Plaknit is my thesis log for processing PlanetScope mosaics at scale—this page now centers on the workstreams, documentation, and talks that grow out of that repo.
Every commit in plaknit captures how I'm building the Satellite Anthrax Surveillance thesis—from PlanetScope ordering and QA to the automation that makes those mosaics repeatable.
GDAL-powered masking cleans each strip with its corresponding UDM raster before Orfeo Toolbox mosaics everything together using RAM hints tuned for multi-hundred scene jobs.
Lightweight plaknit.analysis utilities compute normalized-difference indices straight from raster files or NumPy arrays, so I can benchmark vegetation stress without spinning up new notebooks.
The plaknit.classify module handles training and inference against Planet stacks, keeping polygons, model binaries, and raster outputs in sync for disease-surveillance mapping.
The MkDocs site and supporting labs double as my own field manual—what I teach is what runs plaknit.
plaknit.mosaic, plaknit.analysis, and plaknit.classify. Browse functionsI share plaknit progress and thesis findings with the geospatial community to pressure-test the workflow and invite collaborators.
If you want to adapt plaknit to a new landscape or need a workshop on PlanetScope processing, I would love to talk through it.