Assessment of Landslide Susceptibility under Climate Change in the Laonong river watershed in Taiwan

The study focuses on the distribution of landslide susceptibility in the future under climate change in the Laonong river watershed (abbreviated as LRW) in southwestern Taiwan. LRW is a mountainous watershed prone to sediment disaster and had caused serious sediment disaster during 2009 Typhoon Morakot. The study used the downscaled daily rainfall data provided by Taiwan Climate Change Estimation Information and Adaptation Knowledge Platform (TCCIP) as the daily rainfall data in the future in LRW. We combined the daily rainfall data and the landslide susceptibility model in the LRW to assess the distribution of landslide susceptibility in the future in the LRW.

The landslide susceptibility model was composed of 9 landslide-related factors, including elevation, slope, aspect, geological setting, landuse, Topographic Wetness Index, distance from the rivers, landslide frequency, and daily rainfall. This study built the landslide susceptibility model of LRW based on the daily rainfall data and landslide inventory after 2009 Typhoon Morakot. The AUC (area under receiver operating characteristic curve) of the landslide susceptibility model is 0.712, and the accuracy by using the confusion martix is 0.731.

The study also compared the rainfall characteristic in the past (the rainfall data from 1998 to 2022) and the future (the downscaled rainfall data from 2023 to 2100) in the LRW. No significant difference shows between the characteristic of average annual rainfall in the past and the future, but the monthly rainfall is obviously concentrated in the rainy season, i.e. from May to October. The occupied percentage of accumulated rainfall in the rainy seasons to annual rainfall in the future is larger by 0.3% to 4.1% than that in the past. The daily rainfall with 50 years return period in the future is larger by 56% to 125% than that in the past.

The study combined the daily rainfall data in the future under climate change scenarios SSP126 and SSP585 and the landslide susceptibility model based on 2009 Typhoon Morakot to assess the distribution of landslide susceptibility in the LRW in the future. The area of middle-high and high landslide susceptibility in the LRW increased obviously based on the distribution of landslide susceptibility in the future under climate change. The average landslide susceptibility value in the future in the LRW is larger by 1.7 times than that in the past.