THOR: A Versatile Foundation Model for Earth Observation Climate and Society Applications

TL;DR

THOR is a versatile, compute-adaptive foundation model for Earth observation that unifies multiple satellite data sources and enables flexible deployment with dynamic resolution and cost trade-offs.

Contribution

It introduces a novel architecture and training strategy that unifies heterogeneous satellite data and allows flexible inference without retraining.

Findings

Achieves state-of-the-art performance on Earth observation benchmarks.

Demonstrates effective multi-sensor data integration from Sentinel satellites.

Enables dynamic resolution and compute trade-offs during inference.

Abstract

Current Earth observation foundation models are architecturally rigid, struggle with heterogeneous sensors and are constrained to fixed patch sizes. This limits their deployment in real-world scenarios requiring flexible computeaccuracy trade-offs. We propose THOR, a "computeadaptive" foundation model that solves both input heterogeneity and deployment rigidity. THOR is the first architecture to unify data from Copernicus Sentinel-1, -2, and -3 (OLCI & SLSTR) satellites, processing their native 10 m to 1000 m resolutions in a single model. We pre-train THOR with a novel randomized patch and input image size strategy. This allows a single set of pre-trained weights to be deployed at inference with any patch size, enabling a dynamic trade-off between computational cost and feature resolution without retraining. We pre-train THOR on THOR Pretrain, a new, large-scale multi-sensor dataset and demonstrate state-of-the-art performance on downstream benchmarks, particularly in data-limited regimes like the PANGAEA 10% split, validating that THOR's flexible feature generation excels for diverse climate and society applications.