Climate associations with North American wildfires between 1986-2013

Release Date:

This article is part of the Spring 2019 issue of the Earth Science Matters Newsletter.

maps of topography, vegetation, and burn area in North America

Figure 1. The regional model domains. Left column: North America 50 km grid; right column: 15 km grids, clockwise from upper left: Pacific Northwest, Northern Rocky Mountains, Southern Rocky Mountains, and Pacific Southwest. Top row: model elevation; middle row: prescribed vegetation; bottom row: all fires > 1 ha in burn area set by lightning (left) and human and unknown causes (right) for the period 1986-2013. Vegetation types based on USGS Global Land Cover Characterization data set (Loveland et al., 1999 and are similar in resolution to the Level II Ecoregions). (From Figure 1 in Hostetler et al., 2018)

(Credit: From Figure 1 in Hostetler et al., 2018. Public domain.)

Wildfire is a major source of ecosystem disturbance with increasingly wide-ranging effects on society. The occurrence and distribution of wildfires are controlled by the state of fuels in burnable biomes, atmospheric circulation, energy and water balances of the Earth’s surface, land use and land use change, and fire suppression. USGS scientists and colleagues from the University of Oregon compiled a new fire dataset and combined it with regional climate simulations to investigate the association of wildfire with the seasonal cycle and interannual climate variability. This effort is a product of a longer-term USGS project that combines data analysis and modeling to improve understanding of how changing land use and climate affect the Earth system over a wide range of temporal and spatial scales.

The study used the regional climate model, RegCM3, to provide high-resolution, continuous, internally consistent climate fields over a 50 x 50 km grid covering most of North America, along with four 15 x 15 km grids covering most of the West (Figure 1). Joint analysis of wildfire data and RegCM3 simulations diagnosed interactions and feedbacks among atmospheric circulation, surface climate, energy balance, and soil moisture, along with their controls on wildfire.

The daily fire dataset (date, point of origin, and ultimate size) combined data from United States and Canada from 1986 to 2013. The dataset contains over 2 million wildfires from all ignition sources that burned more than 100 million hectares over the period (see Figure 1). The daily data were aggregated into monthly totals and the RegCM3 grid cells.

Lightning-set fires amount to 24-56% of the total number of fires and 66-82% of the total burned area in the four domains in the western US. Although wildfires occur throughout the year, fire activity increases in the Southwest and boreal forest in April and peaks May through August, when lightning from thunderstorms is common. Fall through spring, fires set by non-lightning ignition sources are found primarily over the south, southeast, and Florida. Human-set fires extend the fire season into months and areas in which climate conditions are not otherwise conducive to fire.

The model runs clearly illustrate the association of wildfire with the seasonal climate cycle and how wildfire occurrence is influenced by variability in mid-tropospheric circulation, receipts of solar radiation, precipitation, surface energy balance, and soil moisture (Figure 2). Furthermore, the analysis differentiates the response of North American wildfire to warm and cold phases of El Niño–Southern Oscillation events. During El Niño, wet conditions from the Southwest across the Gulf suppressed fire activity while early season fire activity was enhanced during warm, dry La Niña events in these regions.

maps of net radiation and soil moisture across North America

Figure 2. March through September composite anomalies for net radiation (B) and soil moisture (H) for high fire years (first row) and low fire years (bottom row) in the Northern Rockies 15-km domain. The colored anomaly maps are overlaid with the area burned by lightning and human and unknown ignition sources. The composite anomalies are computed by subtracting the 1986 2013 average from the average of the 9 highest (lowest) fire years. (From Figure 5 in Hostetler et al., 2018)

(Credit: From Figure 5 in Hostetler et al., 2018. Public domain.)

This modeling approach can be extended to test hypotheses and investigate paleo and future climate-wildfire interactions under differing climates and vegetation distributions. Additionally, it may inform wildfire potential and hazard models that are used to characterize seasonal wildfire outlook assessments.

The paper, “Atmospheric and surface climate associated with 1986‑2013 wildfires in North America” was published in the Journal of Geophysical Research Biogeosciences and is available here:

<< Back to Spring 2019 Newsletter

Related Content