Multiscale Remote Sensing and GIS-Based Reconstruction of the Valgussera Avalanche Event (Italian Alps): Integrating Sentinel-2/-3, UAV Surveys and Forecast Verification

Avalanche hazard assessment in alpine environments increasingly requires the integration of satellite remote sensing with in-situ and UAV-based observations, given increasing snowpack variability under climate change and growing off-piste and ski-touring activity. This study presents a multiscale, multisensor GIS workflow applied to the 29 January 2026 avalanche event in Valgussera (Foppolo, Bergamo, Italian Orobie Alps), aimed at reconstructing pre-event snow and weather conditions and verifying the operational avalanche forecast against field observations over a nine-day study period (20-29 January 2026).

A single cloud-free Sentinel-2 L2A scene, acquired on 20 January 2026, was processed to derive the Normalized Difference Snow Index (NDSI) and a custom SWIR–NIR–Red composite for snow-cover mapping and dry/wet snow discrimination. One Sentinel-3 SLSTR scene, acquired on 28 January 2026, supported the analysis of surface brightness temperature. These satellite datasets were integrated in QGIS with Meteomont field snow surveys and the 28 January 2026 forecast bulletin, Italian Air Force forecast grids, a digital terrain model, and a single UAV survey (RGB photogrammetric and thermal payload) performed on 29 January 2026 immediately after the avalanche release. A discordance coefficient compared forecast and observed avalanche danger, and the open-source model AvaFrame simulated the avalanche dynamics.

NDSI classification indicated 85% snow cover over the regional alert zone before the event; thermal and multispectral analyses revealed widespread surface wetting at mid-to-low, sun-exposed elevations. Forecast verification showed a systematic 7-13 cm underestimation of fresh snow accumulation, although the spatial danger pattern remained largely consistent with observations. UAV-based reconstruction and AvaFrame simulation reproduced the event's intensity parameters despite limited trajectory agreement.

The proposed multiscale approach demonstrates how integrating optical and thermal satellite data with UAV surveys and numerical modelling can support both process-based understanding and quantitative verification of operational avalanche forecasts, offering a transferable methodology for alpine natural hazard monitoring.