table anddepth to water table over 11 years. The study was conducted based on the data collectedfrom piezometric wells of Sari-Neka Plain in Mazandaran Province, Iran. The investigatedmethods included a set of geostatistical approaches involving simple Kriging, ordinaryKriging, Radial Basis Function (RBF), and a deterministic interpolation method calledInverse Distance Weighting (IDW) with powers of 1 and 5. Subsequent to quality controland data normalization, the most appropriate variogram was chosen based on low RSS andhigh r2 while the most suitable interpolation technique was determined regarding the crossvalidation, Mean Absolute Error (MAE), and Mean Bias Error (MBE). The resultsdemonstrated that Simple Kriging was the most suitable method for zoning the depth togroundwater over the years 2001, 2006, and 2012. Meanwhile, the most suitable methodsfor zoning the water table included IDW with a power of 1for the year 2001, RBF for theyear 2006, and IDW with a power of 5 for the year 2012. The important finding was thatthe interpolation methods showed a lower error for estimating water table than estimatingdepth to groundwater. This study also revealed a drop in water table in the study area overthe 11 years’ period. Meanwhile, new water table classes have been added and extendedbetween the years 2006 and 2012 that had not existed five years earlier. The highest watertable losses were observed in three points at 13m depth to water table in the middle andnorthern parts of the study area.

Keywords:

Interpolation Methods , Variogram , Geostatistics , Groundwater , MAE , MBE