Abstract: Based on investigations of mountain torrents in 2013-2015 and 2016-2019, this study analyzes the temporal and spatial characteristics of 831 historical mountain torrent disasters in Chongqing. The spatial changes of mountain torrent disasters are researched using an ArcGIS gravity model and the standard deviation ellipse model, while the driving factors are identified using the geographic detector model. The frequency of disasters can be divided into three stages based on statistical analysis of disaster frequency over time: a low-frequency period from 1926 to 1977, a low-frequency fluctuation period from 1977 to 2006, and a high-frequency fluctuation period from 2006 to 2017, which is related to literature records, rainfall conditions, and social and economic development. Wavelet analysis shows that the mountain flood disasters in recent years change periodically every 3.7 years. Despite of increased frequency of disasters since 2000, the number of deaths and missing persons due to disasters has stabilized, reflecting the effectiveness of mountain torrent prevention and control measures. The disasters concentrated in the Pengshui and Wulong areas in Southeast Chongqing in May and moved northward in June. After July, the focus of disasters moved westward to west Chongqing and then gradually forward to northeast Chongqing from August to September. The geographic detector model analysis shows that the river network density and elevation factors have a significant impact on mountain flood disasters. After the superposition of each factor, the release force increases non-linearly. This study comprehensively summarizes the historical patterns of mountain torrents in Chongqing, providing important insights into the temporal and spatial evolution pattern of mountain torrents and their driving factors in Chongqing. The results could provide technical support and theoretical guidance for mountain torrent prevention and control.

Key words: mountain torrent disaster, survey results, gravity model, standard deviation ellipse model, geographic detector model, temporal and spatial evolution characteristics, Chongqing