Quantifying Large-Scale Patterns of Forest Fire Occurrence

NCSU Department of Forestry and Environmental Resources; USDA Forest Service Forest Health Monitoring Program and Forest Health Protection

SUMMARY: Forest fire occurrence outside the historic range of frequency and intensity can result in extensive economic and ecological impacts. Quantifying and monitoring broad-scale patterns of fire occurrence across the United States can help provide a fuller understanding of the ecological and economic impacts of fire, and of the appropriate management and prescribed use of fire. Specifically, large-scale assessments of fire occurrence can help identify areas where specific management activities may be useful, or where research into the ecological and socioeconomic impacts of fires may be necessary.

It is a challenge, however, to summarize fine-scale fire occurrence data (from the Moderate Resolution Imaging Spectroradiometer [MODIS] Active Fire Detections for the United States database) in a way that is useful for national forest health monitoring efforts. To address this challenge, an Eastern Threat Center cooperator at North Carolina State University developed an approach for quantifying statistically significant geographic hot spots of fire occurrence at a continental scale. The approach employs the Getis-Ord Gi* spatial clustering statistic of a nationwide lattice of hexagonal cells (ideal for use in spatial neighborhood analysis) to find areas with more satellite-derived MODIS fire occurrence detections than expected by chance. The numbers of fire occurrences per area of forest (fire occurrence densities) are also summarized by ecoregion, both for the conterminous United States and for Alaska (and by island in Hawaii and Puerto Rico). The value for each region for each year is then compared with the values from past years to assess whether fire occurrence densities are higher, lower or the same relative to all the previous years for which data were collected.

fire_hotspots_2017_061818.pngHotspots of fire occurrence across the conterminous United States for 2017. Values are Getis-Ord Gi* scores, with values greater than 2 representing significant clustering of high numbers of fire occurrences. (No areas of significant clustering of low numbers of fire occurrences, -2, were detected). Click image to enlarge.

fire_means_SD_deviations_061818.png(A) Mean number and (B) standard deviation of forest fire occurrences per 100 km2 (10 000 hectares) of forested area from 2001 through 2016, by ecoregion section within the conterminous 48 States. (C) Degree of 2017 fire occurrence density excess or deficiency by ecoregion relative to 2001–16 and accounting for variation over that time period.
Click image to enlarge.

This project is supported by Eastern Threat Center funding.

STATUS: Ongoing

PROGRESS: The quantification of large-scale patterns of forest fire occurrence, using MODIS Active Fire detections data, has become a standard component within the annual Forest Health Monitoring national reports. Chapters presenting the results are included in the 2008-2018 editions. A paper published in the journal Landscape Ecology describes this approach in detail. This research has been described in a number of other publications and presentations.


Forest Health Monitoring National Technical Reports

MODIS Active Fire Mapping Program: Continental United States Fire Detection GIS Data

Kevin Potter, NCSU Department of Forestry and Environmental Resources, kevinpotter@fs.fed.us or (919) 549-4071

Updated June 2018

Document Actions
Personal tools