Sentinel-2 Analysis-Ready Data Pipeline

Data Pipeline

Designed and built a containerised pipeline that ingests Sentinel-2 imagery, applies cloud masking and reprojection, and publishes cloud-optimised GeoTIFFs and STAC metadata for downstream science teams.

-80%

Processing Time

500+

Scenes / Day

100%

Reproducible

Gallery

01 / 03

Scene Ingestion

Automated discovery and download of Sentinel-2 tiles from open archives.

Tip: use ← → keys for keyboard navigation.

The story

Scroll through problem, approach, and outcome.

Problem

Science teams were manually downloading and pre-processing satellite scenes — slow, inconsistent, and impossible to reproduce or audit.

Approach

Built an orchestrated ETL pipeline in Airflow that ingests scenes from open archives, runs cloud masking and reprojection with Rasterio and GDAL, and writes cloud-optimised GeoTIFFs plus STAC metadata to object storage — all containerised with Docker and deployed on GCP.

Outcome

Delivered a hands-off, scheduled pipeline producing analysis-ready data with consistent QA, freeing scientists to focus on analysis instead of data wrangling.

Tech Stack

PythonRasterioGDALAirflowDockerGCP