TerraTorch: The Geospatial Foundation Models Toolkit

TL;DR

The paper tackles the challenge of adapting Geospatial Foundation Models to Earth observation, weather, and climate data, which are highly heterogeneous and domain-specific. It introduces TerraTorch, a modular toolkit built on PyTorch Lightning that enables end-to-end fine-tuning and systematic benchmarking through a model factory, reusable data modules, and the Iterate hyperparameter optimization extension, with GEO-Bench integration. The key contributions are the four-component design (tasks, model factory, datasets, Iterate), support for multiple GeoFMs and backbones, and automated HPO for reproducible benchmarking. Experiments demonstrate that architecture and pretraining data choices materially affect downstream performance and that Iterate can provide meaningful hyperparameter gains, highlighting the practical value of a plug-and-play EO toolkit. By open-sourcing TerraTorch, the work lowers barriers to adopting GeoFMs in research and production pipelines and enables reproducible, scalable workflows for EO analytics.

Abstract

TerraTorch is a fine-tuning and benchmarking toolkit for Geospatial Foundation Models built on PyTorch Lightning and tailored for satellite, weather, and climate data. It integrates domain-specific data modules, pre-defined tasks, and a modular model factory that pairs any backbone with diverse decoder heads. These components allow researchers and practitioners to fine-tune supported models in a no-code fashion by simply editing a training configuration. By consolidating best practices for model development and incorporating the automated hyperparameter optimization extension Iterate, TerraTorch reduces the expertise and time required to fine-tune or benchmark models on new Earth Observation use cases. Furthermore, TerraTorch directly integrates with GEO-Bench, allowing for systematic and reproducible benchmarking of Geospatial Foundation Models. TerraTorch is open sourced under Apache 2.0, available at https://github.com/IBM/terratorch, and can be installed via pip install terratorch.

Figures (2)

• Figure 1: Overview of the TerraTorch components and structure.
• Figure 2: Effect of hyperparameters on the validation mIoU per dataset. The best runs are marked with a diamond.