河北地质大学学报2025，Vol.48Issue(4)：69-75,7.DOI:10.13937/j.cnki.hbdzdxxb.2025.04.009
地震动峰值速度预测:基于地形特征的随机森林方法
Prediction of Peak Ground Vibration Velocities:A Random Forest Approach Based on Topographic Features
摘要
Abstract
Peak Ground Velocity(PGV)is an intensity parameter for assessing the destructiveness of earthquakes,and its fast and accurate prediction is crucial for earthquake early warning.Studies have shown that topography has a significant effect on the intensity of ground shaking.However,traditional empirical prediction models are difficult to consider complex factors such as topography,which limits the accuracy of prediction results.In this paper,a random forest-based PGV prediction model is proposed by establishing a nonlinear mapping relationship between PGV and magnitude,epicenter distance,and multiple topographic factors.The model was trained and validated using 31 404 strong surface earthquake records from the KiK-net database in Japan,and compared and analyzed with the widely used BA08 and CY08 models.The results show that the correlation of the RF model in predicting PGV is improved by about 18%,the standard deviation of the error is reduced by about 35%,and the problems of"overestimation of small values"and"underestimation of large values",which are commonly found in the traditional models,are effectively avoided.关键词
地震动峰值速度/地形特征/随机森林/地震动预测Key words
peak ground velocity/topographic features/random forests/site conditions分类
天文与地球科学引用本文复制引用
鲍余,王延伟,桂飞..地震动峰值速度预测:基于地形特征的随机森林方法[J].河北地质大学学报,2025,48(4):69-75,7.基金项目
广西自然科学基金(2021GXNSFFA196001) （2021GXNSFFA196001）
