Lossy Neural Compression for Geospatial Analytics: A Review

Over the past decades, there has been an explosion in the amount of available Earth Observation (EO) data. The unprecedented coverage of the Earth’s surface and atmosphere by satellite imagery has resulted in large volumes of data that must be transmitted to ground stations, stored in data centers, and distributed to end users. Modern Earth System Models (ESMs) face similar issues, operating at high resolutions in space and time, producing petabytes of data per simulated day. Data compression has gained relevance, with neural compression emerging from deep learning and information theory, making EO data and ESM outputs ideal candidates due to abundant unlabeled data. In this review, we outline recent developments in neural compression applied to geospatial analytics. We introduce the main concepts in neural compression and seminal works in its traditional applications to image and video compression domains. We then discuss the unique characteristics of EO and ESM data, contrasting them with “natural images”, and explain the additional challenges and opportunities they present. We also review current applications of neural compression across various EO modalities and explore the limited efforts in ESM compression. The advent of self-supervised learning (SSL) and foundation models (FM) has advanced methods to efficiently distill representations from vast unlabeled data. We connect these developments to neural compression for EO, highlighting the similarities in the two fields and the potential of transferring compressed feature representations for machine–to–machine communication. Based on insights drawn from this review, we devise future directions relevant to applications in EO and ESM.