TESSERA (Temporal Embeddings of Surface Spectra for Earth Representation and Analysis)

Description

TESSERA is a global geospatial foundation model that turns the European Space Agency's raw Sentinel‑1 (SAR) + Sentinel‑2 (optical) time series into compact, gap-free, 128‑dimensional, 10‑meter embeddings, producing an annual “representation map” for every 10m land pixel worldwide from 2017–2024 (rolling forward each new year) enabling detailed analysis of land cover dynamics, vegetation health, and environmental monitoring over time.Each 10m pixel encodes a full year’s spectral‑temporal behavior across the two major sensor modalities (optical + radar), enabling downstream tasks (carbon quantification, land cover, canopy height, change detection, hazards, habitat mapping, biodiversity) to start from a compact annual descriptor instead of petabytes of raw scenes. This collapses costs and latency for analysts and builders while enhancing accuracy and enabling indexing.

Update Frequency

Annually (as new Sentinel data becomes available)

License

CC‑BY‑SA

Documentation

TESSERA Documentation How to use the TESSERA dataset, model details, and example applications. https://geotessera.readthedocs.io/en/latest/index.html How to generate the embeddings using AWS data https://github.com/dClimate/tessera-embeddings

Managed By

"Energy and Environment Group, Department of Computer Science and Technology, University of Cambridge" "dClimate"

See all datasets managed by Energy and Environment Group, Department of Computer Science and Technology, University of Cambridge.

Contact

Questions about the data can be raised as an issue at https://github.com/ucam-eo/geotessera/issues

How to Cite

TESSERA (Temporal Embeddings of Surface Spectra for Earth Representation and Analysis) was accessed on DATE from https://registry.opendata.aws/tessera. https://arxiv.org/abs/2506.20380

Usage Examples

Tutorials

• Basic Data Analysis, Point Sampling Workflow, GIS Integration Workflow, Coverage Analysis with Boundary Visualization, Custom Analysis Workflows, Cloud-Native Analysis with Zarr by Cambridge Centre for Earth Observation
  
• Get to Know a Dataset: Tessera by Frank Feng, Anil Madhavapeddy
  
  

Tools & Applications

• GeoTessera Python Package by Cambridge Centre for Earth Observation
  
• Interactive Tessera Embedding Classifier by Cambridge Centre for Earth Observation
  
• Solar Panel Detection with GeoTessera by Cambridge Centre for Earth Observation
  

Publications

• TESSERA Temporal Embeddings of Surface Spectra for Earth Representation and Analysis by Zhengpeng F.(Cambridge), Atzberger C.(dClimate), Jaffer S.(Cambridge), Knezevic J.(Cambridge), Sormunen S.(Aalto Uni), Young R.(Cambridge), et al