Comparison of the effectiveness of GIS-based interpolation methods for estimating the spatial distribution of agrochemical soil properties

Abstract

Various interpolation methods were compared in the ArcGIS (a complex of geoinformation software products of the American company ESRI) geographical information system environment for assessing agrochemical soil properties on the territory of LLC “Neral-Chishmy” agricultural enterprise. The study aims to compare the effectiveness of interpolation methods based on GIS technologies for assessing the spatial distribution of agrochemical properties in the soil. Six interpolation methods were used to generate the spatial distribution of the studied indicator: inverse distance weighting, local polynomial interpolation, radial basis functions method, simple Kriging, ordinary Kriging, Universal Kriging. Cross-check was used to evaluate the interpolation methods accuracy by comparing the values of mean error, mean square error, mean square normalized error. It was found that the model created using the Local Polynomial Interpolation method had the lowest values of the average difference between the measurement and the interpolated value and the mean square error equal to 165.9908, indicating how closely the model predicts the measured values. An artificial neural network was created for spatial and temporal modelling of agrochemical properties in the soil, introducing new technologies into agricultural production. It is established that universal Kriging is the most suitable method for interpolating the spatial distribution of agrochemical properties of soils, namely humus, with the lowest RMSE value of 0.7198. By 2025, current agricultural technologies will gradually reduce humus content in layer 0...40 cm. Implementing the interpolation method for assessing the agrochemical properties spatial distribution in the soil will significantly increase the humus content.