Morphological parametrisation of snow cover variability as preparatory factor for landslides: the Campotosto (Central Italy) test site

Landslides are natural land-forming processes and their interaction with humans makes them one of the most common geo-hazards. Landslides are controlled by three macro-categories of factors, which are the “predisposing”, “preparatory”, and “triggering” ones. Earthquakes and heavy rainfall usually act as triggers, but their action is typically anticipated by a combination of predisposing factors and preparatory processes, that are investigated in PE3 RETURN – VS2, respectively, by WP2.2 and WP2.3. In particular, while predisposing factors are static, preparatory factors are especially significant as they change over time, gradually reducing the slope stability without initiating the movement. Snow accumulation and melting are generally discussed in scientific literature as triggering factors for landslides, particularly shallow ones (see, for example, Cardinali et al., 2000), although the approach presented here focuses on their contribution also as preparatory factors. In fact, in the context of DS Spoke, snow is recognized as an important impact-oriented hazard indicator for landslides, especially in mountainous and hilly areas, alongside other common weather-climatic parameters related to rainfall and temperature. In mountain and hill slopes, snow loading and, particularly, seasonal snow melting can increase the soil pore water pressure, causing a reduction of available strength. Their influence on slope stability is time-dependent and, in detail, changes cyclically throughout the year. Snow usually start accumulating in the late autumn, with peaks especially in late winter, then it melts in spring to summer, resulting in water infiltration into the soil and resistance loss. However, in Italy and especially in the Apennines, more complex seasonal trends can occur, including intraseasonal transitions between accumulation and melt and frequent episodes of complete melt out even in mid-winter. In seismically active areas, earthquakes acting as triggers for shallow landslides can encounter different levels of soil weakness throughout the year depending on the snowpack and related hydraulic conditions, resulting in distinct landslide scenarios. The same variability of effects applies to heavy rainfall and is intrinsically related to the influence of preparatory factors. The main aim of this research is to assess the preparatory effect of snow cover on landslide phenomena, taking into account the temporal and spatial variability of snow accumulation and melting. The test site is located close to the Campotosto Lake (Italy), in the Central Apennines, where considerable, albeit highly variable from season to season, snowfall per year is recorded, and strong earthquakes are expected due to the proximity of the Mt. Gorzano seismogenic source. The analysis for the quantification of the preparatory role of snow covers for landslides was conducted in two steps. On one side, snow accumulation data were analysed through a Bayesian Estimator of Abrupt change, Seasonal change, and Trend (BEAST; Zhao et al., 2019), to reliably detect both abrupt changepoints in the seasonality and trend of snow height (HS) for each cell of the area of interest. In this way, it is possible to identify and discriminate representative time windows in which snow accumulation, or, alternatively, snow smelting is expected. On the other side, HS data were grouped using the above mentioned time windows, and compared with geomorphometric parameters (e.g. aspect, slope, etc.), in order to identify a spatio-temporal correlation between the terrain and the snow characteristics, facilitating the understanding of the dynamics of landslide processes in relation to snow preparation. TINITALY v. 1.1 (Tarquini et al., 2023) was used as DTM, while snow data were taken from IT-SNOW (Avanzi et al., 2023), because: i) both of them are freely available throughout Italy, ii) IT-SNOW is the most resolute temporal and spatial snow cover data for Italy. In detail, IT-SNOW is a snow reanalysis providing daily information on snow depth and mass, to date between the years 2010-2023. In this study, it was used in the original ~200 m cell size raster, extracted directly from S3M Italy (the underlying operational model of IT-SNOW), and then resampled, to spatially compare it with geomorphometric parameters (~10 m cell size). The here reported preliminary results are really encouraging in the perspective of modelling forward landslide scenarios.