基于统计分析的城市道路结冰预报模型研究  

Research on Urban Road Icing Forecast Model Based on Statistical Analysis

作　　者：张兴山[1] 王娟怀[2] 叶培龙[3] 桑文军 李猛[1] 包永虎 ZHANG Xingshan;WANG Juanhuai;YE Peilong;SANG Wenjun;LI Meng;BAO Yonghu(Handan Meteorological Bureau Handan,Hebei 056001;Guangdong Climate Center Guangzhou,Guangdong 510080;Lanzhou Central Meteorological Observatory Lanzhou,Gansu 730020;Civil Aviation Flight University of China Chengdu,Sichuan 618300)

机构地区：[1]河北省邯郸市气象局,河北邯郸056001 [2]广东省气候中心,广东广州510080 [3]兰州中心气象台,甘肃兰州730020 [4]中国民用航空飞行学院,四川成都618300

出　　处：《现代农业研究》2022年第11期110-115,共6页Modern Agriculture Research

基　　金：河北省气象局科研项目(项目编号:18KY31)。

摘　　要：为准确预报冬季道路清扫和喷雾抑尘作业过程中产生的道路结冰情况,保障交通运输及人民生命财产安全,需要构建精度高、实用性强的结冰预报模型。本文利用河北省邯郸1961-2016年国家气象站的常规气象要素数据,采用统计分析、Logistic回归、SVM(支持向量机)分类方法分析邯郸冬季土壤下垫面结冰的气象条件和结冰概率特征,构建邯郸市地面结冰预报模型,并利用2017-2018年当年11月到次年3月的实况数据进行模型检验,显示:1)邯郸市结冰始于当年11月初,结束于次年4月初;临界完全结冰条件为日最低气温≤-1℃或日最低地温≤-5℃,日最低气温≥3℃或日最低地温≥1℃为结冰不发生的条件;2)结冰过渡区域的概率随气温、地温呈S型特征分布。确定性预报中的Lin⁃ear核函数SVM分类器AUC(曲线下面积)达0.982,预报准确率较高;3)在实况检验中结冰概率预报和确定性预报的AUC均接近1,预测效果好,概率性预报略优于确定性预报。这些研究结果可为邯郸市冬季道路清扫和喷雾抑尘作业时间提供一定的科学参考。In order to obtain accurate forecast of the road icing caused by cleaning and dust suppression by spraying in winter,thus ensure the safety in transportation and people's life and property,it is necessary to build a high-precision and practical prediction model of road icing.In this paper,the meteorological conditions and icing probability characteristics of surface icing in winter in Handan city were based on the conventional meteorological data of Handan national meteorological station from 1961 to 2016 by using statistical analysis,Logistic regression and support vector machine(SVM)classification method,and the prediction model of road icing in Handan City was constructed,and then the model was test by using the observation data from 2017 to 2019,the results are as follows:1)The period of road icing in Handan city is from early November of the year to early April of the next year.The critical condition for whole icing test is that the daily minimum temperature is less than-1℃or the daily minimum ground temperature is less than-5℃.On the other hand,there is no ice formation when the daily minimum temperature is more than 3℃or the daily minimum ground temperature is more than 1℃.2)The probability of the icing transition area shows an S-Shaped characteristic distribution with air temperature and ground temperature.There is a higher prediction accuracy for the area under the curve(AUC)of the linear kernel function SVM classifier in deterministic prediction which is 0.982.3)In real-time test,the AUC of icing probability forecast and certainty forecast are close to 1 though the former is slightly better than the latter.These results can provide a scientific reference for the operation time of road cleaning and dust suppression by spraying in winter in Handan.

关 键 词：道路结冰 回归拟合 预报模型 冬季 

分 类 号：U421.4[交通运输工程—道路与铁道工程]