Application of Empirical Bayesian Kriging to Map Soil Contamination by Heavy Metals (Case study: Esfarayen city)

Kriging methods especially ordinary kriging are more popular than other interpolation methods because of less uncertainty and the least possible variance. However, various complications in the procedure of this method limit its uses. Therefore, empirical Bayesian kriging has been introduced to overcome the problems of ordinary kriging.

In this study, the efficiency of ordinary kriging and empirical Bayesian kriging was investigated to determine the spatial pattern of concentrations of Ni, Cu and Zn in the soils surrounding the Esfarayen industrial complex. For this purpose, 35 surface soil samples (0-20 cm) were collected in area of nearly 87 Km2 and the metal concentrations were determined in the soil samples using a Flame Atomic Absorption Spectrophotometer (FAAS). The ordinary and empirical Bayesian kriging were utilized for interpolating and cross validation including RMSE, NSE and PBIAS were used to assess and compare the efficiency of two methods.

The results showed the mean concentration of Ni, Zn and Cu were 23.61, 58.47 and 12.51 mg/kg respectively. Based on the results the mean concentration of Zn and Cu were more than background concentrations of the metals, whereas Ni concentration was less than background concentration.
Discussion and

The experimental vario-gram of Cu, Zn and Ni concentrations were best-fitted by exponential, exponential and spherical models respectively. The results of cross validation indicated that the empirical Bayesian kriging was more accurate than ordinary kriging to estimate the elements concentration, though this difference was not considerable.