THE EFFECT OF SPATIAL RESOLUTION OF ENVIRONMENTAL VARIABLES ON THE PERFORMANCE OF MACHINE LEARNING MODELS IN DIGITAL MAPPING OF SOIL PHOSPHORUS

Spatially accurate estimation of soil phosphorus is important in terms of agricultural sustainability as well as in the management of eutrophication processes in watershed management planning. Traditional soil maps may not provide data for these needs. With the scientific approach introduced by the digital soil mapping methodology, areas with high phosphorus content can be spatially estimated. We report that spatial resolution does not significantly affect the performance of two different machine learning algorithms at a time when the availability of environmental variables is becoming more and more important. In the modeling process, the use of local land use information would improve the spatial estimation of phosphorus classes.