Evaluation of spatial interpolation techniques for mapping soil pH

Abstract

Soil pH has a major effect on plant nutrient availability by controlling the chemical structure of the nutrient. Adjusting soil acidity or alkalinity improves soil nutrition without adding extra fertilizers. Soil nutrients needed by plants in the largest amount are referred to as macronutrients. In addition to macronutrients, plants also need trace nutrients and both macro and trace nutrient availability is controlled by soil pH. Understanding of spatial variability of soil properties is important in site-specific management. Analysis of spatial variation of soil properties is fundamental to sustainable agricultural and rural development. The special variability of soil property is often measured using various interpolation methods resulting in map generation. Selecting a proper spatial interpolation method is crucial in surface analysis, since different methods of interpolation can lead to different surface results. Among statistical methods, geo-statistical kriging-based techniques have been frequently used for spatial analysis and surface mapping. In this work, three common interpolation methods are used to study the spatial distributions of soil pH in a vineyard. Interpolation techniques were used to estimate the pH measurement in unsampled points and create a continuous dataset that could be represented over a map of the entire study area. The method investigated includes; Inverse Distance Weighting (IDW), Radial base Function (RBF) and Ordinary Kriging (OK). The performance of conventional statistics showed that soil pH had a law variation in this study. Experimental anisotropic semivariograms were fitted with the Spherical, Exponential, Gaussian and Exponential models and the Exponential model was found as the best fitted model using the cross-validation method. The performances of interpolation methods were evaluated and compared using the cross-validation. The results showed that RBF method performed better than IDW and OK for prediction of the spatial distribution of topsoil pH (Figure 1).