Comparative study of interpolation methods for mapping soil pH in the apple orchards of Murree, Pakistan

Abstract

Soil pH is considered as a core indicator for nutrient bioavailability. Prevailing alkaline pH due to calcareousness in Pakistan is considered as one of the limiting factor for nutrient availability to plants. Exploring the spatial variability of soil variables serves as scientific basis for the generation of soil management strategies. Selection of best interpolation method to predict the soil properties at un-sampled locations is an important issue in the site specific investigations. This article evaluates Inverse distance weighting, global and local polynomial interpolation, radial basis function and kriging to determine the optimal interpolation method for mapping soil pH. Performance of the interpolation methods was analyzed using soil test (pH) data from 180 surface soil samples collected from 30 representative orchards grown in tehsil Murree. For inverse distance weighting, powers of 1, 2 and 3 were used and the number of neighbors for all methods ranged from 15 to 25. The conclusion of our study suggested that increased power of inverse distance weighting resulted in an increase in the prediction accuracy. Local polynomial interpolation method was more suitable as compared to global polynomial interpolation. Radial basis function with regularized and spline tension gave equivalent prediction accuracy. Higher errors (mean and mean absolute errors) were observed in case of ordinary kriging as compared to other interpolation methods. Digital maps generated by the higher powers of inverse distance weighting, local polynomial interpolation, and radial basis function were of higher accuracy.