The California Interagency Wildfire Severity (CaliWiSe) Database

OVERVIEW

This dataset contains multi-band raster images representing burn severity and spectral response across wildfire perimeters in California. Each image corresponds to a unique fire event and includes several bands including burn severity, moisture, and vegetation indices.

The data were generated using Google Earth Engine, primarily from Landsat 4–8 Surface Reflectance Tier 1 collections data. All images are scaled and exported as float-encoded multi-band GeoTIFFs.

AUTHORS
Mitchell J. Hung & A. Park Williams

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

SOURCE WILDFIRE POLYGONS
MTBS and CALFIRE FRAP perimeters, 1985–2024.

FILE STRUCTURE / DIRECTORY GUIDE

Each year’s data are organized into compressed archives corresponding to the source fire perimeter database:

├── calfire_YYYY.zip → CAL FIRE FRAP wildfire perimeters for year YYYY
├── mtbs_YYYY.zip → California MTBS wildfire perimeters for year YYYY
├── band_info.csv → Band names and scale factors for all exported layers
├── README.md → Dataset documentation

Each .zip archive contains one GeoTIFF per fire event within that year’s source dataset:

DATASET CONTENTS
Each exported .tif file contains the following bands:

CBI - Composite Burn Index (bias-corrected model output) [scale: 100]
dnbr - Delta Normalized Burn Ratio [scale: 1000]
rbr - Relativized Burn Ratio [scale: 1000]
rdnbr - Relativized Delta Normalized Burn Ratio [scale: 1000]
dndvi - Delta Normalized Difference Vegetation Index [scale: 1000]
devi - Delta Enhanced Vegetation Index [scale: 1000]
dndmi - Delta Normalized Difference Moisture Index [scale: 1000]
dmirbi - Delta Mid-Infrared Bi-Spectral Index [scale: 1000]
post_nbr - Post-fire Normalized Burn Ratio [scale: 1]
post_mirbi - Post-fire Mid-Infrared Bi-Spectral Index [scale: 1000]

Note: To retrieve original float values, divide by the appropriate scale factor.

DATA DESCRIPTION

Composite Burn Index (CBI)
Estimated using a Random Forest regression model (Parks et al. 2019).

Pixel Masking
The Landsat QA_PIXEL band and Hansen Global Forest Change water mask were used to remove pixels obscured by:
- Cloud
- Cloud shadow
- Water
- Snow

Spectral Indices
Collections: Landsat 4, 5, 7, and 8 (Surface Reflectance Tier 1)

Formulas:
NDVI = (NIR - RED) / (NIR + RED)
NBR = (NIR - SWIR2) / (NIR + SWIR2)
RBR = (SWIR1 - RED) / (SWIR1 + RED)
NDMI = (NIR - SWIR1) / (NIR + SWIR1)
EVI = 2.5 * ((NIR - RED) / (NIR + 6RED - 7.5BLUE + 1))
MIRBI = (SWIR1 - SWIR2) / (SWIR1 + SWIR2)

Temporal definitions:
Pre-fire = Mean fire season value for the year prior to the fire.
If flagged (cloud/shadow/water), the second year prior was used.
Post-fire = Mean fire season value for the year following the fire.
If flagged, the second year following was used.
Delta = Post-fire minus Pre-fire.

EXPORT DETAILS

Projection : EPSG:4326
Pixel resolution : 30 meters
Region : Fire perimeter bounding boxes
Format : GeoTIFF
Data type : Float32
Date range : 1985–2024

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. (2019).
"Giving Ecological Meaning to Satellite-Derived Fire Severity Metrics across North American Forests."
Remote Sens. 11, 1735. https://doi.org/10.3390/rs11141735