Geostatistical modeling of fertility of an agricultural land using maps of Kriging interpolating soil chemical analysis data

Abstract

Since soil fertility can vary greatly over relatively short distances, the classical statistical method is not suitable for analyzing data from chemical analyzes of soils. The Kriging interpolation method, on the other hand, is capable of interpolating data to generate, with a certain level of confidence, georeferenced maps of these properties. In this study, 25 soil analyzes belonging to the same field were carried out to obtain data on pH, the effective cation exchange capacity, availability of nutrients and the presence of anti-nutrients; which were tabulated together with their respective coordinates. Statgraphics® 18 and PSPP software exhibited high values ​​of standard deviation, kurtosis, or skewness with respect to the mean of each data series, which invalidated the reliability of possible regressions and correlations. On the other hand, the variogram analyzes made using Vesper® yielded nugget and plateau data that indicated the existence of spatial correlation, making reliable the use of the ordinary Kriging form: “a + b (x)”, to generate a raster layer of data interpolated by each variable, through QGIS and SAGA software, using a one thousandth cell size for good resolution. This function calculates each interpolation as the average of the surrounding values ​​plus their spatial lag, then adding the estimated variance between the surrounding points. Based on the raster, contour curves were drawn to delimit the areas with similar values. The maps allowed the identification of terrain conditions based on location.