Intensifying Fire Season Aridity Portends Ongoing Expansion of Severe Wildfire in Western US Forests

Abstract

Area burned by wildfire has increased in western US forests and elsewhere over recent decades coincident with warmer and drier fire seasons. However, high-severity fire-fire that kills all or most trees-is arguably a more important metric of fire activity given its destabilizing influence on forest ecosystems and direct and indirect impacts to human communities. Here, we quantified area burned and area burned severely in western US forests from 1985 to 2022 and evaluated trends through time. We also assessed key relationships between area burned, extent and proportion burned severely, and fire season climate aridity. Lastly, using the strong relationships between fire season aridity and both area burned and area burned severely, we predicted future fire activity under ongoing warming. While annual area burned increased 10-fold over our study period, area burned severely increased 15-fold. Disproportionate increases in severe fire occurred across a wide range of forest types from 1985 to 2022. Importantly, we found that the proportion of area burned severely increased with fire extent at the scale of individual fires and total annual area burned. The relationships between fire season aridity and fire were strong, and our models predicted further increases in fire activity, leading to 2.9- and 4-fold increases in area burned and area burned severely, respectively, under mid-21st century climate. Without a substantial expansion of management activities that effectively reduce fire severity (e.g., thinning of understory and fire-intolerant trees combined with prescribed fire), wildfires will increasingly drive forest loss and degrade ecosystem services including carbon storage, biodiversity conservation, and water yield, with major impacts to human communities.

Keywords: climate change; fire–climate relationships; forests; high‐severity wildfire; wildfire; wildfire trends.

FIGURE 1

Graphical depiction of high–severity fire, or stand‐replacing fire, in the context of our study (a–c). Here, the forest (a) experiences increasing flame lengths from left to right, corresponding with increasing canopy fuel consumption, tree mortality, and fire severity (b). High–severity fire corresponds to a composite burn index (CBI) ≥ 2.25 (c). Study area showing forests in the western US (d).

FIGURE 2

Area burned and area burned severely from 1985 to 2022 in western US forests (a). Smooth lines show the Theil‐Sen model fit (back‐transformed) which is used to calculate the overall increase in area burned and high–severity area burned from 1985 to 2022. Proportion burned at high severity as a function of annual area burned (b).

FIGURE 3

Key relationships among fire size, proportion burned severely, and time. Proportion burned at high severity as a function of fire size (binned; interquartile range shown by vertical lines) (a) showing how larger fires tend to have proportionally more high–severity fire. Median fire size by year (b); fire size increases through time (p < 0.001). Only forested pixels are used in the calculation of fire size, thereby maintaining consistency among all analyses; fires with < 100 ha (1 km²) of forest cover are excluded. Example of a smaller fire with low proportion burned severely (0.11): The 1999 High Complex Fire burned 1332 ha in California (c). Example of a very large fire with high proportion burned severely (0.59): The 2020 North Fork Complex burned 128,101 ha in California (d).

FIGURE 4

The relationships between area burned and fire season aridity (a) and area burned severely and fire season aridity (b). Smooth black lines (a, b) show back‐transformed model fit between log‐transformed area burned (and area burned severely) and fire season aridity. These models were used to project mid‐21st century outcomes (Figure 5).

FIGURE 5

Modeled recent (1985–2022) and mid‐21st century area burned (a, b) and area burned severely (c, d). In panels (a, c), the black triangles and horizontal lines represent the mean and median, respectively; the black dots show 2020 and 2021 predictions, and the open circle (d) represents an outlier for the mid‐21st century area burned severely prediction. Note that the modeled mid‐21st century mean annual area burned and mean annual area burned severely approximately matches fire activity in 2020 and 2021 (a, c).