Yirgacheffe: a declarative approach to geospatial data
summary
Yirgacheffe is a Python declarative geospatial library that expresses raster ops as numpy‑style expressions, performs lazy evaluation, and automatically chunks and parallelises work to avoid manual tiling and bookkeeping. It powers ecology pipelines with 120,000 area-of-habitat runs and “life” maps built from global rasters (400,000×200,000 pixels, about 150 GB per layer, ~300 GB total), supports cloud masking via scene classification, and skips irrelevant tiles, but remains limited by Python multiprocessing inefficiencies and repeated evaluation of large intermediate rasters.
AI-written summary · about
chapters (10)
- 0:00Yirgacheffe introduction
- 1:05Declarative geospatial processing
- 2:13Installing Yirgacheffe
- 3:57Removing clouds with layers
- 5:44Optimizing raster evaluations
- 6:49Calculating habitat areas
- 9:19Handling large datasets
- 10:48Python memory limits
- 11:29Beyond Python constraints
- 11:50Q&A
related talks
20mCloud Native Patterns for Geospatial Analytics and AI
34mGeoscience Big Data Computing with Python
41mUsing Python to Build a GIS Data Pipeline for Rural-Urban Classification
19mThe Power of GeoDjango
14mMosaic: A Framework for Geospatial Analytics at Scale
46mGeoSpatially enabling your Big Data Apache clusters with LocationTech
22mTake your Business to a New Geography
48mEmpowering AI Innovation Through Open Geospatial Ecosystems