气候变化领域集成服务门户(CCPortal): Estimating daily high-resolution PM2.5 concentrations over Texas: Machine Learning approach

CCPortal

DOI	10.1016/j.atmosenv.2021.118209
	Estimating daily high-resolution PM2.5 concentrations over Texas: Machine Learning approach
	Ghahremanloo M.; Choi Y.; Sayeed A.; Salman A.K.; Pan S.; Amani M.
发表日期	2021
ISSN	1352-2310
卷号	247
英文摘要	PM2.5 is an important atmospheric constituent associated to human health. Therefore, the capability of estimating PM2.5 concentrations at high spatiotemporal resolutions, particularly in places with no ground stations, would be invaluable. Although several studies have involved the estimation of PM2.5, few have estimated PM2.5 concentrations at high spatial resolutions. In this study, we leverage the aerosol optical depth (AOD) and random forest (RF) algorithm to estimate daily 1-km PM2.5 concentrations over Texas from 2014 to 2018. For this purpose, we use collection 6 Multi-Angle Implementation of Atmospheric Correction (MAIAC) AOD products from the Moderate Resolution Imaging Spectroradiometer (MODIS). To address the different sources and speciation of PM2.5 over Texas, we use several important control parameters. As a result, the accuracy of RF remains consistent throughout the study area. After estimating ground-level PM2.5 levels, we apply a ten-fold cross-validation approach to obtain a correlation coefficient (R) of 0.83–0.90 and a mean absolute bias (MAB) of 1.47–1.77 μg/m3. Our results show that RF is highly capable of estimating ground-level PM2.5 concentrations. In addition to the RF model, we also compare the capability of commonly used models, including multiple linear regression (MLR) and mixed effects model (MEM), for estimating the PM2.5 concentrations of global regions. Results indicate that RF, compared to the other models, has the highest accuracy, MEM the second-highest, and MLR the third. We also leverage the USEPA Environmental Benefits Mapping and Analysis Program Community Edition (BenMAP-CE) to estimate the impact of changes in PM2.5 levels on the number of respiratory-related premature mortalities in Texas in 2014–2018. Considering 2014 as the baseline year, the BenMAP analyses reveal that PM2.5 reductions could have prevented a large number of premature mortalities, particularly among adults aged 25–99, from 2014 to 2018 in Texas. © 2021 Elsevier Ltd
关键词	BenMAP High-resolution PM2.5 Machine learning MODIS AOD Remote sensing Texas
语种	英语
scopus关键词	Decision trees; Linear regression; Machine learning; Atmospheric constituents; Correlation coefficient; High spatial resolution; Machine learning approaches; Moderate resolution imaging spectroradiometer; Multi-angle implementation of atmospheric corrections; Multiple linear regressions; Spatio-temporal resolution; Radiometers; concentration (composition); estimation method; ground-based measurement; machine learning; MODIS; mortality; optical depth; particulate matter; pollution effect; respiratory disease; spatial resolution; spatiotemporal analysis; adult; aerosol; aged; article; controlled study; correlation coefficient; cross validation; human; optical depth; particulate matter 2.5; premature mortality; random forest; remote sensing; species differentiation; Texas; Texas; United States
来源期刊	ATMOSPHERIC ENVIRONMENT
文献类型	期刊论文
条目标识符	http://gcip.llas.ac.cn/handle/2XKMVOVA/248597
作者单位	Department of Earth and Atmospheric Sciences, University of Houston, Houston, TX 77004, United States; School of Atmospheric Physics, Nanjing University of Information Science and Technology (NUIST), Nanjing, 210044, China; Wood Environment & Infrastructure Solutions, Ottawa, ON K2E7L5, Canada
推荐引用方式 GB/T 7714	Ghahremanloo M.,Choi Y.,Sayeed A.,et al. Estimating daily high-resolution PM2.5 concentrations over Texas: Machine Learning approach[J],2021,247.
APA	Ghahremanloo M.,Choi Y.,Sayeed A.,Salman A.K.,Pan S.,&Amani M..(2021).Estimating daily high-resolution PM2.5 concentrations over Texas: Machine Learning approach.ATMOSPHERIC ENVIRONMENT,247.
MLA	Ghahremanloo M.,et al."Estimating daily high-resolution PM2.5 concentrations over Texas: Machine Learning approach".ATMOSPHERIC ENVIRONMENT 247(2021).