The Western United States Wildfire Severity (WestWiSe) Database

Authors

Mitchell Hung & A. Park Williams

For questions or issues, please contact:
Mitchell Hung (mjhung@stanford.edu)

Source Wildfire Perimeters

WUMI fires are a merged perimeter set combining MTBS, CALFIRE FRAP, WFAIP, and WFIGS, covering 1985–2024. Perimeters are sourced from the WUMI2024a dataset: https://datadryad.org/dataset/doi:10.5061/dryad.63xsj3vd4

Perimeter selection (`index.geojson`)

The master fire index lives at `index.geojson`.
Each row provides two fire properties:

- `fireid`: `{YYYYMMDD}_{lat*1000}_{lon*1000}`, same id used in the `.tif` filenames (see below)
- `agency`: which source dataset supplied the perimeter

For each `fireid` in the WUMI2024a wildfire list, the geometry was taken from the first available perimeter shapefile in this priority order:

`mtbs` → `calfire` → `usgs` → `iafph` → `circle`

`circle` is a fallback: for fires with no official polygon in any of the above sources, a circular buffer was built from the reported point location and burned area.

---

1. Annual zip archives

**Files:** `{YEAR}.zip`, 1985–2024.

Each zip bundles the GeoTIFFs for each fire in that year. 

Dataset Contents

Each `.tif` is a EPSG:4326, ~30-m GeoTIFF. Band definitions and scale factors are provided in `band_info.csv`. To retrieve original float values, divide by the appropriate scale factor.

Data Description

**Composite Burn Index (CBI)** was estimated using a Random Forest regression model per the method described in Parks et al. 2019.

---

2. `severity_burnarea_epsg5070_monthly.nc`

Monthly, forest-masked burned area by severity class, gridded to 1 km, EPSG:5070, 1985–2024.

- Severity thresholds: CBI low [0.1,1.25), moderate [1.25,2.25], high >2.25; dNBR low [41,176], moderate [177,366], high ≥367; RBR low [30,149], moderate [150,299], high ≥300
- Forest mask: NLCD classes 41 (deciduous), 42 (evergreen), 43 (mixed forest), applied using the year before the fire-year.

How it was created

For each year/month, all GeoTIFFs were grouped onto the same 30-m resolution grid. For each pixel, severity masks were max-accumulated (so a pixel double-covered by two overlapping fires wasn't double-counted), with priority high > mod > low. Then, pixel counts were summed by severity class within a larger 1-km grid and values were scaled to km² by the native pixel area (27 m²).

---

References

Parks, S.A.; Holsinger, L.M.; Koontz, M.J.; Collins, L.; Whitman, E.; Parisien, M.-A.; Loehman, R.A.; Barnes, J.L.; Bourdon, J.-F.; Boucher, J.; et al. Giving Ecological Meaning to Satellite-Derived Fire Severity Metrics across North American Forests. Remote Sens. 2019, 11, 1735. https://doi.org/10.3390/rs11141735

United States Geological Survey, Annual National Land Cover Database (NLCD) Collection 1 Products. U.S. Geological Survey. https://doi.org/10.5066/P94UXNTS.