Aerosol Optical Depth Retrieval for Sentinel-2 Based on Convolutional Neural Network Method
Atmospheric aerosol significantly affects the climate environment and public health, and Aerosol Optical Depth (AOD) is a fundamental optical characteristic parameter of aerosols, so it is important to develop methods for obtaining AOD. In this work, a novel AOD retrieval algorithm based on a Convol...
Published in: | Atmosphere |
---|---|
Main Authors: | , , |
Format: | Text |
Language: | English |
Published: |
Multidisciplinary Digital Publishing Institute
2023
|
Subjects: | |
Online Access: | https://doi.org/10.3390/atmos14091400 |
id |
ftmdpi:oai:mdpi.com:/2073-4433/14/9/1400/ |
---|---|
record_format |
openpolar |
spelling |
ftmdpi:oai:mdpi.com:/2073-4433/14/9/1400/ 2023-10-09T21:44:12+02:00 Aerosol Optical Depth Retrieval for Sentinel-2 Based on Convolutional Neural Network Method Jie Jiang Jiaxin Liu Donglai Jiao agris 2023-09-05 application/pdf https://doi.org/10.3390/atmos14091400 eng eng Multidisciplinary Digital Publishing Institute Aerosols https://dx.doi.org/10.3390/atmos14091400 https://creativecommons.org/licenses/by/4.0/ Atmosphere Volume 14 Issue 9 Pages: 1400 AOD CNN Sentinel-2 deep learning Text 2023 ftmdpi https://doi.org/10.3390/atmos14091400 2023-09-10T23:54:15Z Atmospheric aerosol significantly affects the climate environment and public health, and Aerosol Optical Depth (AOD) is a fundamental optical characteristic parameter of aerosols, so it is important to develop methods for obtaining AOD. In this work, a novel AOD retrieval algorithm based on a Convolutional Neural Network (CNN) method that could provide continuous and detailed aerosol distribution is proposed. The algorithm utilizes data from Sentinel-2 and Aerosol Robotic Network (AERONET) spanning from 2016 to 2022. The CNN AOD data are consistent with the AERONET measurements, with an R2 of 0.95 and RMSE of 0.049 on the test dataset. CNN demonstrates superior performance in retrieving AOD compared with other algorithms. CNN retrieves AOD well on high reflectance surfaces, such as urban and bare soil, with RMSEs of 0.051 and 0.042, respectively. CNN efficiently retrieves AOD in different seasons, but it performs better in summer and winter than in spring and autumn. In addition, to study the relationship between image size and model retrieval performance, image datasets of 32 × 32, 64 × 64 and 128 × 128 pixels were created to train and test the CNN model. The results show that the 128-size CNN performs better because large images contain rich aerosol information. Text Aerosol Robotic Network MDPI Open Access Publishing Atmosphere 14 9 1400 |
institution |
Open Polar |
collection |
MDPI Open Access Publishing |
op_collection_id |
ftmdpi |
language |
English |
topic |
AOD CNN Sentinel-2 deep learning |
spellingShingle |
AOD CNN Sentinel-2 deep learning Jie Jiang Jiaxin Liu Donglai Jiao Aerosol Optical Depth Retrieval for Sentinel-2 Based on Convolutional Neural Network Method |
topic_facet |
AOD CNN Sentinel-2 deep learning |
description |
Atmospheric aerosol significantly affects the climate environment and public health, and Aerosol Optical Depth (AOD) is a fundamental optical characteristic parameter of aerosols, so it is important to develop methods for obtaining AOD. In this work, a novel AOD retrieval algorithm based on a Convolutional Neural Network (CNN) method that could provide continuous and detailed aerosol distribution is proposed. The algorithm utilizes data from Sentinel-2 and Aerosol Robotic Network (AERONET) spanning from 2016 to 2022. The CNN AOD data are consistent with the AERONET measurements, with an R2 of 0.95 and RMSE of 0.049 on the test dataset. CNN demonstrates superior performance in retrieving AOD compared with other algorithms. CNN retrieves AOD well on high reflectance surfaces, such as urban and bare soil, with RMSEs of 0.051 and 0.042, respectively. CNN efficiently retrieves AOD in different seasons, but it performs better in summer and winter than in spring and autumn. In addition, to study the relationship between image size and model retrieval performance, image datasets of 32 × 32, 64 × 64 and 128 × 128 pixels were created to train and test the CNN model. The results show that the 128-size CNN performs better because large images contain rich aerosol information. |
format |
Text |
author |
Jie Jiang Jiaxin Liu Donglai Jiao |
author_facet |
Jie Jiang Jiaxin Liu Donglai Jiao |
author_sort |
Jie Jiang |
title |
Aerosol Optical Depth Retrieval for Sentinel-2 Based on Convolutional Neural Network Method |
title_short |
Aerosol Optical Depth Retrieval for Sentinel-2 Based on Convolutional Neural Network Method |
title_full |
Aerosol Optical Depth Retrieval for Sentinel-2 Based on Convolutional Neural Network Method |
title_fullStr |
Aerosol Optical Depth Retrieval for Sentinel-2 Based on Convolutional Neural Network Method |
title_full_unstemmed |
Aerosol Optical Depth Retrieval for Sentinel-2 Based on Convolutional Neural Network Method |
title_sort |
aerosol optical depth retrieval for sentinel-2 based on convolutional neural network method |
publisher |
Multidisciplinary Digital Publishing Institute |
publishDate |
2023 |
url |
https://doi.org/10.3390/atmos14091400 |
op_coverage |
agris |
genre |
Aerosol Robotic Network |
genre_facet |
Aerosol Robotic Network |
op_source |
Atmosphere Volume 14 Issue 9 Pages: 1400 |
op_relation |
Aerosols https://dx.doi.org/10.3390/atmos14091400 |
op_rights |
https://creativecommons.org/licenses/by/4.0/ |
op_doi |
https://doi.org/10.3390/atmos14091400 |
container_title |
Atmosphere |
container_volume |
14 |
container_issue |
9 |
container_start_page |
1400 |
_version_ |
1779321403520057344 |