Appraisal of the geostatistical methods to estimate Mazandaran coastal ground water quality

Authors

1 1*, A. Shahnazari2

2 0

Abstract

The present study was carried out to evaluate three interpolation methods including weighted moving average (WMA) with the power of 2 and 3, Kriging and Cokriging methods. Data of 23 wells in Mazandaran province were collected in fall and spring 2006. Seven parameters including electrical conductivity (EC), pH, total dissolved solids (TDS), sodium adsorption ratio (SAR), total hardness (TH), chloride concentration (Cl- ) and sulphate concentration (SO42- ) have been chosen as groundwater quality indices in the study area. Variogram analysis and extracting the spatial distribution maps of groundwater quality parameters were done using Geostatistics extension program in GIS environment. All interpolation methods have been evaluated based on mean bias error (MBE) and mean absolute error (MAE) criteria. The spherical model for semi-variograms had the less value of RSS (residual sum of square) for Cl-, EC, pH, SAR and SO42- parameters. TDS and TH parameters followed a Gaussian model. All semi-variograms and cross variograms had high confident level due to little values of nugget effects (Co) relative to sill. The covariance matrix demonstrated that magnesium concentration (Mg2+), sodium concentration (Na+), Total anions, Cl-, EC and TDS parameters have been the best covariate for estimating TH, SO42-, Cl-, PH, TDS and EC parameters, respectively. Co-Kriging was the best method for estimating all parameters far apart TH for which Kriging method was the best. Spatial distribution maps of groundwater quality indices demonstrated that the groundwater in the study area is slightly basic and the values of EC exceeded the permeable limit in more than 40% of the study area. Also there was sodium hazard and high concentration of TDS in the north-east part. Therefore, further studies are needed to recognize the pollution sources in order to reclaim the polluted part in the study area.

Keywords