NEAR REAL TIME PREDICTION OF SPATIAL DISTRIBUTION PROBABILITY OF EARTHQUAKE-INDUCED LANDSLIDES—TAKE THE LUSHAN EARTHQUAKE ON JUNE 1, 2022 AS AN EXAMPLE

At 17:00 on June 1^st, 2022, following the Lushan Earthquake in 2013, an M_S6.1 earthquake occurred again in Lushan County, Ya'an City, Sichuan Province after 9 years. Earthquake is one of the most important factors that trigger geological hazards in mountainous areas, which usually leads to a large number of casualties and property losses. Rapidly and accurately obtaining the spatial distribution of earthquake-induced geological hazards is crucial for post-earthquake emergency rescue and temporary resettlement planning. Based on the global earthquake-induced landslide database, this paper employed Deep Forest algorithm to establish a near real-time prediction model for the spatial distribution probability of earthquake-induced landslides. The model was applied to the rapid prediction of geological hazards induced by the "6.1" Lushan Earthquake, and the prediction results of the spatial distribution probability of geological hazards were obtained within 1 hour after the earthquake. Meanwhile, we arrived at the seismic zone as soon as possible to conduct emergency investigation and model verification of geological hazards. The survey indicates that the geological hazards induced by this earthquake mainly consist of small collapses and landslides. The high-risk areas are mainly distributed in the intersection region of the northern Lushan County and the western Baoxing County. The number of geological hazards in the upper fault is significantly higher than that in the lower fault. Comparing the prediction results with the field survey in the basin of Baoxing Donghe, it can be concluded that the accuracy of the model is more than 80%. In particular, all large-scale landslides exactly occurred in the high-risk areas predicted by the model. The results confirm that the model enables to make up for the lack of timeliness of post-earthquake field investigation and remote sensing data acquisition and provides scientific support for emergency rescue.