Wildfire spread understanding via scverse/geospatial analogies (burn-time-as-pseudotime, Rothermel-style kernels). POC: Palisades/Eaton/Bordeaux.
  • Jupyter Notebook 91.9%
  • Python 8%
  • Shell 0.1%
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
gjotomcat 9cbec3455e acq: per-fire raster data (4 fires) + multi-resolution xarray cubes
Capture DEM/wind/fuel/SAR/optical GeoTIFFs for palisades/eaton/saumos and new
biscarrosse_2026 (EMSR902, Landes) into data/geodata/<fire>/; AOIs = 1km hull
buffer clipped to GAUL country+county. acq_cube.py harmonizes to 10m/30m/100m/1km
xarray cubes (variable,y,x) as Zarr with burn/aoi masks. dNBR sanity confirmed.
data/geodata/ gitignored; browser tooling + work-tracker updated.
2026-08-18 00:48:37 +00:00
.opencode/skills/browser-automation acq: per-fire raster data (4 fires) + multi-resolution xarray cubes 2026-08-18 00:48:37 +00:00
data CellRank kernel directionality: rho 0.970 (Palisades with_seeds) lifts beyond distance-only geometry 0.422; diffusion doesn't beat Dijkstra within 0.02 threshold; Bordeaux 2-snapshot fallback documented 2026-08-17 04:44:15 +00:00
notebooks notebooks: squidpy intro, real-fire graphs, fire-CA squidpy prototype 2026-08-16 01:49:26 +00:00
plans acq: per-fire raster data (4 fires) + multi-resolution xarray cubes 2026-08-18 00:48:37 +00:00
scripts acq: per-fire raster data (4 fires) + multi-resolution xarray cubes 2026-08-18 00:48:37 +00:00
.env.example fire-scverse: burn-time-as-pseudotime POC + covariate layers (Palisades/Eaton/Bordeaux) 2026-08-07 04:46:02 +00:00
.gitignore acq: per-fire raster data (4 fires) + multi-resolution xarray cubes 2026-08-18 00:48:37 +00:00
AGENTS.md acq: per-fire raster data (4 fires) + multi-resolution xarray cubes 2026-08-18 00:48:37 +00:00
lessons_learned.md acq: per-fire raster data (4 fires) + multi-resolution xarray cubes 2026-08-18 00:48:37 +00:00
next_steps.md acq: per-fire raster data (4 fires) + multi-resolution xarray cubes 2026-08-18 00:48:37 +00:00
opencode.json acq: per-fire raster data (4 fires) + multi-resolution xarray cubes 2026-08-18 00:48:37 +00:00
requirements.lock.txt fire-scverse: burn-time-as-pseudotime POC + covariate layers (Palisades/Eaton/Bordeaux) 2026-08-07 04:46:02 +00:00
results.md CellRank kernel directionality: rho 0.970 (Palisades with_seeds) lifts beyond distance-only geometry 0.422; diffusion doesn't beat Dijkstra within 0.02 threshold; Bordeaux 2-snapshot fallback documented 2026-08-17 04:44:15 +00:00
results_cellrank.md Plan B/C: CellRank kernel directionality + Bordeaux perimeter enrichment 2026-08-17 04:37:21 +00:00
run_background.sh directions A-C unattended run: diffusion kernel, squidpy link, Bordeaux truth probe 2026-08-14 13:45:23 +00:00
scverse_analogies.md fire-scverse: burn-time-as-pseudotime POC + covariate layers (Palisades/Eaton/Bordeaux) 2026-08-07 04:46:02 +00:00
wildfire_spread_basics.md fire-scverse: burn-time-as-pseudotime POC + covariate layers (Palisades/Eaton/Bordeaux) 2026-08-07 04:46:02 +00:00
work_log.md acq: per-fire raster data (4 fires) + multi-resolution xarray cubes 2026-08-18 00:48:37 +00:00