A Comparison of Inverse Distance Weighting and Ordinary Kriging for Characterising within-paddock Spatial Variability of Soil Properties in Western Australia

Abstract

Site-specific crop management requires matching resource application and agronomic practices with soil and crop requirements, as they vary in space and time within a field. As such, information on the composition of soils at either farm or paddock scale is essential. Soil composition over an entire paddock might not be uniform. For instance, it may not be efficient to fertilise an entire paddock if only the northeast corner show deficiencies. Furthermore, it is not possible to sample every centimetre of the paddock, as this would be a very time consuming and costly procedure. Ideally, we should be able to collect enough sample points so that continuos maps of soil properties can be accurately produced using spatial interpolation techniques, and good judgements can be made about the soil composition of an entire paddock. Two spatial interpolators, namely kriging and inverse distance weighting (IDW) are compared in this paper, evaluating their accuracy in the prediction of seasonally stable soil properties at paddock level. Organic matter content and soil reaction (pH) are the soil properties selected for this investigation, as previous research by Yule et al. (1996) and Brady and Weil (1999) has found these properties to be correlated to cropping potential. The selected spatial interpolators are tested on an existing data set (e.g. 100 soil samples) collected using a stratified random approach over a 60 ha paddock, located in the southwest of Western Australia. The research approach consists of three main steps. Firstly, basic descriptive statistical techniques are applied to identify the presence of outliers. Secondly, selected interpolators are applied to generate continuous maps of the soil properties, and finally the mean root mean square error (RMSE) is used as the validation index, and to select the best spatial interpolator for the soil properties selected. Therefore, for each soil property, every fourth sample is removed from the main data set and put aside to validate the result of the interpolation method. Likewise, the map of uncertainty generated by using the kriging interpolation method is used as an additional source for evaluating the quality of the prediction maps. The RMSE results show IDW to perform slightly better than kriging for pH (e.g. 0.38 vs 0.39) and organic matter (1.25 vs 1.34)