Comparing Deterministic and Geostatistical Methods in Spatial Distribution Study of Soil Physical and Chemical Properties in Arid Rangelands (Case Study: Masileh Plain, Qom, Iran)

Accurate knowledge of spatial distribution of soil physical and chemical properties is needed for suitable management and proper use of rangelands in Masileh plain, Qom, Iran. In present study, for the spatial modeling of chemical and physical parameters such as sodium (Na), calcium (Ca), soluble potassium (K), magnesium (Mg), Electrical Conductivity (EC), Saturation Percentage (SP%), silt, clay and sand percent, 49 soil samples were collected from a depth of 0-30 cm of soil surface in a systematic pattern with 1030 m interval and then, they were compared using deterministic methods (radial basis function, inverse distance weighting and local polynomial interpolation) and geostatistical methods (ordinary, universal and disjunctive kriging, and Cokriging). First normality of data was test using Kolmogorov-Smirnov method. Log transformation was used for non-normal data and finally, spatial structure of the data was determined. Then, cross variogram of parameters was calculated by variography analysis. Then, results were evaluated by MBE and MAE calculation for the predicted and observed data. The results demonstrated that geostatistical methods lead to notable findings rather than deterministic ones. According to the results, the best method for modeling calcium and electrical conductivity parameters was Cokriging method while the best method for saturation percent, magnesium, sodium and silt as well as clay percent parameters was disjunctive kriging. Moreover, ordinary kriging method was suitable for the zonation of potassium and sand percent.