Williams et al. (2025) The Western United States MTBS-Interagency database of large wildfires, 1984–2024 (WUMI2024a)

Identification

• Journal: Earth system science data
• Year: 2025
• Date: 2025-12-19
• Authors: Park Williams, Caroline S. Juang, Karen C. Short
• DOI: 10.5194/essd-17-7359-2025

Research Groups

• Department of Geography, University of California, Los Angeles, CA, USA
• Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, CA, USA
• Department of Earth and Environmental Sciences, Columbia University, New York, NY, USA
• USDA Forest Service, Rocky Mountain Research Station, Missoula, MT, USA

Short Summary

This paper introduces WUMI2024a, a quality-controlled, publicly available database of 22,234 large (≥1 km²) wildfires in the western United States from 1984–2024, compiled by merging seven government datasets to provide comprehensive wildfire occurrences, perimeters, and burned-area maps. It addresses limitations of existing datasets by offering improved spatial and temporal coverage and enhanced data quality for wildfire research and modeling.

Objective

• To develop and present the WUMI2024a database, a comprehensive, quality-controlled, and publicly available dataset of large (≥1 km²) wildfire occurrences, perimeters, and burned-area maps for the western United States from 1984–2024, addressing limitations of existing datasets for wildfire research and modeling.

Study Configuration

• Spatial Scale: The 11 westernmost states of the contiguous United States (Arizona, California, Colorado, Idaho, Montana, New Mexico, Nevada, Oregon, Utah, Washington, and Wyoming).
• Temporal Scale: 1984–2024

Methodology and Data

• Models used: The study primarily involved a data compilation, merging, and quality-control methodology rather than predictive models. This included:
  • Automated detection and removal of duplicate wildfire entries based on spatial proximity (e.g., within 100 km, 50 km, or 25 km), temporal proximity (e.g., within 5 days, 3 days, or 1 day), matching names (allowing for variations), and size similarity (within 10%).
  • Rigorous visual inspection and manual correction of fire lists and perimeter maps to identify additional duplicates and errors.
  • Prioritization rules for merging data from different sources (MTBS, CalFire, WFIGS, USGS, IAFPH) to ensure the highest quality perimeter data.
  • Estimation of circular perimeters for fires lacking observed perimeter data, with adjustments to exclude unburnable areas (open water, permanent ice, barren ground) using land cover data.
  • Production of 1 km resolution fractional burned area maps for each fire.
• Data sources:
  • Monitoring Trends in Burn Severity (MTBS) project (Landsat-based, 30 m resolution burned area maps for fires > 4.04 km², 1984–2024).
  • US Forest Service (USFS) Fire Program Analysis Fire Occurrence Dataset (FPA FOD), version 6 (point-based ignition locations, dates, and sizes, 1992–2020).
  • Wildland Fire Application Information Portal (WFAIP) (point-based federal agency reports, used for 1984–1991).
  • California Department of Forestry and Fire Protection (CalFire) Resource Assessment Program (FRAP) (fire perimeters for California, 1898–2024).
  • Wildland Fire Interagency Geospatial Services (WFIGS) Group (fire perimeters, 2018–present).
  • US Geological Survey (USGS) compiled dataset of US fire perimeters (from 40 sources, 1800s–2020).
  • National Interagency Fire Center (NIFC) Interagency Fire Perimeter History (IAFPH) (fire perimeters, pre-1900s–2023).
  • USGS annual National Land Cover Database (NLCD) (for land cover types).
  • Ruefenacht et al. (2008) 250 m resolution map of forest classifications.

Main Results

• The WUMI2024a database comprises 22,234 wildfires ≥1 km² in the western United States, covering the period 1984–2024.
• Observed fire perimeters are available for 71% of the fires in the database.
• Over 47% of all fires (more than 10,300 wildfires) have perimeters and burned area maps derived from 30 m satellite data provided by the MTBS project.
• An additional 24% of fires include perimeter observations from non-MTBS sources.
• Only 29% of fire occurrences lack perimeter observations, for which circular perimeters are assumed; these fires account for less than 5% of the total burned area in the database.
• The proportion of fires without perimeter data is lowest in California/Nevada (20%) and highest in the Four Corners region (41%).
• The proportion of fires lacking perimeter observations and their associated burned areas significantly declined from 1984 through 2020, with all fires having perimeters after 2020.
• The database includes two lists of wildfire events: one prioritizing parent fires (21,940 fires) and an alternative list prioritizing sub-fires (22,234 fires) for improved accuracy of ignition dates and frequencies.
• Ignition locations for 16.8% of FPA FOD or WFAIP fires were adjusted, with a median adjustment distance of 406 m and 95% of adjustments within 6.6 km.
• Monthly 1 km resolution maps of total area burned across the western US are provided.

Contributions

• Presents the first comprehensive, quality-controlled database of large (≥1 km²) wildfires for the western United States that spans the full period of 1984–2024.
• Significantly increases the proportion of fires with observed perimeters by merging and quality-controlling seven government datasets, improving upon previous versions and individual datasets.
• Provides high-resolution (30 m) satellite-derived burned area maps for over 47% of fires, accurately accounting for unburned areas within fire perimeters.
• Offers 1 km resolution maps of burned area and observed or estimated perimeters for each fire, enhancing spatial detail for various applications.
• Designed for easy updates, ensuring its continued utility for future research.
• Enables improved assessments of wildfire frequency and area burned trends, facilitates statistical analyses and modeling of wildfire activity, and supports linkage to independent observations of fire progression, severity, and hydrology.
• Addresses critical limitations of existing wildfire datasets, including inconsistent temporal coverage, limited size ranges, and lack of comprehensive perimeter data and quality control.

Funding

• Zegar Family Foundation
• John D. and Catherine T. MacArthur Foundation
• Gordon and Betty Moore Foundation (11974 and 13283)
• UCLA Sustainable LA Grand Challenge
• USGS Southwestern Climate Adaptation Science Center (G24AC00611 and G24AC00080)
• US National Park Service (P24AC00743-00)

Citation

@article{Williams2025Western,
  author = {Williams, Park and Juang, Caroline S. and Short, Karen C.},
  title = {The Western United States MTBS-Interagency database of large wildfires, 1984–2024 (WUMI2024a)},
  journal = {Earth system science data},
  year = {2025},
  doi = {10.5194/essd-17-7359-2025},
  url = {https://doi.org/10.5194/essd-17-7359-2025}
}