Model-derived hydrological signatures of debris avalanches and debris flows for enhanced landslide prediction

The Norwegian landslide forecasting and warning service provides daily regional predictions of shallow landslides (i.e. debris avalanches and debris flows) triggered by intense rainfall, snowmelt and high soil moisture conditions.While landslide initiation is clearly linked to hydrological processes through infiltration and increasing soil-water pressure, no distinct signature from rainfall–runoff models has yet been identified for use at local scale alongside existing landslide forecasting models. Progress is limited because few landslides occur in catchments with calibrated hydrological models, leaving little basis for relating landslide triggers to simulated hydrological states.To address this gap, the Norwegian Water Resources and Energy Directorate (NVE) has developed a system to parameterise the Distance Distribution Dynamics (DDD) rainfall-runoff model for ungauged basins. The DDD model use a parsimonious set of parameters that can be estimated from landscape and climatic characteristics. We configure the DDD model for landslide-affected catchments, using samples from the Norwegian landslide database (containing landslide type, location, time of occurrence and observation quality), and simulate time series of hydrological variables at 1 hour temporal resolution from 2014 onward.The DDD model simulates hydrological variables such as soil moisture in saturated and unsaturated zones, snow parameters, flood values, and runoff. By examining these variables at the time of landslide events, we aim to identify hydrological signatures associated with landslide initiation. Preliminary results indicate that, subsurface saturation, high flows, and the incremental rate in subsurface saturation relative to the incremental rate in runoff, are key factors in triggering shallow landslides. Analysis of historical events across multiple regions supports dependencies between simulated hydrological states and landslide occurrence. Ultimately, integrating simulated hydrological states into operational forecasting could enhance landslide prediction and improve early warning systems.