First Aerosol Retrieval From PACE HARP2 Over Land With Physics-Informed Deep Learning Method

Wenjing Man; Minghui Tao; Xiaoguang Xu; Jun Wang; Shijie Zhou; Ziyue Tang; Yi Wang; Lunche Wang; Jhoon Kim; Liangfu Chen

doi:10.1109/TGRS.2026.3702267

Back

First Aerosol Retrieval From PACE HARP2 Over Land With Physics-Informed Deep Learning Method

Journal article

Peer reviewed

First Aerosol Retrieval From PACE HARP2 Over Land With Physics-Informed Deep Learning Method

Wenjing Man, Minghui Tao, Xiaoguang Xu, Jun Wang, Shijie Zhou, Ziyue Tang, Yi Wang, Lunche Wang, Jhoon Kim and Liangfu Chen

IEEE transactions on geoscience and remote sensing, Vol.64, 4108510

2026

DOI: 10.1109/TGRS.2026.3702267

View Online

Abstract

The advanced Hyperangular Rainbow Polarimeter (HARP2) onboard NASA's Plankton, Aerosol, Cloud, ocean Ecosystem (PACE) satellite launched in February 2024, provides an enhanced detection of global aerosols with wide-angle imaging and polarization measurements. However, there are still no HARP2 aerosol products over land due to the lack of an efficient retrieval algorithm. In this study, we develop a HARP2 aerosol retrieval algorithm over land based on a physics-informed deep learning (PDL) method. By making full use of HARP2 measurements and prior information, an efficient and reliable retrieval of aerosol optical/microphysical parameters is achieved with the PDL algorithm over China during 2024. Ground validations show very high correlations of 0.935 and 0.943, and root-mean-square error (RMSE) values of 0.151 and 0.120 between HARP2 PDL Aerosol Optical Depth (AOD) and fine mode AOD and AErosol RObotic Network (AERONET) products. Also, HARP2 PDL retrievals of coarse mode AOD and spectral single scattering albedo (SSA) have high consistency with AERONET inversions (R = 0.696 and 0.632). HARP2 retrievals effectively capture variations of particle size and absorption in typical dust and fire events. With high efficiency and robust accuracy, the PDL algorithm can support operational retrieval of global HARP2 aerosol products over land.

Algorithms

Remote Sensing

Aerosol

Aerosol retrieval

Aerosols

Cloud

deep belief network (DBN)

Educational institutions

Grounding

Measurement

Modeling

ocean Ecosystem (PACE) Hyperangular Rainbow Polarimeter (HARP2)

physics-informed deep learning (PDL)

Plankton

Polarization

Satellites

Surfaces

Details

Title: Subtitle: First Aerosol Retrieval From PACE HARP2 Over Land With Physics-Informed Deep Learning Method
Creators: Wenjing Man - China University of Geosciences
Minghui Tao - China University of Geosciences
Xiaoguang Xu - University of Maryland, Baltimore County
Jun Wang - University of Iowa
Shijie Zhou - China University of Geosciences
Ziyue Tang - China University of Geosciences
Yi Wang - China University of Geosciences
Lunche Wang - China University of Geosciences
Jhoon Kim - Yonsei University
Liangfu Chen - State Key Laboratory of Remote Sensing Science
Resource Type: Journal article
Publication Details: IEEE transactions on geoscience and remote sensing, Vol.64, 4108510
DOI: 10.1109/TGRS.2026.3702267
ISSN: 0196-2892
eISSN: 1558-0644
Publisher: IEEE
Number of pages: 1
Grant note: 42271382; 41830109 / National Natural Science Foundation of China (10.13039/501100001809)
Language: English
Date published: 2026
Academic Unit: Electrical and Computer Engineering; Civil and Environmental Engineering; Physics and Astronomy; Chemical and Biochemical Engineering
Record Identifier: 9985175436202771

Metrics

3 Record Views