Spatial variability of soil properties determined by the interpolation methods in the agricultural lands

Geostatistical methods are widely applied to determine the spatial variability of soil parameters in the unknown points, as their main benefit to classical statistics, through incorporating limited measured data. The objective of the present research was to investigate the spatial variability of various soil parameters determined by different geostatistical methods including ordinary kriging (OK) and radial basis function (RBF) in the east of the Karun River, south of Ahvaz, Iran for the purpose of recommending fertilization practices. A total of 250 soil samples (0–30 cm) were randomly collected from the study area and were analyzed for pH, salinity (EC), available potassium (AK), CaCO3, sand, silt, and clay. The precision of the prepared maps was determined by the cross-validation analysis using root mean square error (RMSE), mean absolute error (MAE), mean bias error (MBE) and Q–Q plot. The C.V. values indicated the highest spatial distribution for EC (49.4%), AK (39.6%) and sand (56%). However, the pH variable had a little variability (C.V. of 3.7%), and clay (24.7%) and silt (21%) had moderate variability. The variogram analyses indicated the effective ranges for pH (610.9), EC (255.7), AK (253.16), CaCO3 (31.6), sand (611), silt (254) and clay (252). According to the results RBF method could provide higher precision for EC, CaCO3, sand, and silt, and OK is a more precise technique for the estimation of pH, AK, and clay. Such results could be applied for the proper handling of agricultural lands and increasing the efficiency of yield production by recommending efficient fertilization techniques.