A Sequence-to-Sequence Transformer Model for Satellite Retrieval of Aerosol Optical and Microphysical Parameters from Space DOI Creative Commons
Luo Zhang, Haoran Gu, Zhengqiang Li

et al.

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(24), P. 4659 - 4659

Published: Dec. 12, 2024

Aerosol optical and microphysical properties determine their radiative capabilities, climatic impacts, health effects. Satellite remote sensing is a crucial tool for obtaining aerosol parameters on global scale. However, traditional physical statistical retrieval methods face bottlenecks in data mining capacity as the volume of satellite observation information increases rapidly. Artificial intelligence are increasingly applied to parameter retrieval, yet most current approaches focus end-to-end single-parameter without considering inherent relationships among multiple properties. In this study, we propose sequence-to-sequence joint algorithm based transformer model S2STM. Unlike conventional methods, leverages encoding–decoding capabilities model, coupling multi-source such polarized satellite, meteorological, surface characteristics, incorporates physically coherent consistency loss function. This approach transforms numerical regression into relationship mapping. We observations from Chinese polarimetric (the Particulate Observing Scanning Polarimeter, POSP) simultaneously retrieved key parameters. Event analyses, including dust pollution episodes, demonstrate method’s responsiveness hotspot regions events. The results show good agreement with ground-based products. method also adaptable instruments various configurations (e.g., multi-wavelength, multi-angle, multi-dimensional polarization) can further improve its spatiotemporal generalization performance by enhancing spatial balance ground station training datasets.

Language: Английский

Global aerosol retrieval over land from Landsat imagery integrating Transformer and Google Earth Engine DOI
Jing Wei, Zhihui Wang, Zhanqing Li

et al.

Remote Sensing of Environment, Journal Year: 2024, Volume and Issue: 315, P. 114404 - 114404

Published: Sept. 24, 2024

Language: Английский

Citations

7
Passive Remote Sensing of Marine Liquid Cloud Geometric Thickness Using the O2–O2 Band: First Results From TROPOMI DOI Creative Commons
Wenwu Wang,

Chong Shi,

Jian Xu

et al.

Geophysical Research Letters, Journal Year: 2025, Volume and Issue: 52(3)

Published: Jan. 30, 2025

Abstract Observations on cloud geometric thickness are crucial for understanding the radiative balance and aerosol indirect effects, currently, retrieval studies passive instruments remain constrained due to lack of incident radiation penetrability. In this work, we firstly analyze relationship between droplets distribution penetrability based physical model, then fully utilize advantages hyperspectral O 4 measurements build a physically machine learning model retrieve thickness. The algorithm retrieves from TROPOMI observations first time, retrievals compared with active observations. It is found that mean absolute error using 2B‐CLDPROF‐LIDAR cloud‐top height as input 0.49 km, which shows potential band

Language: Английский

Citations

0
A High-Resolution Aerosol Retrieval Algorithm Via Deep Learning DOI
Bing Tu, Chengxin Hu, Bo Liu

et al.

Published: Jan. 1, 2025

Language: Английский

Citations

0
Monitoring of aerosol optical-microphysical properties from DPC/GF-5(02): A case study of dust event in North China Plain DOI Creative Commons

Yujia Cao,

Cheng Chen, Haixiao Yu

et al.

Atmospheric Environment, Journal Year: 2025, Volume and Issue: unknown, P. 121193 - 121193

Published: March 1, 2025

Language: Английский

Citations

0
Enhancing global aerosol retrieval from satellite data via deep learning with mutual information estimation DOI
Xiaohu Sun, Yong Xue, Lin Sun

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2025, Volume and Issue: 139, P. 104534 - 104534

Published: April 14, 2025

Language: Английский

Citations

0
Aerosol Retrieval Method Using Multi-Angle Data from GF-5 02 DPC over the Jing–Jin–Ji Region DOI Creative Commons
Zhongting Wang, Shikuan Jin, Cheng Chen

et al.

