Enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimation

Satellite-based data can provide continuous aerosol observations but suffer from significant uncertainties across various regions. Transfer learning improves model generalization, yet its application in atmospheric research remains limited. Here, we introduce an innovative framework for retrieving g...

Full description

Saved in:
Bibliographic Details
Main Authors: Xiaohu Sun, Yong Xue, Lin Sun
Format: Article
Language:English
Published: Elsevier 2025-05-01
Series:International Journal of Applied Earth Observations and Geoinformation
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S1569843225001815
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850135604504821760
author Xiaohu Sun
Yong Xue
Lin Sun
author_facet Xiaohu Sun
Yong Xue
Lin Sun
author_sort Xiaohu Sun
collection DOAJ
description Satellite-based data can provide continuous aerosol observations but suffer from significant uncertainties across various regions. Transfer learning improves model generalization, yet its application in atmospheric research remains limited. Here, we introduce an innovative framework for retrieving global aerosol optical depth (AOD) which named the Aerosol domain-Adaptive Network (AAdaN). The framework utilizes a neural network to estimate mutual information, and aligns spatial covariate shift via a transfer loss term. Then, we assess the retrieval potential in unknown scenarios using independent land cover type, and the proposed model demonstrates satisfactory results. The cross-validation shows strong agreement with in-situ measurements, both in sample-based and site-based evaluations. Specifically, the site-based ten-fold cross-validation of our AOD retrievals indicate that all accuracy metrics are satisfactory, with a Pearson correlation of 0.766 and a Root-Mean-Square Error of 0.118, and that about 76.05 % of the retrievals meet the expected error criteria [±(0.05 + 20 %)]. Additionally, the proposed AAdaN achieves stable, high-accuracy aerosol retrievals across various surface and atmospheric conditions, and can generate spatially continuous AOD distributions. This study significantly improves spatial generalization and offers valuable insights for future model development.
format Article
id doaj-art-03d1aa224c134c4ba4ccd3d41b684baa
institution OA Journals
issn 1569-8432
language English
publishDate 2025-05-01
publisher Elsevier
record_format Article
series International Journal of Applied Earth Observations and Geoinformation
spelling doaj-art-03d1aa224c134c4ba4ccd3d41b684baa2025-08-20T02:31:22ZengElsevierInternational Journal of Applied Earth Observations and Geoinformation1569-84322025-05-0113910453410.1016/j.jag.2025.104534Enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimationXiaohu Sun0Yong Xue1Lin Sun2School of Atmospheric Physics, Nanjing University of Information Science & Technology, 210044 Nanjing, ChinaSchool of Emergency Management, Nanjing University of Information Science & Technology, Nanjing 210044, China; Corresponding author.College of Geodesy and Geomatics, Shandong University of Science and Technology, Qingdao, Shandong 266590, ChinaSatellite-based data can provide continuous aerosol observations but suffer from significant uncertainties across various regions. Transfer learning improves model generalization, yet its application in atmospheric research remains limited. Here, we introduce an innovative framework for retrieving global aerosol optical depth (AOD) which named the Aerosol domain-Adaptive Network (AAdaN). The framework utilizes a neural network to estimate mutual information, and aligns spatial covariate shift via a transfer loss term. Then, we assess the retrieval potential in unknown scenarios using independent land cover type, and the proposed model demonstrates satisfactory results. The cross-validation shows strong agreement with in-situ measurements, both in sample-based and site-based evaluations. Specifically, the site-based ten-fold cross-validation of our AOD retrievals indicate that all accuracy metrics are satisfactory, with a Pearson correlation of 0.766 and a Root-Mean-Square Error of 0.118, and that about 76.05 % of the retrievals meet the expected error criteria [±(0.05 + 20 %)]. Additionally, the proposed AAdaN achieves stable, high-accuracy aerosol retrievals across various surface and atmospheric conditions, and can generate spatially continuous AOD distributions. This study significantly improves spatial generalization and offers valuable insights for future model development.http://www.sciencedirect.com/science/article/pii/S1569843225001815Aerosol optical depthTransfer learningMutual informationModerate resolution imaging spectroradiometer
spellingShingle Xiaohu Sun
Yong Xue
Lin Sun
Enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimation
International Journal of Applied Earth Observations and Geoinformation
Aerosol optical depth
Transfer learning
Mutual information
Moderate resolution imaging spectroradiometer
title Enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimation
title_full Enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimation
title_fullStr Enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimation
title_full_unstemmed Enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimation
title_short Enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimation
title_sort enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimation
topic Aerosol optical depth
Transfer learning
Mutual information
Moderate resolution imaging spectroradiometer
url http://www.sciencedirect.com/science/article/pii/S1569843225001815
work_keys_str_mv AT xiaohusun enhancingglobalaerosolretrievalfromsatellitedataviadeeplearningwithmutualinformationestimation
AT yongxue enhancingglobalaerosolretrievalfromsatellitedataviadeeplearningwithmutualinformationestimation
AT linsun enhancingglobalaerosolretrievalfromsatellitedataviadeeplearningwithmutualinformationestimation