Retrieval of Aerosol Single-Scattering Albedo from MODIS Data Using an Artificial Neural Network
Aerosol single-scattering albedo (SSA) is one of the largest sources of uncertainty in the evaluation of the aerosol radiative forcing effect. The SSA signal, coupled with aerosol optical depth (AOD) and surface reflectance in satellite images, is difficult to retrieve by the look-up table approach....
Published in: | Remote Sensing |
---|---|
Main Authors: | , , |
Format: | Text |
Language: | English |
Published: |
Multidisciplinary Digital Publishing Institute
2022
|
Subjects: | |
Online Access: | https://doi.org/10.3390/rs14246341 |
id |
ftmdpi:oai:mdpi.com:/2072-4292/14/24/6341/ |
---|---|
record_format |
openpolar |
spelling |
ftmdpi:oai:mdpi.com:/2072-4292/14/24/6341/ 2023-08-20T03:59:12+02:00 Retrieval of Aerosol Single-Scattering Albedo from MODIS Data Using an Artificial Neural Network Lin Qi Ronggao Liu Yang Liu agris 2022-12-14 application/pdf https://doi.org/10.3390/rs14246341 EN eng Multidisciplinary Digital Publishing Institute Atmospheric Remote Sensing https://dx.doi.org/10.3390/rs14246341 https://creativecommons.org/licenses/by/4.0/ Remote Sensing; Volume 14; Issue 24; Pages: 6341 remote sensing retrieval thick aerosol single-scattering albedo intelligent computing artificial neural network Text 2022 ftmdpi https://doi.org/10.3390/rs14246341 2023-08-01T07:48:23Z Aerosol single-scattering albedo (SSA) is one of the largest sources of uncertainty in the evaluation of the aerosol radiative forcing effect. The SSA signal, coupled with aerosol optical depth (AOD) and surface reflectance in satellite images, is difficult to retrieve by the look-up table approach. In this study, we proposed an artificial neural network- (ANN) based approach that retrieves SSA over land based on MODIS (moderate resolution imaging spectroradiometer) visible (red band) reflectance variations among nearby pixels that have different surface reflectivities. Using the training dataset generated by the radiative transfer model, the ANN model was trained to establish the relationship among SSA, surface reflectance, and top of atmosphere (TOA) reflectance. Then, based on the trained ANN model, SSA can be retrieved using the surface and apparent reflectance of several heterogeneous pixels. According to sensitivity analysis, this method works well on nonuniform land surfaces with high AODs. The root mean square error (RMSE) of retrieved and measured SSA (from 28 sites of AErosol RObotic NETwork, AERONET) was 0.042, of which the results with an error less than 0.03 accounted for 51%. In addition, the SSA retrieval method was applied to several thick aerosol layer events over different areas (South Asia, South America, and North China Plain) and compared with the ozone monitoring instrument near-UV aerosol data product (OMAERUV). The comparison results of the images show that the retrieval method of visible wavelength proposed in this study has similar outcomes to those from the ultraviolet wavelengths in these regions. The retrieval algorithm we propose provides an effective way to produce an SSA product in visible wavelength and might help to better estimate the aerosol radiative and optical properties over high heterogeneous areas, which is important for the aerosol radiative impact estimate at a regional scale. Text Aerosol Robotic Network MDPI Open Access Publishing Remote Sensing 14 24 6341 |
institution |
Open Polar |
collection |
MDPI Open Access Publishing |
op_collection_id |
ftmdpi |
language |
English |
topic |
remote sensing retrieval thick aerosol single-scattering albedo intelligent computing artificial neural network |
spellingShingle |
remote sensing retrieval thick aerosol single-scattering albedo intelligent computing artificial neural network Lin Qi Ronggao Liu Yang Liu Retrieval of Aerosol Single-Scattering Albedo from MODIS Data Using an Artificial Neural Network |
topic_facet |
remote sensing retrieval thick aerosol single-scattering albedo intelligent computing artificial neural network |
description |
Aerosol single-scattering albedo (SSA) is one of the largest sources of uncertainty in the evaluation of the aerosol radiative forcing effect. The SSA signal, coupled with aerosol optical depth (AOD) and surface reflectance in satellite images, is difficult to retrieve by the look-up table approach. In this study, we proposed an artificial neural network- (ANN) based approach that retrieves SSA over land based on MODIS (moderate resolution imaging spectroradiometer) visible (red band) reflectance variations among nearby pixels that have different surface reflectivities. Using the training dataset generated by the radiative transfer model, the ANN model was trained to establish the relationship among SSA, surface reflectance, and top of atmosphere (TOA) reflectance. Then, based on the trained ANN model, SSA can be retrieved using the surface and apparent reflectance of several heterogeneous pixels. According to sensitivity analysis, this method works well on nonuniform land surfaces with high AODs. The root mean square error (RMSE) of retrieved and measured SSA (from 28 sites of AErosol RObotic NETwork, AERONET) was 0.042, of which the results with an error less than 0.03 accounted for 51%. In addition, the SSA retrieval method was applied to several thick aerosol layer events over different areas (South Asia, South America, and North China Plain) and compared with the ozone monitoring instrument near-UV aerosol data product (OMAERUV). The comparison results of the images show that the retrieval method of visible wavelength proposed in this study has similar outcomes to those from the ultraviolet wavelengths in these regions. The retrieval algorithm we propose provides an effective way to produce an SSA product in visible wavelength and might help to better estimate the aerosol radiative and optical properties over high heterogeneous areas, which is important for the aerosol radiative impact estimate at a regional scale. |
format |
Text |
author |
Lin Qi Ronggao Liu Yang Liu |
author_facet |
Lin Qi Ronggao Liu Yang Liu |
author_sort |
Lin Qi |
title |
Retrieval of Aerosol Single-Scattering Albedo from MODIS Data Using an Artificial Neural Network |
title_short |
Retrieval of Aerosol Single-Scattering Albedo from MODIS Data Using an Artificial Neural Network |
title_full |
Retrieval of Aerosol Single-Scattering Albedo from MODIS Data Using an Artificial Neural Network |
title_fullStr |
Retrieval of Aerosol Single-Scattering Albedo from MODIS Data Using an Artificial Neural Network |
title_full_unstemmed |
Retrieval of Aerosol Single-Scattering Albedo from MODIS Data Using an Artificial Neural Network |
title_sort |
retrieval of aerosol single-scattering albedo from modis data using an artificial neural network |
publisher |
Multidisciplinary Digital Publishing Institute |
publishDate |
2022 |
url |
https://doi.org/10.3390/rs14246341 |
op_coverage |
agris |
genre |
Aerosol Robotic Network |
genre_facet |
Aerosol Robotic Network |
op_source |
Remote Sensing; Volume 14; Issue 24; Pages: 6341 |
op_relation |
Atmospheric Remote Sensing https://dx.doi.org/10.3390/rs14246341 |
op_rights |
https://creativecommons.org/licenses/by/4.0/ |
op_doi |
https://doi.org/10.3390/rs14246341 |
container_title |
Remote Sensing |
container_volume |
14 |
container_issue |
24 |
container_start_page |
6341 |
_version_ |
1774721485411188736 |