How and to What Extent Does Topography Control the Results of Soil Function Assessment: A Case Study From the Alps in South Tyrol (Italy)

Soil function assessments (SFA) are becoming increasingly important as a tool to integrate soil-related issues in decision-making processes in order to maintain soil quality. We present the SEPP (Soil Evaluation for Planning Procedures) tool, which calculates a level of fulfillment for 14 soil functions based on the information generally collected in soil pit descriptions. By using a statistical modeling approach based on support vector machine classification, we investigate how and to what extent topography, as representated by local terrain parameters and landform classes computed with the GRASS GIS tool r.geomorphon algorithm, controls soil parameters and hence the output of the SEPP tool. A feature selection procedure is applied which highlights those topographic attributes best suited for modeling the various soil function fulfillment levels. By evaluating the model for each soil function using cross-validation we show that the prediction accuracy varies from function to function. While some terrain attributes are directly implemented in the SFA algorithms of SEPP, others are implemented indirectly due to the link between topography and land use. Minimal curvature and slope were found to be first indicators of function fulfillment level for a number of soil functions.