Hide and Seek: Investigating Redundancy in Earth Observation Imagery

TL;DR

This paper reveals that Earth Observation data contains significant multidimensional redundancy, which can be exploited to achieve high performance with substantially reduced computational costs across various tasks and conditions.

Contribution

It systematically demonstrates the pervasive redundancy in EO data and quantifies its impact on model efficiency and scalability, highlighting a fundamental property of EO data.

Findings

Redundancy in EO data is substantial and consistent.

Exploiting redundancy achieves ~98.5% of baseline performance with 4x fewer GFLOPs.

Redundancy is a structural property across tasks, locations, sensors, and architectures.

Abstract

The growing availability of Earth Observation (EO) data and recent advances in Computer Vision have driven rapid progress in machine learning for EO, producing domain-specific models at ever-increasing scales. Yet this progress risks overlooking fundamental properties of EO data that distinguish it from other domains. We argue that EO data exhibit a multidimensional redundancy (spectral, temporal, spatial, and semantic) which has a more pronounced impact on the domain and its applications than what current literature reflects. To validate this hypothesis, we conduct a systematic domain-specific investigation examining the existence, consistency, and practical implications of this phenomenon across key dimensions of EO variability. Our findings confirm that redundancy in EO data is both substantial and pervasive: exploiting it yields comparable performance ($\approx98.5\%$ of baseline) at a fraction of the computational cost ($\approx4\times$ fewer GFLOPs), at both training and inference. Crucially, these gains are consistent across tasks, geospatial locations, sensors, ground sampling distances, and architectural designs; suggesting that multi-faceted redundancy is a structural property of EO data rather than an artifact of specific experimental choices. These results lay the groundwork for more efficient, scalable, and accessible large-scale EO models.