Precipitation plays a crucial role in landslides, which most studies have disregarded the impact of compound temporal precipitation. Here, we developed models that incorporate compound temporal precipitation, constructed by jointly characterizing long-term accumulated precipitation and short-term precipitation immediately preceding landslides. Results show that integrating compound temporal precipitation improves predictive accuracy, with AUC increases of 1.0–6.9% across different experimental settings. The influence of compound temporal precipitation exhibits clear spatial and seasonal heterogeneity. Southern China is more sensitive to compound temporal precipitation, especially in the southeastern coastal region. The effect intensifies during summer and autumn when precipitation seasonality is strongest. Overall, this study demonstrates that integrating compound temporal precipitation not only boosts predictive accuracy but also provides new insights into the spatiotemporal variability of landslide susceptibility. These findings underscore the necessity of incorporating compound temporal precipitation in future frameworks for landslide prediction.
Wang et al. (Fri,) studied this question.