TiMo: Spatiotemporal Foundation Model for Satellite Image Time Series

TL;DR

TiMo is a hierarchical vision transformer model designed for satellite image time series, capturing multiscale spatiotemporal patterns to improve various Earth observation tasks.

Contribution

We introduce TiMo, a novel hierarchical transformer with a spatiotemporal gyroscope attention mechanism and a large-scale pre-training dataset, enhancing SITS analysis capabilities.

Findings

TiMo outperforms state-of-the-art methods in multiple tasks.

Pre-training on MillionST improves generalization.

Effective modeling of multiscale spatiotemporal patterns.

Abstract

Satellite image time series (SITS) provide continuous observations of the Earth's surface, making them essential for applications such as environmental management and disaster assessment. However, existing spatiotemporal foundation models rely on plain vision transformers, which encode entire temporal sequences without explicitly capturing multiscale spatiotemporal relationships between land objects. This limitation hinders their effectiveness in downstream tasks. To overcome this challenge, we propose TiMo, a novel hierarchical vision transformer foundation model tailored for SITS analysis. At its core, we introduce a spatiotemporal gyroscope attention mechanism that dynamically captures evolving multiscale patterns across both time and space. For pre-training, we curate MillionST, a large-scale dataset of one million images from 100,000 geographic locations, each captured across 10 temporal phases over five years, encompassing diverse geospatial changes and seasonal variations. Leveraging this dataset, we adapt masked image modeling to pre-train TiMo, enabling it to effectively learn and encode generalizable spatiotemporal representations.Extensive experiments across multiple spatiotemporal tasks-including deforestation monitoring, land cover segmentation, crop type classification, and flood detection-demonstrate TiMo's superiority over state-of-the-art methods. Code, model, and dataset will be released at https://github.com/MiliLab/TiMo.