Assessing the accuracy of geostatistical techniques for mapping soil macronutrients on basaltic landscape of central India

Abstract

A study was conducted to interpolate and to explore the analysis of spatial variability of major soil nutrients in Basaltic Terrain of Bemetara district, Chhattisgarh. A total of 182 soil samples (0-25 cm) were collected randomly using GPS. Soil chemical properties i.e. available nutrients (N, P and K) were measured in laboratory. Data were interpolated by Ordinary Kriging (Spherical, Circular, and Gaussian). The performance of methods was evaluated using Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Goodness of prediction (G) obtained from a cross-validation procedure. The results showed that Circular, spherical, and Gaussian models were found best fit for available N, P, and K, respectively. All variables showed strong spatial dependence. Cross validation of kriged map showed that spatial prediction of soil nutrients using semi variogram parameters is better than assuming mean of observed value for any unsample location. Therefore it is a suitable alternative method for accurate estimation of soil properties in unsampled positions as compared to direct measurement which has time and costs concerned.