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(8), P. 1415 - 1415

Published: April 16, 2025

The Directional Polarimetric Camera (DPC) aboard the Chinese GaoFen-5 02 satellite is designed to monitor aerosols and particulate matter (PM). In this study, we retrieved aerosol optical depth (AOD) over Jing–Jin–Ji (JJJ) region using multi-angle data from DPC, employing a combination of dark dense vegetation (DDV) retrieval methods. added value our method included novel hybrid methodology good practical performance. process involves three main steps: (1) deriving AOD DPC collected at nadir angle linear parameters land surface reflectance between blue red bands MOD09 product; (2) after performing atmospheric correction with AOD, calculating variance normalized all observation angles; (3) leveraging calculated obtain final values. images JJJ were successfully January June 2022. To validate method, compared results products AErosol RObotic NETwork (AERONET) Beijing-RADI site, as well MODerate-resolution Imaging Spectroradiometer (MODIS) generalized atmosphere properties (GRASP)/models same site. terms validation metrics, correlation coefficient (R2) root mean square error (RMSE) indicated that achieved high accuracy, an R2 greater than 0.9 RMSE below 0.1, closely aligning performance GRASP.

Language: Английский

Citations

0
Characteristics of aerosols and planetary boundary layer dynamics during biomass burning season DOI Creative Commons

Muhammad Zeeshaan Shahid,

Muhammad Imran Shahzad,

Sundas Jaweria

et al.

Natural Hazards, Journal Year: 2025, Volume and Issue: unknown

Published: May 2, 2025

Language: Английский

Citations

0
Development of an Algorithm for the Simultaneous Retrieval of Cloud-Top Height and Cloud Optical Thickness Combining Radiative Transfer and Multisource Satellite Information From O₄ Hyperspectral Measurements DOI
Wenwu Wang, Chong Shi, Huazhe Shang

et al.

IEEE Transactions on Geoscience and Remote Sensing, Journal Year: 2024, Volume and Issue: 62, P. 1 - 11

Published: Jan. 1, 2024

Remote sensing of cloud properties based on multispectral or hyperspectral observations from satellites is important for earth radiation budget and climate change studies. Currently, most retrieval algorithms the measurements are developed O 2 -A band to derive optical thickness (COT) top height (CTH) via optimal estimation theory. Nevertheless, there few studies COT CTH using xmlns:xlink="http://www.w3.org/1999/xlink">4 band, where direct computation slant column density spectral information in blue provide a faster yet flexible inversion strategy. In this study, we develop novel algorithm neural networks (CRANN-O4) simultaneous derivation CTH. CRANN-O4 employs transfer learning strategy that combines radiative model (RTM) multisource satellite data, which deep network module pretrained simulation data RTM enhance its adaptability interpretability, following fine-tuning scheme data. To evaluate performance, apply TROPOMI make an intercomparison with official products, generated band. The results indicate CRANN-O4-derived spatial distributions generally similar product but more consistent SNPP-VIIRS product. RMSEs derived by approximately 15.88 2.33 km, respectively, while those 20.85 3.00 respectively. addition, validation CALIOP demonstrates better agreement than product, RMSE decreasing 2.7 km 2.2 km. methodology presented study provides innovative insight into parameter instruments channels, such as FY-3F/OMS.

Language: Английский

Citations

3
First lunar-light mapping of nighttime dust season oceanic aerosol optical depth over North Atlantic from space DOI
Meng Zhou, Jun Wang, Xi Chen

et al.

Remote Sensing of Environment, Journal Year: 2024, Volume and Issue: 312, P. 114315 - 114315

Published: July 27, 2024

Language: Английский

Citations

2
Wide and Deep Learning Model for Satellite-Based Real-Time Aerosol Retrievals in China DOI Creative Commons
Nana Luo,

Junxiao Zou,

Zhou Zang

et al.

Atmosphere, Journal Year: 2024, Volume and Issue: 15(5), P. 564 - 564

Published: April 30, 2024

Machine learning methods have been recognized as rapid for satellite-based aerosol retrievals but not widely applied in geostationary satellites. In this study, we developed a wide and deep model to retrieve the optical depth (AOD) using Himawari-8. Compared traditional methods, embedded “wide” modeling component tested proposed across China independent training (2016–2018) test (2019) datasets. The results showed that improves accuracy enhances interpretability. estimates exhibited better (R2 = 0.81, root-mean-square errors (RMSEs) 0.19, within estimated error (EE) 63%) than those of deep-only models 0.78, RMSE 0.21, EE 58%). comparison with extreme gradient boosting (XGBoost) Himawari-8 V2.1 AOD products, there were also significant improvements. addition higher accuracy, interpretability was superior model. other seasons, contributions spring concentrations interpreted. Based on application model, near-real-time variation over could be captured an ultrafine temporal resolution.

Language: Английский

Citations

0