2026-10-12T01:55:57Z https://www.scidb.cn/oai

10.57760/sciencedb.j00100.00047 2026-05-07T09:58:23Z

2026-05-07 Rainfall Event and Landslide Triggering Dataset from Automatic Rainfall Stations in Guangdong Province, China (2018-2023) doi:10.57760/sciencedb.j00100.00047 en This dataset contains 57,438 rainfall event records (including 284 landslide-triggering events) from 361 automatic rainfall stations in Guangdong Province, China, covering the period from July 2018 to October 2023. Rainfall events were delineated using the CTRL-T method (an adaptive event segmentation approach based on the statistical distribution of dry spell durations), with season-specific thresholds applied: 48 hours for the wet season (April–September) and 96 hours for the dry season (October–March), in accordance with Guangdong's subtropical monsoon climate.Each record contains the following fields: anonymized station ID (StationID; original hydrological station codes have been de-identified, coded S001–S361), event start time (Event_Start), event end time (Event_End), duration in hours (Duration_Hours), cumulative event rainfall in millimeters (Total_Rainfall_mm), binary landslide triggering label (Is_Triggering_Event; 0 = non-triggering, 1 = triggering), triggering confidence score (Trigger_Confidence; range 0–1), and associated landslide count (Landslide_Count).This dataset is the supporting data associated with: Yuan et al., "Spatiotemporal landslide early warning model for South China based on CTRL-T rainfall event identification and deep learning," Acta Geographica Sinica, Vol. 81, No. 5, 2026. It can be used for training and validation of rainfall-induced landslide early warning models and related research. landslide early warning; rainfall event; CTRL-T method; deep learning; Guangdong Province Yuan Shaoxiong EMBARGO https://mit-license.org dataset Science Data Bank