区域降雨诱发滑坡阈值特征分析  

Characteristics analysis of rainfall-induced landslide thresholds in local regions

作　　者：刘海知 徐辉 包红军[1,4] 徐为 闫旭峰[3] 徐成鹏[1,4] LIU Haizhi;XU Hui;BAO Hongjun;XU Wei;YAN Xufeng;XU Chengpeng(National Meteorological Center,Beijing 100081,China;China Institute of Geo-Environmental Monitoring,Beijing 100081,China;College of Water Resource and Hydropower,Sichuan University,Chengdu 610065,China;CMA-HHU Joint Laboratory for HydroMeteorological Studies,Beijing 100081,China)

机构地区：[1]国家气象中心,北京100081 [2]中国地质环境监测院,北京100081 [3]四川大学水利水电学院,四川成都610065 [4]中国气象局-河海大学水文气象研究联合实验室,北京100081

出　　处：《自然灾害学报》2021年第4期181-190,共10页Journal of Natural Disasters

基　　金：国家重点研发计划(2019YFC1510702)。

摘　　要：本文运用全国加密气象站降雨观测数据对数千起滑坡灾害的致灾降雨因子展开了系统性分析,通过改进致灾降雨场次划分方法,确定本场致灾降雨过程。在静态临界阈值模型算法基础上,考虑前期降雨对滑坡发生的重要影响,以百分位法对前期降雨条件进行区分,将前期有效雨量作为阈值模型的前端输入标准,建立动态的降雨致灾临界阈值模型,并利用评价指标对阈值模型过程进行了验证回算。研究结果表明:(1)致灾降雨因子的量级分布较为集中,在致灾过程中存在明显的优势区间,近90%的致灾本场降雨历时和强度分别在48h以下和10 mm·h^(-1)以下,近80%的致灾本场降雨的前15日有效雨量在100 mm以下。(2)前期降雨对本场降雨的影响主要体现在历时上,极端的前期降雨条件下,本场降雨的强度和历时都明显减小,本场降雨过程在致灾过程中不占主导地位。(3)短时降雨过程的致灾平均雨强分布较为离散,降雨过程持续时间越长,平均雨强的离散程度越小。(4)平均雨强-降雨历时-前期有效雨量阈值模型对诱发滑坡的降雨过程有一定的识别能力,准确率和召回率分别达到0.6以上和0.8以上。This paper uses the rainfall observation data of the national encrypted weather station to systematically analyze the disaster-causing rainfall factors of thousands of landslide disasters,and determine the disaster-causing rainfall process by improving the method of dividing the disaster-causing rainfall events.Based on the static threshold model algorithm,considering the important impact of antecedent rainfall on the occurrence of landslides,the percentile method is used to distinguish the antecedent rainfall conditions,and the antecedent effective rainfall is used as the front-end input standard of the threshold model to establish a dynamic rainfall disaster threshold model,using the threshold model to verify and backcalculate the disaster-causing rainfall process of landslide events.The main conclusion is:(1)The magnitude distribution of the disaster-causing rainfall factors is relatively concentrated,and there is an obvious interval in the disaster-causing process.Nearly 90%of the disaster-causing rainfall has a duration of less than 48h and intensity of less than 10 mm·h^(-1).Nearly 80%of the disaster-causing rainfall has a 15 days of effective rainfall of less than 100mm.(2)The impact of the antecedent rainfall on the current rainfall is mainly reflected in the duration.Under the extreme antecedent rainfall conditions,the intensity and duration of the current rainfall are significantly reduced,and the current rainfall process does not dominate the disaster-causing process.(3)The distribution of disaster-causing average rain intensity during short-term rainfall is relatively discrete.The longer the duration of the rainfall process,the smaller the dispersion of average rain intensity.(4)The average rainfall intensity-rainfall duration-antecedent effective rainfall threshold model has a certain ability to identify the rainfall process that induces landslides,the accuracy and recall rates are above 0.6 and 0.8 respectively.

关 键 词：滑坡 降雨指标 阈值模型 预警区 

分 类 号：X43[环境科学与工程—灾害防治] X915.5