Logo image
A Generalized Aerosol Algorithm for Multi‐Spectral Satellite Measurement With Physics‐Informed Deep Learning Method
Journal article   Open access   Peer reviewed

A Generalized Aerosol Algorithm for Multi‐Spectral Satellite Measurement With Physics‐Informed Deep Learning Method

Jianfang Jiang, Minghui Tao, Xiaoguang Xu, Zhe Jiang, Wenjing Man, Jun Wang, Lunche Wang, Yi Wang, Yalin Zheng, Jinhua Tao, …
Geophysical research letters, Vol.50(24), e2023GL106806
12/28/2023
DOI: 10.1029/2023GL106806
url
https://doi.org/10.1029/2023GL106806View
Published (Version of record) Open Access

Abstract

Abstract The multi‐spectral satellite sensors such as MODIS have a large swath, high spatial resolution, and well onboard calibration, enabling aerosol retrievals with daily global coverage. Despite numerous available bands, MODIS aerosol algorithms over land typically only utilize measurements from 2 to 3 spectral wavelengths to retrieve Aerosol Optical Depth (AOD) based on prescribed aerosol models. To make full use of multi‐spectral measurements and prior information, we developed an aerosol algorithm based on physics‐informed deep learning (PDL) approach. With physical constraint from radiative transfer simulation, PDL can construct model functions between the whole spectral measurements and each retrieved aerosol parameter separately. AERONET validations in eastern China show that MODIS PDL algorithm can accurately retrieve AOD and fine AOD ( R  = 0.936) at 1 km resolution and has reliable performance in coarse AOD as well as notable sensitivity to aerosol absorption. The flexible and efficient PDL method provides a generalized algorithm for common multi‐spectral satellite measurements. Plain Language Summary Atmospheric aerosols play a crucial role in the Earth's radiative energy balance, hydrological cycle, and air quality as well as biogeochemical cycles. Owing to short lifetime and diverse emission sources, aerosol amount and properties vary largely over space and time, making a great challenge in quantifying its climate and environmental effects. With the recognition of aerosols' important role, dedicated satellite instruments have been launched to obtain global aerosol observations since late 1990s. Despite a lack of angular and polarized information, current widely used satellite aerosol products are mainly derived from multi‐spectral measurement due to their daily global coverage and long‐term well‐calibrated observation. However, common algorithms only utilize 2–3 spectral measurements and retrieve one quantitative Aerosol Optical Depth with prescribed aerosol models. By combining the physical constraints from Radiative Transfer simulation and modeling ability of Deep Learning, we developed a generalized physics‐informed DL method that can make full use of multi‐spectral measurements and prior information for aerosol retrievals. Key Points A generalized aerosol algorithm is developed based on physics‐informed deep learning (PDL) method for multi‐spectral satellite measurement MODIS PDL retrieval has not only high accuracy in AOD and fine AOD, but also marked sensitivity to coarse AOD and aerosol absorption PDL method can make full use of the multi‐spectral satellite measurement and prior information with high computational efficiency

Details

Metrics

Logo image