Landslide sediment transfer in Val di Sole, eastern Central Alps

Landslides exert prominent controls on the morphology of mountain drainage basins and drive the overall sediment flux across landscape components. In this context, historical landslide inventories are critical for quantifying landslide sediment dynamics through time and assessing relevant contributions to regional sediment budgets. We present an historical inventory that combines the mapping of deep-seated gravitational slope deformations (DSGSD) and shallow landslides in Val di Sole (707 km2), eastern Central Alps, Italy. The study area was selected for two reasons: (i) the composite geological setting including intrusive, metamorphic, and sedimentary formations, which provides the opportunity to evaluate lithological effects on landslide sediment transfer; and (ii) the homogeneous distribution of forest cover, which warrants higher reliability of landslide identification through time. In so doing we avoid areas dominated by large rock walls and sedimentary linkages (e.g., unvegetated talus slopes and screes), typically associated to a chronic flux of colluvial sediment. In these conditions, owing to the subtle contrast between parent material and freshly eroded debris, the identification of fresh landslide scars in sequential photosets is difficult and unreliable. For the same reason, evaluation of the visibility time window for clusters of landslides would become highly uncertain. Data collection involved interpretation of seven sequential photosets (1959, 1969, 1973, 1983, 1996, 2000, 2006) and LiDAR-derived hillshade rasters in conjunction with field measurements. Fieldwork served to measure landslide depths and obtain a volumetric transformation factor for remotely-sensed landslide areas. Inspection of LiDAR hillshades allowed to identify and delineate the perimeter of large slope deformations, otherwise masked by a number of environmental conditions during aerial photo interpretation (e.g., forest cover, shadow, and snow). In the compilation of the inventory, the use of high-resolution hillshades and aerial photos did not prove to be mutually exclusive. If on hand, the former method reduces uncertainty in the identification and delineation of landslide features, on the other it is too expensive to be repeated in time and obtain a multi-temporal dataset. Aerial photographs have limitations in terms of landslide identification, but they are relatively affordable, they have good coverage for large parts of the country since 1950s, and allow to analyse landslide sediment dynamics through time. Results show that the two methodologies should be employed in an integrated approach. We identify a total of 220 DSGSD and 3369 shallow rapid failures that cover an area of 174 km2 and 92 km2 respectively. The corresponding sediment flux exhibits lithologic and topographic controls. Metamorphic, sedimentary, and intrusive terrain are associated with progressively lower rates of sediment production and subsequent delivery to the drainage network. In this context, local confounding modulated by the presence of thick glacigenic deposits is common. Interestingly, lithologies with high density of DSGSD events exhibit higher rates of shallow-rapid failures. This finding suggests that DSGSD events by favouring/preparing the occurrence of shallow rapid failures may control, in an indirect way, landslide sediment flux at the regional scale.