Can Permafrost Soil Thaw be Characterized by Hyperspectral Reflectance and Plant Community Structure?

Abstract

I investigated (1) whether stages of permafrost thaw were consistently associated with plant community composition and other land surface characteristics; (2) whether those different land surface characteristics could be consistently distinguished with remote sensing tools in a sub-arctic mire. I utilized plant area cover and topography to identify five distinct site-types as being characteristic of different stages of permafrost thaw, and 50 one square-meter plots were measured for species-specific area cover and pole-based hyperspectral reflectance. A Tukey-HSD comparison test showed that plant functional group richness decreased with permafrost thaw, and could readily be used to differentiate between stages of thaw. A discrete, stepwise canonical classification function with bootstrap cross validation showed a mean classification error rate of 7.3% +/- 7.3% (6.8%-9.65% 95% Confidence Interval). These results showed successful ground-truthing methods for regional-scale landscape classification, allowing for high temporal and spatial resolution of circumpolar permafrost thaw monitoring.