摘要
春节人口迁徙是中国一年一度在全国范围内的人口大型迁徙活动。基于空间全生命周期流行病学理论框架,集成多源时空数据和机器学习算法实现日尺度的PM2.5高时空分辨率制图,结合腾讯位置大数据评估长三角区域人群PM2.5污染短期暴露健康风险。结果如下:(1)基于随机森林构建的PM2.5高时空分辨率制图模型空间交叉验证结果R2达到0.8以上,具有良好的精度;(2)春节人口迁徙行为使得全国范围内人口在短期内大规模流动,导致长三角地区PM2.5短期暴露健康风险增加和减少的迁徙人口分别有6070万人和6175万人;(3)春节人口迁徙行为导致的PM2.5污染短期暴露健康风险危害不容忽视,增加风险均值在0.25—0.39,高值在0.9以上;(4)PM2.5污染短期暴露健康效应具有强烈的时空异质性,两天内的健康风险变化极大(0.84,-0.75);(5)在长三角地区的部分城市中,大部分迁徙人口的春节迁徙行为导致PM2.5污染短期暴露健康风险增加,如上海和苏州,暴露人口约有1320万人和600万人。以期为PM2.5污染短期暴露健康效应估算提供实证分析,丰富了空间全生命周期流行病学的应用案例。
Human mobility during the Spring Festival is a large-scale annual human mobility across China.This paper evaluated the human health risks of short-term exposure to PM2.5 in the Yangtze River Delta through a combination of the daily-level high spatiotemporal mapping of PM2.5 based on integrated multi-sources spatiotemporal data and machine learning algorithm,and Tencent mobility data based on the theoretical framework of spatial life-course epidemiology.The results showed that:(1)the accuracy of the random forest model for PM2.5 spatiotemporal mapping by spatial cross-validation is more than 0.8.(2)Human mobility behavior enables large-scale mobility across the country in a short period during the Spring Festival,which leads to 60.7 million and 61.75 million populations who have increased and decreased the health risk of short-term exposure to PM2.5 respectively.(3)We should not ignore the health risks of short-term exposure to PM2.5 caused by human mobility behavior during the Spring Festival,because the estimated average increased risks are from 0.25 to 0.39.The highest risk is more than 0.9.(4)This reveals strong spatiotemporal heterogeneity in the health effect of short-term exposure to PM2.5.Even the value of risks could change from 0.84 to-0.75 in two days.(5)Most of the travel populations in the Yangtze River Delta have increased the health risk of short-term exposure to PM2.5 caused by human mobility during the Spring Festival,such as Shanghai and Suzhou,with the exposed populations reaching to nearly 13.2 million and 6 million.This paper provides a practice case study to evaluate the health risk of short-term exposure to PM2.5 and enriches the application of spatial life-course epidemiology.
作者
戴劭勍
李佳佳
杨维旭
陈方煜
江辉仙
DAI Shaoqing;LI Jiajia;YANG Weixu;CHEN Fangyu;JIANG Huixian
出处
《上海城市规划》
2020年第5期22-29,共8页
Shanghai Urban Planning Review
基金
福建省自然科学基金项目“基于移动终端的大型公共建筑物智能消防疏散系统关键技术研究”(编号2018J01740)资助。
关键词
PM2.5短期污染暴露
春节人口迁徙
时空制图
机器学习
short-term exposure to PM2.5 pollution
human mobility during the Spring Festival
spatiotemporal mapping
machine learning