Susiluoto et al. (2025) Improved Atmospheric Correction for Remote Imaging Spectroscopy Missions with Accelerated Optimal Estimation

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Remote Sensing
• Year: 2025
• Date: 2025-11-14
• Authors: Jouni Susiluoto, Niklas Bohn, Amy Braverman, Philip G. Brodrick, Nimrod Carmon, M. R. Gunson, Hai Nguyen, David R. Thompson, M. Turmon
• DOI: 10.3390/rs17223719

Research Groups

[Information not provided in the paper text.]

Short Summary

This paper introduces Accelerated Optimal Estimation (AOE), a Bayesian algorithm that significantly speeds up hyperspectral surface reflectance retrieval and improves convergence compared to standard Optimal Estimation (OE), while also validating the accuracy of Gaussian uncertainty estimates from OE-type algorithms using Markov Chain Monte Carlo (MCMC).

Objective

• To develop a faster and more accurate Bayesian algorithm for hyperspectral surface reflectance retrieval from space-based imaging spectrometer data, and to validate the accuracy of Gaussian uncertainty estimates associated with Optimal Estimation (OE) type algorithms.

Study Configuration

• Spatial Scale: Pixel to regional (demonstrated with an AVIRIS-NG scene).
• Temporal Scale: Instantaneous (snapshot) for individual retrievals, applicable to continuous data streams from imaging spectrometers.

Methodology and Data

• Models used: Optimal Estimation (OE), Accelerated Optimal Estimation (AOE), Markov Chain Monte Carlo (MCMC).
• Data sources: Remotely measured radiance data from space-based imaging spectrometers, specifically an AVIRIS-NG scene.

Main Results

• Accelerated Optimal Estimation (AOE) speeds up the Optimal Estimation (OE) reflectance inversion process by up to two orders of magnitude compared to a reference OE implementation (ROE).
• AOE provides improved convergence over a number of selected test targets.
• Gaussian uncertainty estimates from OE-type algorithms are accurate under given atmospheric conditions, as validated by comparison with non-Gaussian posterior distributions obtained with Markov Chain Monte Carlo (MCMC).
• AOE demonstrates effective scalability to a larger AVIRIS-NG scene, showcasing its ability to handle complex, large-scale data.

Contributions

• Introduction of Accelerated Optimal Estimation (AOE), a novel Bayesian algorithm that significantly enhances the computational efficiency (up to two orders of magnitude faster) and convergence of hyperspectral surface reflectance retrieval compared to existing Optimal Estimation (OE) methods.
• Empirical validation of the accuracy of Gaussian uncertainty estimates in OE-type algorithms for hyperspectral retrievals through comparison with Markov Chain Monte Carlo (MCMC) results, addressing a previously unvalidated assumption.
• Demonstration of AOE's practical applicability and scalability for processing complex, large-scale remote sensing datasets, such as those from AVIRIS-NG.

Funding

[Information not provided in the paper text.]

Citation

@article{Susiluoto2025Improved,
  author = {Susiluoto, Jouni and Bohn, Niklas and Braverman, Amy and Brodrick, Philip G. and Carmon, Nimrod and Gunson, M. R. and Nguyen, Hai and Thompson, David R. and Turmon, M.},
  title = {Improved Atmospheric Correction for Remote Imaging Spectroscopy Missions with Accelerated Optimal Estimation},
  journal = {Remote Sensing},
  year = {2025},
  doi = {10.3390/rs17223719},
  url = {https://doi.org/10.3390/rs17223719}
}