COS 112-3 - Variability in modern fire-climate relationships across boreal forest and arctic tundra biomes

Wednesday, August 9, 2017: 2:10 PM
B116, Oregon Convention Center
Adam M. Young, Forest, Rangeland, and Fire Sciences, University of Idaho, Moscow, ID, Philip Higuera, Ecosystem and Conservation Sciences, University of Montana, Missoula, MT, John Abatzoglou, Geography, University of Idaho, Moscow, ID, Luigi Boschetti, Natural Resources and Society, University of Idaho, Moscow, ID, Paul Duffy, Neptune and Company, Inc., Bellvue, CO and Feng Sheng Hu, Departments of Plant Biology and Geology, Program in Ecology, Evolution, and Conservation Biology, University of Illinois at Urbana-Champaign, Urbana, IL
Background/Question/Methods

Anticipating how fire activity may respond to climatic change in northern-high-latitude biomes is critical for projecting 21st-cenutury Earth-system dynamics. A key challenge in these efforts is understanding how vegetation mediates the relationship between climate and fire activity. Here we investigate how fire-climate relationships vary among vegetation types within and between boreal forest and arctic tundra biomes, spanning North America and Eurasia from 2002-2012. We used global satellite-derived datasets of burned area (MCD64) and landcover (MCD12Q1), a summer fire weather index (FWI), and ecoregion boundaries to quantify fire-climate relationships. We used logistic regression to model the linear relationship between the log-odds of fire occurrence and summer FWI, stratifying each model by landcover type, biome, and continent. We identified where and how vegetation modified fire-climate relationships by comparing slope-parameter estimates among models.

Results/Conclusions

Our statistical models successfully discriminated between burned and unburned pixels (AUC from 0.55-0.77) and identified varying fire-climate relationships across northern-high latitudes. In North American boreal forests, the response of the log-odds of fire occurrence to an increase in FWI was similar among different landcover types, with mean (SE) slope-parameter estimates of 0.13 (0.01), 0.14 (0.01), and 0.13 (0.02) for needle-leaf forests, mixed forests, and woodlands, respectively. In contrast, in eastern-Eurasian boreal forests, the sensitivity of fire activity to changes in FWI was significantly higher in needle-leaf forests (0.19 ±0.01) than in mixed forests (0.13 ±0.01). Fire-climate relationships also varied significantly across the tundra biome, with slope-parameter estimates of 0.37 (0.05) and 0.71 (0.06) in North America and Eurasia, respectively. Our results quantify the spatial variability of fire-climate relationships across boreal forest and tundra biomes, highlighting how vegetation can modify the sensitivity of fire activity to climatic changes. Future work will focus on exploring how these varying relationships impact 21st-century projections of fire activity, and incorporating the effects of non-climatic controls, including anthropogenic ignitions.