摘要
基于2008—2020年青海省的灾情记录,利用灾损指数分析了青海省洪涝灾害的时空分布特征,确定了青海省的洪涝灾害高风险区,同时采用2017—2020年多源融合的CLDAS降水数据,利用机器学习算法建立了洪涝灾害预报模型,确定了致灾雨量阈值,结果表明:(1)青海省洪涝灾害2018年最多,达98次,2014年最少,仅16次,7—8月是洪涝灾害的高风险时段;空间上,基于年平均灾害次数的高风险区为海南州—海西州东部,基于年平均灾损指数的高风险区为海东市—西宁市;(2)利用多种机器学习算法,得到基于CLDAS数据的1、2和24 h雨强是预警灾害的降水因子,海南州—海西州东部1 h或2 h最大雨强达到6.8 mm,或者24 h最大雨强达到11.1 mm,是预警洪涝灾害的降水阈值。海东市—西宁市及邻近地区1 h或2 h最大雨强达到13 mm,或者24 h最大雨强达到18.2 mm,是预警洪涝灾害的降水阈值。
Based on the disaster records during 2008-2020 in Qinghai province,the disaster index is used to study the spatial and temporal distribution characteristics of flood disasters,and high-risk areas are identified.Additionally,a flood disaster prediction model is constructed by machine learning algorithms utilizing the multi-source fusion CLDAS precipitation data from 2017 to 2020,and the disaster-causing rainfall threshold of high-risk area is calculated.The findings indicate that:(1)The highest number of flood disasters,sum to 98,occurred in 2018,while the lowest number,sum to 16,occurred in 2014.The most devastating floods occur in July and August.Based on the annual mean number of disasters in Qinghai,Hainan-eastern Haixi prefecture is the high-risk area of flood disasters,and based on the annual average disaster index,Haidong-Xining is another high-risk area.(2)The 1,2,and 24 h rain intensity of CLDAS data are significant parameters for disasters prediction using a variety of machine learning techniques.The precipitation threshold of Hainan and the eastern part of Haixi prefecture is that the maximum rain intensity of 1 h or 2 h reaches 6.8 mm,or that of 24 h reaches 11.1 mm,while the threshold of Haidong-Xining and nearby areas is that the maximum rain intensity of 1 h or 2 h reaches 13 mm,or that of 24 h reaches 18.2 mm.
作者
朱科旭
管琴
白爱娟
ZHU Kexu;GUAN Qin;BAI Aijuan(School of Earth and Space Sciences,University of Science and Technology of China,Hefei 230026,China;School of Atmospheric Sciences,Chengdu University of Information Technology,Chengdu 210025,China;Qinghai Meteorological Obversatory,Xining 810000,China;School of Earth Sciences,Yunnan University,Kunming 650500,China)
出处
《沙漠与绿洲气象》
2024年第1期81-88,共8页
Desert and Oasis Meteorology
基金
自然科学基金气象联合基金项目(U2242202)
青海省科技计划应用基础研究项目(2020-ZJ-739)
成都信息工程大学教师科技创新能力提升计划重大项目(KYTD202201)。
关键词
青海省
洪涝灾害
CLDAS降水数据
机器学习
致灾雨量
Qinghai province
flood disasters
CLDAS precipitation data
machine learning
disaster-causing rainfall
