TerraQ

About TerraQ...

...a spatiotemporal question-answering engine for Sentinel-1 and Sentinel-2 archives.

...built to make Earth Observation data more accessible to the public and enhance the productivity of experts.

Screenshot of the TerraQ application

Natural Language Interface

Interact using only natural language, no expert knowledge required.

Sentinel mission images

Access 1.5 milion Sentinel-1 and Sentinel-2 images of Europe.

Rapid execution

Built using a range of industry standard and custom built tools to ensure responsiveness.

Open Source

All parts of the system are publicly available and built using open research.

View on GitHub

Architecture

TerraQ leverages a pipeline architecture combining classical and modern natural language processing techniques.

The TerraQ application consists of four components:

The Web UI serves as the point of interaction for the user. Through it the user can make textual requests in plain English, visualize results and download images.
The TerraQ pipeline is responsible for understanding the user's request and translating it to a machine-executable SPARQL query.
The Knowledge Graph is a purpose-built data repository that contains information about satellite images, administrative divisions, natural features and more.
The RDF engine is tasked with executing the query over the target Knowledge Graph and returning a result.

Architecture diagram showcasing the components of TerraQ.

Publications

Sergios-Anestis Kefalidis, Konstantinos Plas and Manolis Koubarakis. TerraQ: Spatiotemporal Question-Answering on Satellite Image Archives. IEEE International Geoscience and Remote Sensing Symposium 2025. Paper.