AQUAVERSE-AC: a machine learning-based atmospheric correction framework for inland and coastal waters
6 affiliations
- 1NASA Goddard Space Flight Center, Greenbelt, MD, United States
- 2Science Systems and Applications Inc. (SSAI), Lanham, MD, United States
- 3Earth Observation and Geoinformatics Division, National Institute for Space Research, Sao Jose Dos Campos, Brazil
- 4Department of Physics, University of Maryland Baltimore County, Baltimore, MD, United States
- 5Geo-Sensing and Imaging Consultancy, Trivandrum, Kerala, India
- 6Department of ECE, Chennai Institute of Technology, Chennai, India
Abstract
Accurately deriving remote sensing reflectance (Rrs) from satellite imagery in optically complex inland and coastal waters is critical for water quality monitoring and management. This study presents AQUAVERSE-AC, the atmospheric correction component of the AQUAVERSE (Aquatic inversion scheme for remote sensing of fresh and coastal waters) platform, an innovative machine learning-based atmospheric correction (AC) framework, that leverages Mixture Density Networks (MDNs) to retrieve high-quality Rrs from multispectral data acquired by Landsat-8/9 (OLI) and Sentinel-2A/B (MSI) satellites. By coupling a well-validated ocean-atmosphere radiative transfer (RT) model with extensive simulated and in situ datasets, AQUAVERSE-AC compensates for aerosol scattering and atmospheric absorption across diverse environmental and imaging conditions. The model’s performance was evaluated through a comprehensive matchup analysis against AERONET-OC and GLORIA datasets, wherein AQUAVERSE-AC generally produced lower median symmetric accuracy (ε) compared to the state-of-the-art AC processors. ε ranged from 15% to 26% for key spectral bands (coastal-blue to 705 nm red-edge band), representing a 20%–100% improvement in accuracy compared to alternative methods. Notably, AQUAVERSE-AC exhibited robust performance in the blue spectrum within the evaluated matchups. The results, including generated Rrs maps, were benchmarked against outputs from other widely used AC processors (ACOLITE, POLYMER, SeaDAS, OC-SMART), indicating AQUAVERSE-AC’s competitive performance across diverse aquatic environments, while also providing comprehensive uncertainty estimates. These uncertainty metrics, inherently generated by the MDN framework, provide critical confidence indicators for downstream applications. These results demonstrate the potential of AQUAVERSE-AC as an AC framework for OLI and MSI observations of inland and coastal waters, while further validation is needed to establish its performance across additional regions, atmospheric conditions, and diverse optically complex aquatic systems.
Related
- Atmospheric Predictability beyond 30 Days with Machine Learning
- The Equilibrium Response of Atmospheric Machine‐Learning Models to Uniform Sea Surface Temperature Warming
- SATFRONT: Detecting Atmospheric Fronts with Satellite Data using Machine Learning
- A Deep-Learning Framework for the Generalized Emulation of Atmospheric Radiative Transfer at Microwave Frequencies
- Machine-learning-based approach for solar radiation model uncertainty identification, attribution and bias correction
BibTeX
@article{ashapure2026aquaverseac,
title = {{AQUAVERSE-AC: a machine learning-based atmospheric correction framework for inland and coastal waters}},
author = {Akash Ashapure and Akash Ashapure and Brandon Smith and Brandon Smith and Ryan E. O’Shea and Ryan E. O’Shea and William Wainwright and William Wainwright and Arun M. Saranathan and Arun M. Saranathan and Daniel Andrade Maciel and Peng-Wang Zhai and Sundarabalan V. Balasubramanian and Sundarabalan V. Balasubramanian and Sakib Kabir},
journal = {Frontiers in Remote Sensing},
year = {2026},
month = nov,
doi = {10.3389/frsen.2026.1839175},
url = {https://doi.org/10.3389/frsen.2026.1839175},
note = {CC BY},
}
RIS
TY - JOUR
TI - AQUAVERSE-AC: a machine learning-based atmospheric correction framework for inland and coastal waters
AU - Akash Ashapure
AU - Akash Ashapure
AU - Brandon Smith
AU - Brandon Smith
AU - Ryan E. O’Shea
AU - Ryan E. O’Shea
AU - William Wainwright
AU - William Wainwright
AU - Arun M. Saranathan
AU - Arun M. Saranathan
AU - Daniel Andrade Maciel
AU - Peng-Wang Zhai
AU - Sundarabalan V. Balasubramanian
AU - Sundarabalan V. Balasubramanian
AU - Sakib Kabir
PY - 2026
DA - 2026/11/04
JO - Frontiers in Remote Sensing
DO - 10.3389/frsen.2026.1839175
UR - https://doi.org/10.3389/frsen.2026.1839175
AB - Accurately deriving remote sensing reflectance (Rrs) from satellite imagery in optically complex inland and coastal waters is critical for water quality monitoring and management. This study presents AQUAVERSE-AC, the atmospheric correction component of the AQUAVERSE (Aquatic inversion scheme for remote sensing of fresh and coastal waters) platform, an innovative machine learning-based atmospheric correction (AC) framework, that leverages Mixture Density Networks (MDNs) to retrieve high-quality Rrs from multispectral data acquired by Landsat-8/9 (OLI) and Sentinel-2A/B (MSI) satellites. By coupling a well-validated ocean-atmosphere radiative transfer (RT) model with extensive simulated and in situ datasets, AQUAVERSE-AC compensates for aerosol scattering and atmospheric absorption across diverse environmental and imaging conditions. The model’s performance was evaluated through a comprehensive matchup analysis against AERONET-OC and GLORIA datasets, wherein AQUAVERSE-AC generally produced lower median symmetric accuracy (ε) compared to the state-of-the-art AC processors. ε ranged from 15% to 26% for key spectral bands (coastal-blue to 705 nm red-edge band), representing a 20%–100% improvement in accuracy compared to alternative methods. Notably, AQUAVERSE-AC exhibited robust performance in the blue spectrum within the evaluated matchups. The results, including generated Rrs maps, were benchmarked against outputs from other widely used AC processors (ACOLITE, POLYMER, SeaDAS, OC-SMART), indicating AQUAVERSE-AC’s competitive performance across diverse aquatic environments, while also providing comprehensive uncertainty estimates. These uncertainty metrics, inherently generated by the MDN framework, provide critical confidence indicators for downstream applications. These results demonstrate the potential of AQUAVERSE-AC as an AC framework for OLI and MSI observations of inland and coastal waters, while further validation is needed to establish its performance across additional regions, atmospheric conditions, and diverse optically complex aquatic systems.
SN - 2673-6187
ER -