Open-source processing pipelines for Cyclone Gabrielle
summary
LINZ migrates imagery QA and standardization into the cloud using Argo Workflows on Kubernetes with GDAL containers on AWS EKS and EC2 spot instances, processing ~50,000 km² of mostly 50 cm satellite mosaic data in about an hour. The pipeline converts 16‑bit L3C tiles to a consistent 8‑bit via histogram leveling, outputs Cloud Optimized GeoTIFFs, and publishes to an NZ imagery S3 bucket, cutting delivery from 1–2 months to 1–2 weeks.
AI-written summary · about
chapters (8)
- 0:00Introduction and team roles
- 1:48Moving processing to the cloud
- 6:46Cyclone Gabrielle response
- 8:54High-resolution satellite imagery challenge
- 11:45Cloud processing results and lessons
- 13:51Current workflow and QA process
- 16:24Benefits and open resources
- 17:18Q&A session
related talks
05mAlex Kudriashova Interview
01h 04mBuilding Efficient and Frugal Image Data Processing Pipelines
22mRunning Geospatial Workloads on AWS
14mSatellite data monitoring for analytical models
28mAdvancing the future of space in the cloud
34mGeoscience Big Data Computing with Python
39mSatellite Imagery to Machine-Readable Intelligence
24mDigital Earth Pacific Cloud Native Geospatial Infrastructure for Earth Observation