A pattern recognition approach to generate soil process units for ecosystem modeling

The landscape-scale evaluation and modeling of the impact of agricultural management and climate change on soil-derived ecosystem services requires soil information at a spatial resolution addressing individual agricultural fields. A pattern recognition approach is presented that generates a nationwide data product. It agglomerates the multivariate soil parameter space into a limited number of functional soil process units (SPUs) that facilitate operating agricultural process models. Each SPU is defined by a multivariate parameter distribution along its depth profile from 0 to 100 cm. It has a depth resolution of 1 cm and a spatial resolution of 100 m. The methodological approach is based on an unsupervised classification procedure involving remote sensing, cluster analysis, and machine learning. It accounts for differences in variable types and distributions and involves genetic algorithm optimization to identify those SPUs with the lowest internal variability and maximum inter-unit difference with regards to both, their soil characteristics and landscape setting. The high potential of the method is demonstrated for the agricultural soil landscape of Germany. It can be applied to other landscapes and ecosystem contexts.