Evaluation of Cokriging and Neurofuzzy Model Performance in Estimating the Nitrate Concentration in Karaj Aquifer

Recently, new techniques based on geostatistical methods have been used to estimate groundwater nitrate concentrations in unmeasured areas as well as to determine new sampling locations. In this study the Cokriging and Anfis models have been developed in interpolation step for nitrate parameter spatiovariation in Karaj aquifer. Nitrate concentrations have been estimated annually using samples derived from 179 drinking water wells. For this purpose, values of nitrate concentration in 1384 (2005) have been considered as the initial values. Nitrate concentration in 1379 to 1383 (200-2004) have been applied as the covariate for cokriging model and as the input parameters for neuro fuzzy model. The comparison between cokriging and Anfis results showed that in five neuro fuzzy models, the error function are less than the cokrihging model, especially for the data of 2004.