摘要
准确掌控区域二氧化氮(NO_(2))浓度时空分布,对于区域NO_(2)污染防控措施的制定具有重要意义。空气质量监测站点的稀疏且空间分布不均为计算全面域的NO_(2)带来了较大挑战,尤其对于监测站点数量较少的城市。为全面把控四川省宜宾市近年来NO_(2)时空变化特征,研究使用TROPOMI卫星遥感数据,采用基于机器学习的多重插补链式方程(MICE)克服原始观测数据稀疏不均的问题,重构了2019~2021年宜宾市1 km网格NO_(2)小时浓度。基于站点的留出法验证中R2和RMSE分别为0.67和8.4μg/m^(3)。宜宾市2019~2021年各年的人口加权NO_(2)浓度([NO_(2)]pw)分别为19.1±5.5、14.9±5.3和14.8±6.2μg/m^(3)。多年的季均[NO_(2)]pw在冬季最高,其次依次为秋季、春季及夏季。NO_(2)浓度在8:00~10:00和18:00~23:00点呈现出上升趋势,一般夜间NO_(2)污染比白天更严重。翠屏区主城区是宜宾市最主要的NO_(2)污染区域,岷江及长江流域的NO_(2)浓度也较高。在2020年应对COVID-19的全面封控期间,宜宾NO_(2)浓度大幅度降低,[NO_(2)]pw相比2019年降低了约25%。精确的高分辨率NO_(2)时空分布结果可为当地减排策略的制定提供时量化的数据支撑。
Accurate understanding of the spatiotemporal distribution of ambient NO_(2) is important for implementing air pollution prevention and control measures.However,the sparse and uneven distribution of air quality monitoring stations pose a great challenges to estimate the full-coverage of NO_(2),especially for cities with a small number of monitoring stations.In order to comprehensively understand the spatiotemporal variation of NO_(2) in Yibin,Sichuan Province in recent years,TROPOMI satellite remote sensing data was taken to reconstruct the 1 km grid NO_(2) hourly concentrations in Yibin from 2019 to 2021,taking advantage of machine learning-based multiple interpolation chain equations(MICE)to overcome the sparsity and unevenness of the original observation data.The R~2 and RMSE were 0.67 and 8.4μg/m^(3),respectively,in the site-based holdout-validation.The population-weighted NO_(2)([NO_(2)]_(pw))in Yibin during 2019-2021 were 19.1±5.5,14.9±5.3 and 14.8±6.2μg/m^(3),respectively.The[NO_(2)]_(pw)was the highest in winter,followed by fall,spring,and summer.The hourly NO_(2) concentration showed an upward trend during 8:00-10:00 a.m.and 6:00-11:00 p.m.and generally NO_(2) pollution was generally more serious at night than during the day.The main urban area of Cuiping District was the major NO_(2)-polluted area in Yibin.Also,NO_(2) concentrations were high in the Minjiang and Yangtze River basins.During the COVID-19 lockdown period in 2020,the NO_(2) levels in Yibin decreased significantly a[NO_(2)]_(pw)decrease by approximately 25%compared to the corresponding period in 2019.Accurate and high-resolution spatiotemporal distributions of NO_(2) could provide temporal and quantitative data support for implementing emission reduction strategies.
作者
朱瑢昕
米潭
赵贤杰
蒋霞
江水苹
李文秀
周书华
杨复沫
詹宇
ZHU Rongxin;MI Tan;ZHAO Xianjie;JIANG Xia;JIANG Shuiping;LI Wenxiu;ZHOU Shuhua;YANG Fumo;ZHAN Yu(College of Architecture and Environment,Sichuan University;Yibin Ecological Environment Bureau;Yibin Environmental Monitoring Center Station)
出处
《地球与环境》
CAS
CSCD
北大核心
2024年第2期176-187,共12页
Earth and Environment
基金
国家自然科学基金项目(22076129)。
关键词
二氧化氮
卫星遥感
机器学习
时空分布
宜宾市
NO_(2)
satellite remote sensing
machine learning
spatiotemporal distribution
Yibin
