气象科学2026，Vol.46Issue(4)：376-384,9.DOI:10.12306/2025jms.0049
基于交通流量控制下的交通气象风险等级研究
Research on the classification of traffic-meteorological risk levels controlling for traffic flow
摘要
Abstract
Urban traffic flow is a key confounding variable in research on traffic meteorological risks.Its significant daily fluctuations and difficulties in precise estimation obscure the independent risk contributions of meteorological factors.This study utilized meteorological and traffic accident data from the urban area of Wenzhou from 2019 to 2023.Factor analysis was employed to eliminate multicollinearity among various meteorological elements.A Logistic Regression model(LR)was constructed using samples from weekdays,which exhibit stable commuting patterns,to control for variability in traffic flow.The model's output of high accident probability was converted into three traffic meteorological risk levels(low,medium,and high),and this framework was applied to the entire year.To validate the effectiveness of this approach,a comparative full-time model(incorporating both weekdays and non-weekdays)without traffic flow control was simultaneously developed.Results indicate seasonal differences in the impact of meteorological factors on traffic accidents in Wenzhou's urban area.During the flood season,temperature and precipitation factors contained elements directly linked to accident mechanisms,whereas in the non-flood season,the relationship between meteorological factors and accidents was predominantly more complex and indirect.In addition,the weekday model demonstrated superior goodness-of-fit and test performance compared to the full-time model without traffic flow control.Specifically,the test accuracy improved by 1.6%during the flood season and 9.6%during the non-flood season.By selecting weekday samples to control for the confounding effect of traffic flow,this study enhances the model's predictive capability while effectively quantifying the independent risk of meteorological factors in traffic accidents.This approach helps focus traffic safety assurance efforts on the causal influence of meteorological factors themselves and provides a more scientific and precise decision-making basis for public travel and traffic management.关键词
气象风险/交通流量/混杂变量控制/逻辑回归模型Key words
meteorological risk/traffic flow/confounding effect control/logistic regression model分类
天文与地球科学引用本文复制引用
谷铁城,郑峰,吴贤笃,杨烨坤,郑裕,范亮..基于交通流量控制下的交通气象风险等级研究[J].气象科学,2026,46(4):376-384,9.基金项目
浙江省气象科技计划项目重点项目(2021ZD26) （2021ZD26）
