Construction of a Computer-Aided Dual-Parameter Landslide Prediction Method

Abstract

This study focuses on the typical geological region of granitic weathered soil in the hilly areas of southern China, integrating computer-assisted statistical methods and intelligent computing techniques to analyze geological disaster data and hourly precipitation data from 2016 to 2022. The research indicates that in the study area, geological disasters triggered by precipitation generally occur within 48 hours following the precipitation event. For 90% of the events, the average rainfall intensity is concentrated between 1.2 to 27.7 mm/h, with 72% having an average rainfall intensity of less than 10 mm/h, and only 13% exceeding 20 mm/h. The influence of prior precipitation mainly depends on the cumulative effect of effective rainfall over the previous 11 days. Based on these findings, this paper constructs a meteorological risk prediction model for geological disasters using two parameters: the average rainfall intensity (I) and the duration of rainfall (D). After comparing the prediction results with 62 test samples from 2022, it is noted that the method's TS score is superior to other prediction methods, significantly reducing the false alarm and missed alarm rates for meteorological forecasts in the granitic weathered soil geological region. Moreover, the computer implementation of this model demonstrates its efficiency and reliability in handling large-scale geological data, providing strong computational support for geological disaster risk assessment.