Day-of-burning maps and daily fire ROS for western US and Canadian forested ecoregions 2012-2023
Data files
Jul 03, 2026 version files 2.50 GB
-
BFMS_DOB_archive.zip
2.35 GB
-
CBI_sample.csv
95.85 MB
-
D2Con_sample.csv
53.20 MB
-
Prop.HS_sample.csv
4.48 MB
-
README.md
6.31 KB
Abstract
Most area burned in western and boreal North American forests is the product of a relatively small number of extreme, rapidly spreading fire events that burn disproportionately large areas. But beyond area burned, do these events produce distinct postfire ecological outcomes? We measured linear daily fire growth rates and assessed relationships between fire spread, burn severity, and metrics of postfire forest landscape resilience from 3499 individual fires that occurred between 2012 and 2023 in conifer-dominated forests of Canada and the western US. We find that as linear fire spread rate increases, so too does fire severity and the proportion of area burned at high severity. Distance to surviving live tree seed source also increases with fire spread rate. Thus, not only do the fastest spreading fires burn disproportionately large areas, but these areas have less remaining tree cover and a reduced capacity for recovery, setting the stage for shifts in forest types and towards non-forest vegetation. Given anticipated increases in fast events as fire seasons lengthen and weather becomes more fire-conducive, our findings point toward growing needs to implement effective strategies to reduce undesired effects of fast-growing wildfires, manage postfire landscapes to sustain valued forest ecosystem functions, and foster social-ecological adaptation.
Dataset DOI: 10.5061/dryad.rjdfn2zsp
Description of the data and file structure
Data were developed to assess relationships between wildfire rate of spread (ROS) and effects measured from postfire landscapes including burn severity, proportion of area burned at high severity (≥ 95% canopy mortality) and distance to surviving conifer tree seed sources. We sampled wildfires >400 ha that occurred in forested ecoregions of Canada and the western US between 2012 and 2013.
Files and variables
File: CBI_sample.csv
Description:
Variables
- fire.id: a unique code for each wildfire
- DOB: day of burning
- fire.year: year of fire occurrence
- country: USA or Canada
- CBI: satellite-measured composite burn index, a measure of fire effects to vegetation and soils
- dob.area.ha: the area burned on a given day by a given wildfire
- dob.max.linear.spread.m: the maximum linear distance the wildfire traveled during a day of burning (m)
- NA_L1NAME: level one ecoregion
- NA_L2NAME: level two ecoregion
- speed: fast or slow, based on whether or not it exceeded an ecoregional threshold for linear distance travelled during a 24-hr period
File: D2Con_sample.csv
Description:
Variables
- fire.id: a unique code for each wildfire
- DOB: day of burning
- fire.year: year of fire occurrence
- country: USA or Canada
- dob.area.ha: the area burned on a given day by a given wildfire
- dob.max.linear.spread.m: the maximum linear distance the wildfire traveled during a day of burning (m)
- NA_L1NAME: level one ecoregion
- NA_L2NAME: level two ecoregion
- speed: fast or slow, based on whether or not it exceeded an ecoregional threshold for linear distance travelled during a 24-hr period
- point.ID: sample point identification number
- HS.patch.ID: high severity (CBI ≥ 2.25) patch identification number
- HS.patch.area.ha: area of the high severity patch
- distance.m: distance from a random pixel within the patch to the nearest conifer pixel that burned at CBI <2.25 or was not within the wildfire perimeter
File: Prop.HS_sample.csv
Description:
Variables
- fire.id: a unique code for each wildfire
- DOB: day of burning
- fire.year: year of fire occurrence
- country: USA or Canada
- dob.area.ha: the area burned on a given day by a given wildfire
- dob.max.linear.spread.m: the maximum linear distance the wildfire traveled during a day of burning (m)
- NA_L1NAME: level one ecoregion
- NA_L2NAME: level two ecoregion
- speed: fast or slow, based on whether or not it exceeded an ecoregional threshold for linear distance travelled during a 24-hr period
- total.HS.area.ha: the total area that burned at high severity (CBI ≥ 2.25) within the day-of-burning polygon
- DOB.Conifer.area.ha: area of conifer forest prior to the wildfire within the day-of-burning polygon
- Prop.HS: the proportion of conifer that burned at high severity (CBI ≥ 2.25)
File: BFMS_DOB_archive.zip
Description: Subfolders are named by fire.id. For U.S. fires, fire.id is the MTBS Event ID. For Canadian fires, it is a concatenation of the fire year and NBAC ID (e.g., 2023_435 denotes NBAC ID 435 from 2023), since NBAC IDs are reused across years.
Each subfolder contains the following files, all prefixed with the corresponding fire.id:
- _dob.tif: Day-of-burning raster (30 × 30 m resolution) generated following Parks et al. (2014). Pixel values represent the day of year on which each pixel burned. NoData values are NA.
- _shifted.dob.tif: Day-of-burning raster (30 × 30 m resolution) generated following Parks et al. (2014), with early morning fire detections (00:00–06:00) reassigned to the previous day. Pixel values represent the day of year on which each pixel burned. This file is only generated for fires with overnight hotspot detections. NoData values are NA.
- _dob.gpkg: Day-of-burning polygons generated from fire.id_shifted.dob.tif when available; otherwise generated from fire.id_dob.tif. Attributes include:
- fire.year: Year the fire occurred.
- country: Country where the fire occurred.
- fire.id: Fire identifier (MTBS Event ID for U.S. fires; year_NBAC ID for Canadian fires).
- DOB: Day of year the polygon burned.
- dob.area.ha: Area of the day-of-burning polygon (ha).
- _linear_spread_rev.tif: Raster (30 × 30 m resolution) generated following Harris et al. (2025). Pixel values represent the distance (m) of fire spread from the previous day's fire perimeter or the earliest hotspot detection. NoData values are NA.
Files are provided in region-specific equal-area coordinate reference systems: EPSG:5070 (NAD83 / Conus Albers) for CONUS, EPSG:3338 (NAD83 / Alaska Albers) for Alaska, and ESRI:102001 (Canada Albers Equal Area Conic) for Canada.
Code/software
All data can be analyzed in R (as of May 2026 the most recent version 4.6.0). The recommended package for spatial analysis is terra. The recommended package for generalized linear mixed-effects models is glmmTMB. The recommended package for visualization is ggplot2.
Access information
Data was derived from the following sources:
- Fire perimeter data were provided by Monitoring Trends in Burn Severity (MTBS) at https://doi.org/10.5066/P9IED7RZ and the Canadian National Burned Area Composite (NBAC) at https://gee-community-catalog.org/projects/nbac/
- Day of Burning was calculated using Moderate Resolution Imaging Spectroradiometer (MODIS) obtained from the NASA EOSDIS Land Processes Distributed Active Archive Center at https://doi.org/10.5067/FIRMS/MODIS/MCD14DL.NRT.0061 and Visible Infrared Imaging Radiometer Suite (VIIRS) daily fire detection data obtained from LP DAAC at https://doi.org/10.5067/FIRMS/VIIRS/VNP14IMGT_NRT.002.
- Burn severity was estimated using methods from Parks et al. 2019 https://doi.org/10.3390/rs11141735
