2002 ARCSS All-Hands Workshop

Land cover change in the Western Arctic: Development of a logistic regression model

Monika P. Calef¹, A. David McGuire², T. Scott Rupp³, Edward M. Debevec⁴, Howard E. Epstein⁵, Herman H. Shugart^6
1Department of Environmental Sciences, University of Virginia, 1233 20th Avenue, Apt. 5, Fairbanks, AK, 99701, USA, Phone 907/457-3249, monika@virginia.edu
²Department of Biology and Wildlife, University of Alaska Fairbanks, Fairbanks, AK, USA, ffadm@uaf.edu
³Forest Sciences Department, University of Alaska Fairbanks, Fairbanks, AK, USA, ffsr@uaf.edu
⁴Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, AK, USA, fnemd@uaf.edu
⁵Department of Environmental Sciences, University of Virginia, Charlottesville, VA, USA, hee2b@virginia.edu
⁶Department of Environmental Sciences, University of Virginia, Charlottesville, VA, USA, hhs@virginia.edu

To develop the capability to predict future land cover changes in the Western Arctic, it is important to understand how patterns of land cover change that have occurred in recent decades are associated with climate and fire history. We used logistic regression to develop an empirical model of land cover change in Alaska and adjoining Western Canada. The model predicts land cover based on elevation, aspect, slope, time since last wildfire, fire return interval, drainage class, growing-season air temperature and precipitation. Land cover prediction was limited to four main vegetation types: tundra (including shrubs), deciduous forest, black spruce forest, and white spruce forest. Preliminary results indicate that the vegetation predictions based on the logistic regression model estimate land cover with 70% accuracy in the Western Arctic.

Return to Abstracts page.