Automatic Delineation of Slope Units and Terrain Classification of Italy (poster)

Automatic delineation of slope units and terrain classification of Italy Automatic subdivision of landscapes into terrain units, in large areas and with homogeneous criteria, remains a challenge. Slope units (SUs) are irregular terrain units bounded by drainage and divide lines, with the requirement of maximizing geomorphological homogeneity within a single unit and heterogeneity between different ones (Guzzetti et al., 1999). SUs are suitable to describe a variety of processes: to assess different natural hazards, for hydrological and erosion modelling, for geo-environmental, ecological, forestry, agriculture, and in general for studies requiring the identification of homogeneous terrain domains facing distinct directions. Nevertheless, their use has been hampered because of the lack of reliable software for their automatic delineation. For a given digital landscape, no unique SU partition exists. Any such partition is characterized by a granularity i.e., the size distribution of the SUs. Granularity varies in different landscapes and for different purpose of the terrain partition. We recently developed the r.slopeunits software (Alvioli et al, 2016), which can draw slope unit polygons on a digital surface. It provides an unambiguous and reproducible definition of slope units based on quantitative hydrologic and topographic criteria, allowing for a flexible yet well-defined SU delineation through an adaptive algorithm. The code is designed to quickly produce results on large areas. It requires a digital elevation model and a few input parameters, whose values must be optimized in a sound way, by means of multiple software runs and a proper objective function to be maximized. We propose a parameter-free SU map of the entire Italian territory, containing about 300,000 SU polygons. We optimized the map considering only objective morphometric quantities. To assess the validity of the nation-wide SU partition, we classified the whole of Italy into different categories using a clustering procedure, initialized using only morphometric properties of the SUs. We compared the classification form clustering with zonal statistics of independent quantities: terrain elevation and slope, drainage density and lithology. The analysis confirmed that the SU partition was geomorphologically sound, suggesting that the proposed method is effective for a general-purpose SU delineation of a large, complex and heterogeneous landscape. The whole SU map is available for download (CNR-IRPI URL).