DeepAir: deep learning and satellite imagery to estimate high-resolution PM2.5 at scale

Air pollution, specifically PM _2.5 , has become a significant global concern owing to its detrimental impacts on public health. Even so, the high-resolution monitoring of air pollution is still a challenge on a global scale. To cope with this, machine learning (ML) techniques have been utilized to...

Full description

Saved in:
Bibliographic Details
Main Authors: Wenxuan Guo, Zhaoping Hu, Ling Jin, Yanyan Xu, Marta C Gonzalez
Format: Article
Language:English
Published: IOP Publishing 2025-01-01
Series:Machine Learning: Science and Technology
Subjects:
Online Access:https://doi.org/10.1088/2632-2153/adb67a
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850027749697126400
author Wenxuan Guo
Zhaoping Hu
Ling Jin
Yanyan Xu
Marta C Gonzalez
author_facet Wenxuan Guo
Zhaoping Hu
Ling Jin
Yanyan Xu
Marta C Gonzalez
author_sort Wenxuan Guo
collection DOAJ
description Air pollution, specifically PM _2.5 , has become a significant global concern owing to its detrimental impacts on public health. Even so, the high-resolution monitoring of air pollution is still a challenge on a global scale. To cope with this, machine learning (ML) techniques have been utilized to infer the concentration of air pollutants at a fine scale. In this study, we propose DeepAir , a learning framework for estimating PM _2.5 concentrations at a fine scale with sparsely distributed observations. DeepAir integrates a pre-trained convolutional neural network with the LightGBM method. This framework estimates the PM _2.5 concentration of a given patch, utilizing a synergy of geographical information, meteorological conditions, and satellite observations. We select California as the focal region and train the model with data from 2014 to 2017 provided by 130 PM _2.5 observation stations in the state. Upon training, the model can be applied to estimate the daily PM _2.5 concentrations at 1 km resolution across California. Our methodology meticulously incorporates meteorological variables, with a particular emphasis on wildfire propagation, and contemplates the complex interplay of various features. To ascertain the efficacy of our model, we employ the 10-fold cross-validation technique, which confirms that our model surpasses traditional ML and standalone deep learning methods.
format Article
id doaj-art-a4ba3cdcb2964edb8d3bfb3e670440f8
institution DOAJ
issn 2632-2153
language English
publishDate 2025-01-01
publisher IOP Publishing
record_format Article
series Machine Learning: Science and Technology
spelling doaj-art-a4ba3cdcb2964edb8d3bfb3e670440f82025-08-20T03:00:03ZengIOP PublishingMachine Learning: Science and Technology2632-21532025-01-016101505710.1088/2632-2153/adb67aDeepAir: deep learning and satellite imagery to estimate high-resolution PM2.5 at scaleWenxuan Guo0https://orcid.org/0000-0001-6336-3819Zhaoping Hu1Ling Jin2Yanyan Xu3https://orcid.org/0000-0001-5429-3177Marta C Gonzalez4MoE Key Laboratory of Artificial Intelligence, AI Institute, Shanghai Jiao Tong University , Shanghai 200240, People’s Republic of ChinaMoE Key Laboratory of Artificial Intelligence, AI Institute, Shanghai Jiao Tong University , Shanghai 200240, People’s Republic of ChinaEnergy Technologies Area , Lawrence Berkeley National Laboratory, Berkeley, CA 94720, United States of AmericaMoE Key Laboratory of Artificial Intelligence, AI Institute, Shanghai Jiao Tong University , Shanghai 200240, People’s Republic of China; Energy Technologies Area , Lawrence Berkeley National Laboratory, Berkeley, CA 94720, United States of America; Department of City and Regional Planning, University of California , Berkeley, CA 94720, United States of AmericaEnergy Technologies Area , Lawrence Berkeley National Laboratory, Berkeley, CA 94720, United States of America; Department of City and Regional Planning, University of California , Berkeley, CA 94720, United States of America; Department of Civil and Environmental Engineering, University of California , Berkeley, CA 94720, United States of AmericaAir pollution, specifically PM _2.5 , has become a significant global concern owing to its detrimental impacts on public health. Even so, the high-resolution monitoring of air pollution is still a challenge on a global scale. To cope with this, machine learning (ML) techniques have been utilized to infer the concentration of air pollutants at a fine scale. In this study, we propose DeepAir , a learning framework for estimating PM _2.5 concentrations at a fine scale with sparsely distributed observations. DeepAir integrates a pre-trained convolutional neural network with the LightGBM method. This framework estimates the PM _2.5 concentration of a given patch, utilizing a synergy of geographical information, meteorological conditions, and satellite observations. We select California as the focal region and train the model with data from 2014 to 2017 provided by 130 PM _2.5 observation stations in the state. Upon training, the model can be applied to estimate the daily PM _2.5 concentrations at 1 km resolution across California. Our methodology meticulously incorporates meteorological variables, with a particular emphasis on wildfire propagation, and contemplates the complex interplay of various features. To ascertain the efficacy of our model, we employ the 10-fold cross-validation technique, which confirms that our model surpasses traditional ML and standalone deep learning methods.https://doi.org/10.1088/2632-2153/adb67ahigh resolution PM2.5spatial interpolationsatellite imagerydeep learning
spellingShingle Wenxuan Guo
Zhaoping Hu
Ling Jin
Yanyan Xu
Marta C Gonzalez
DeepAir: deep learning and satellite imagery to estimate high-resolution PM2.5 at scale
Machine Learning: Science and Technology
high resolution PM2.5
spatial interpolation
satellite imagery
deep learning
title DeepAir: deep learning and satellite imagery to estimate high-resolution PM2.5 at scale
title_full DeepAir: deep learning and satellite imagery to estimate high-resolution PM2.5 at scale
title_fullStr DeepAir: deep learning and satellite imagery to estimate high-resolution PM2.5 at scale
title_full_unstemmed DeepAir: deep learning and satellite imagery to estimate high-resolution PM2.5 at scale
title_short DeepAir: deep learning and satellite imagery to estimate high-resolution PM2.5 at scale
title_sort deepair deep learning and satellite imagery to estimate high resolution pm2 5 at scale
topic high resolution PM2.5
spatial interpolation
satellite imagery
deep learning
url https://doi.org/10.1088/2632-2153/adb67a
work_keys_str_mv AT wenxuanguo deepairdeeplearningandsatelliteimagerytoestimatehighresolutionpm25atscale
AT zhaopinghu deepairdeeplearningandsatelliteimagerytoestimatehighresolutionpm25atscale
AT lingjin deepairdeeplearningandsatelliteimagerytoestimatehighresolutionpm25atscale
AT yanyanxu deepairdeeplearningandsatelliteimagerytoestimatehighresolutionpm25atscale
AT martacgonzalez deepairdeeplearningandsatelliteimagerytoestimatehighresolutionpm25atscale