基于回归模型的青藏铁路水害气象风险评估  

Meteorological Risk Assessment of Qinghai-Tibet Railway Water Disaster Based on Rregression Model

作　　者：张静 保广裕 周丹 马守存 刘玮 

机构地区：[1]青海省气象服务中心,青海西宁810001

出　　处：《沙漠与绿洲气象》2018年第1期53-60,共8页Desert and Oasis Meteorology

基　　金：2015年青海省气象局重点项目"青海铁路水害气象风险评估关键技术"资助

摘　　要：应用逻辑回归和逐步回归方法对青藏铁路青海段由降水引发铁路水害气象风险评估进行研究,以青藏铁路公司提供的铁路沿线水害统计资料为依据,选取2000—2014年青藏铁路沿线青海境内14个气象站点和28个加密气象站降水资料,开展降水引发的水害气象风险评估,结果表明:青藏铁路水害的发生不仅与降水量级有关,还与降水的性质和持续时间有关,铁路水害主要集中在7—8月,高发路段是乐都-平安、湟源-海晏段。导致青藏铁路沿线水害的主要降水类型为区域性强降水和连阴雨。考虑到降水持续时间对水害的影响,通过计算得出的铁路水害诱发指数和水害有效降水因子,且距水害发生时间愈近,日降水量对水害发生的作用就愈大。采用逻辑回归模型判断青藏铁路青海境内各区段水害的发生有无,其预测模型总判对率均超过86.2%,其次运用逐步回归评估模型计算出水害预报等级,经检验对最高级别1级的预报准确率达88.9%。可见,逻辑回归模型和逐步回归评估模型在青藏铁路青海境内铁路水害预报和评估工作中具有较高的预报准确率。Using logistic regression model and stepwise regression model,the research is on meteorological risk assessment of water damage caused by precipitation along Qinghai-T ibet railway in Qinghai Province.Based on the data of water damage along the railway line provided by Qinghai- Tibet Railway Company, selecting the precipitation data of 14 meteorological stations and 28 encrypted weather stations in Qinghai Province from 2000 to 2014 along Qinghai-T ibet railway for the study of meteorological r is k assessment o f water damage caused by p re c ipitation, the main conclusions are as fol lows： the occurrence of water damage is not only in connection with precipitation intens ity, but also in connection w ith character and duration of ra in fa l l. Water-damage risk is mainly concentrated in July to August and the hazardous sections are Ledu -Pingan and Huangyuan-Haiyan.The main types o f precipitation that caused water damage along the Qinghai- Tibet railway are regional heavy rain and continuous rain.Considering the influence o f the heavy rain duration to water damage, index o f water damage and the effective index o f water damage were calculated.And the closer since water damage occurs, the influence becomes heavier that daily precipitation have on water damage.Using logistic regression model to determine the occurrence of the water damage in Qinghai Railway, and the accuracy o f the prediction model is more than 86.2%; then using stepwise regression assessment model to calculate the water damage forecast level and the forecast accuracy to the highest level is 88.9% .It shows that logistic regression model and stepwise regression assessment model have high forecast accuracy in the forecasting and meteorological r is k assessment o f water damage along Qinghai-Tibet railway in Qinghai Province.

关 键 词：青藏铁路 水害 风险评估 回归模型 

分 类 号：P468[天文地球—大气科学及气象学] TV125[水利工程—水文学及水资源]