Refining evergreen forest categories in historic, present, and future landcover datasets for Anchorage, Fairbanks, and Whitehorse to better identify fire risks

Keywords: fire, modeling, environment, landcover, Alaska, boreal forest, Yukon, climate change, GIS, ABoVE, Flammap
Abstract Type: Paper Abstract

Abstract

Climate is changing especially rapidly at higher latitudes impacting forest growth, fire and insect disturbance, recovery, and successional pathways. Especially alarming to communities in the boreal forest of Alaska and the Yukon are expanding and more frequent wildfires which are driven by warmer, drier summers. However, detailed land cover data that could be used to analyze fire danger and changes through time do not exist. The landcover data with the best temporal coverage for the Western Arctic are NASA’s Arctic-Boreal Vulnerability Experiment (ABoVE), which identify 15 plant functional types annually from 1984 to 2014 at 30 m resolution. As all regional-scale remote-sensing based landcovers, this dataset includes only 3 types of forest: deciduous, evergreen, and mixed. We developed compound modeling approaches to separate highly flammable black spruce, lodgepole pine, and subalpine fir stands from less flammable other evergreens like white spruce and hemlock using community wildfire protection plans, decadal climate, and several other predictors at decadal instances from 1984 to 2054. For future fires, the wildfire prediction model FlamMap’s burn probability module was used to randomly place ignition points which grew to plausible burn scars based on landscape flammability. Modeled wildfire size and number were based on a Monte Carlo analysis of historic fires. After disturbance, vegetation was reset following observed succession in the ABoVE data record. Putting all these approaches together, we were able to produce refined land cover datasets for our three study areas that can be used by communities to assess future wildfire risk.