Numerical-model-derived intensity–duration thresholds for early warning of rainfall-induced debris flows in a Himalayan catchment

Abstract. Debris flows triggered by rainfall are catastrophic geohazards that occur compounded during extreme events. Few early warning systems for shallow landslides and debris flows at the territorial scale use thresholds of rainfall intensity–duration (ID). ID thresholds are mostly defined using hourly rainfall. Due to instrumental and operational challenges, current early warning systems have difficulty forecasting sub-daily time series of weather for landslides in the Himalayas. Here, we present a framework that employs a spatio-temporal numerical model preceded by the Weather Research And Forecast (WRF) Model for analysing debris flows induced by rainfall. The WRF model runs at 1.8 km × 1.8 km resolution to produce hourly rainfall. The hourly rainfall is then used as an input boundary condition in the spatio-temporal numerical model for debris flows. The debris flow model is an updated version of Van Asch et al. (2014) in which sensitivity to volumetric water content, moisture-content-dependent hydraulic conductivity, and seepage routines are introduced within the governing equations. The spatio-temporal numerical model of debris flows is first calibrated for the mass movements in the Kedarnath catchment that occurred during the 2013 North India floods. Various precipitation intensities based on the glossary of the India Meteorological Department (IMD) are set, and parametric numerical simulations are run identifying ID thresholds of debris flows. Our findings suggest that the WRF model combined with the debris flow numerical model shall be used to establish ID thresholds in territorial landslide early warning systems (Te-LEWSs).