No treeline advance over the last 50 years in subarctic western and central Canada and the problem of vegetation misclassification in remotely sensed data

2019-12-27T06:21:25Z (GMT) by Kevin P. Timoney Steven Mamet

In this study we examined (1) whether there has been significant tree cover change over the period 1960–2010 in a 960,000 km2 subarctic study region in western and central Canada, and (2) the degree to which Global Forest Change (GFC) tree cover data agree with other datasets. We compared GFC tree cover to cover estimates from air photos (c. 1960), ground-level plot data (c. 1982–84), annotated low-level oblique photographs (c. 2005–09), and air photo footprints on the World Imagery Base Map (c. 2010). Tree cover changes since 1960 varied by physiographic and ecological regions. Afforestation was modest to non-significant depending on the region. We observed no evidence of northward tree migration. An increase in the areal extent of burned forests, mostly in areas south of the forest-tundra, was the largest change detected. We documented systematic discrepancies between our tree cover estimates and GFC data. GFC underestimates of tree cover typically occurred in areas of low tree density. Areas where GFC data overestimated tree cover were common, especially near the northern limits of trees and in areas dominated by dense or tall shrubs. Predictions of climate-driven vegetation response derived solely from remotely sensed data may not be reliable.