Towards Establishing Empirical Rainfall Thresholds for Shallow Landslides in Guangzhou, Guangdong Province, China

Abstract

Empirical rainfall thresholds for predicting rainfall-triggered shallow landslides are proposed for Guangzhou city, which is prone to widespread geological hazards during the annual flood season due to the subtropical monsoon climate and frequent tropical storms and typhoons. In this study, the cumulated event rainfall (E, in mm), the duration of rainfall event (D, in day) (E–D) thresholds, normalized cumulated event rainfall, and the duration of rainfall event (EMAP–D) thresholds were defined. Thresholds based on five lithological units were obtained at 5%, 20%, and 50% probability levels using quantile regression methods. More than two-thirds of the landslides occurred within units of intrusive rock. The 20-day cumulative rainfall of 97 mm integrating cumulative event rainfall and the duration of rainfall events (CED) is introduced into the three-dimensional spatial threshold. The areas under the receiver operating characteristic curves for the CED threshold and E–D threshold were 0.944 and 0.914, respectively, and the true-positive rate of the CED threshold with the same probability level was slightly lower than that of the E–D threshold, but the CED threshold false-positive rate was much better than the E–D threshold, which can significantly reduce false alarm rate since many non-triggering rainfalls were filtered out.