A physical-based system for flash torrent and debris flow forecasting

Flash torrents and debris flows are typical hazards that threaten human lives and cause significant losses in mountainous areas. This study introduces a system for process simulation and risk forecasting of mountain disasters (e.g., flash torrents and debris flows) based on the full process of hazard formation, movement, and deposition. The geographic information is managed by basic distributed units, while the meteorological data and real-time monitoring data are directly accessed using the Web Graphics Library (WebGL). Compared with traditional forecasting methods (rainfall thresholds), this platform has two main advantages: 1) it is a multipurpose approach that can either forecast hazard occurrence in a large region or simulate the affected area of a small catchment; 2) it is a complete physics-based method that includes the physical simulation of hazard formation and movement processes. A supercomputer platform based on efficient multiple parallel computing greatly facilitates computational efficiency and increases the lead time for disaster early warning. The platform is currently under trial operation in the mountainous regions of China. In Liangshan Prefecture, Sichuan Province, a typical demonstration plot, 15 debris flow events and 52 flash floods were captured by this system, which delivered early warning information to the local residents. The accuracy was more than 75% in 2022 and 2023, significantly supporting local disaster prevention.

Keywords: System platform, flash torrent, debris flow, physical process, forecasting and early warning.