Atmospheric correction with the Bayesian empirical line.

Area of Science:

• Remote Sensing
• Atmospheric Correction
• Spectroscopy

Background:

• Traditional atmospheric correction methods for visible/infrared spectra rely on either physics-based Radiative Transfer Models (RTMs) or empirical in situ measurements.
• These distinct approaches have limitations, particularly in large or remote areas where acquiring coincident field data for empirical methods is challenging.

Purpose of the Study:

• To develop a unified probabilistic framework for atmospheric correction that integrates physics-based and empirical techniques.
• To enhance the accuracy and stability of spectral data correction, making empirical methods more practical for extensive remote sensing campaigns.

Main Methods:

• A probabilistic formulation unifies physics-based and empirical atmospheric correction methods.
• A physics-based solution establishes a prior distribution for reflectances and correction coefficients.
• Bayesian inference incorporates reference measurements, yielding a Maximum A Posteriori (MAP) estimate, with Gaussian assumptions leading to a Tikhonov-regularized closed-form solution.

Main Results:

• The unified probabilistic method provides a more accurate and stable atmospheric correction compared to purely physics-based or empirical methods.
• The technique is simple to implement and yields reliable results using one or more reference spectra.
• Performance was validated using atmospheric simulations and historical data from the Airborne Visible Infrared Imaging Spectrometer (AVIRIS-C).

Conclusions:

• The developed probabilistic approach offers a robust and versatile solution for atmospheric correction in visible/infrared spectroscopy.
• This method significantly improves the practicality and applicability of empirical corrections in challenging remote sensing scenarios.
• The unified framework advances the field of spectral data processing for applications like the HyspIRI mission.