Data-Driven Invertible Neural Surrogates of Atmospheric Transmission

TL;DR

This paper introduces a data-driven, invertible neural network framework that uses differentiable programming to accurately infer atmospheric transmission profiles from spectral data, enabling improved atmospheric correction and spectral data translation.

Contribution

It presents a novel, physics-based, differentiable surrogate model for atmospheric transmission that automatically tunes itself for spectral data analysis.

Findings

Effective atmospheric correction demonstrated

Spectral data can be recast between modalities

Accurate inference of atmospheric transmission profiles

Abstract

We present a framework for inferring an atmospheric transmission profile from a spectral scene. This framework leverages a lightweight, physics-based simulator that is automatically tuned - by virtue of autodifferentiation and differentiable programming - to construct a surrogate atmospheric profile to model the observed data. We demonstrate utility of the methodology by (i) performing atmospheric correction, (ii) recasting spectral data between various modalities (e.g. radiance and reflectance at the surface and at the sensor), and (iii) inferring atmospheric transmission profiles, such as absorbing bands and their relative magnitudes.