Hyperspectral Variational Autoencoders for Joint Data Compression and Component Extraction

TL;DR

This paper introduces a variational autoencoder that compresses hyperspectral satellite data by over 500 times while retaining essential atmospheric information, enabling efficient data storage and retrieval for Earth observation.

Contribution

It presents a novel VAE-based compression method that preserves atmospheric signals in hyperspectral data and analyzes the encoding of atmospheric information in the latent space.

Findings

Achieves x514 compression with low reconstruction error.

Successfully extracts cloud fraction and ozone levels from compressed data.

Highlights challenges in retrieving weaker atmospheric signals like NO2 and HCHO.

Abstract

Geostationary hyperspectral satellites generate terabytes of data daily, creating critical challenges for storage, transmission, and distribution to the scientific community. We present a variational autoencoder (VAE) approach that achieves x514 compression of NASA's TEMPO satellite hyperspectral observations (1028 channels, 290-490nm) with reconstruction errors 1-2 orders of magnitude below the signal across all wavelengths. This dramatic data volume reduction enables efficient archival and sharing of satellite observations while preserving spectral fidelity. Beyond compression, we investigate to what extent atmospheric information is retained in the compressed latent space by training linear and nonlinear probes to extract Level-2 products (NO2, O3, HCHO, cloud fraction). Cloud fraction and total ozone achieve strong extraction performance (R^2 = 0.93 and 0.81 respectively), though these represent relatively straightforward retrievals given their distinct spectral signatures. In contrast, tropospheric trace gases pose genuine challenges for extraction (NO2 R^2 = 0.20, HCHO R^2 = 0.51) reflecting their weaker signals and complex atmospheric interactions. Critically, we find the VAE encodes atmospheric information in a semi-linear manner - nonlinear probes substantially outperform linear ones - and that explicit latent supervision during training provides minimal improvement, revealing fundamental encoding challenges for certain products. This work demonstrates that neural compression can dramatically reduce hyperspectral data volumes while preserving key atmospheric signals, addressing a critical bottleneck for next-generation Earth observation systems. Code - https://github.com/cfpark00/Hyperspectral-VAE