Modelling spatial distribution of soil types and characteristics in a high Alpine valley (Val di Sole, Italy)

Abstract

SUMMARY - Modelling spatial distribution of soil types and characteristics in a high Alpine valley (Val di Sole, Trentino, Italy) - Detailed soil maps in Alpine areas are often not available due to the high variability of the topography, the inaccessibility of parts of the area and consequently high production costs. In the context of growing demand for high-resolution spatial information for environmental planning and modelling, fast and accurate methods are needed to provide high-quality digital soil maps. We performed a spatial analysis to model several characteristics of Alpine soils in Val di Sole, Val di Peio and Val di Rabbi (in total 374 km 2 ). Soil modelling was performed using a non-parametric classification and decision tree analysis (CART: Classification and Regression Tree Analysis). The classification and decision tree analysis used forced splitting rules (according to expert knowledge). Soil type modelling was done using 15 end nodes. Spatial modelling of humus forms could be achieved with 9 terminal nodes. Field and chemical data (115 sites) served as a basis for modelling. In addition, conventional soil mapping was performed on three relatively small test areas. The modelling results could therefore be tested using these maps. Modelling of soils and humus forms was performed successfully with an accuracy of about 65% for soil types and higher values (up to 78%) for the humus forms. The main soil type in the investigation area is a ranker (WRB: