Characterization of Vegetation Change in a Sub-Arctic Mire using Remotely Sensed Imagery

Abstract:

Climate change is impacting northern ecosystems through the thawing of the permafrost, which has resulted in changes to plant communities and greenhouse gas emissions, such as carbon dioxide (CO₂) and methane (CH₄). These greenhouse gases are of concern due to their potential feedbacks which create a warmer climate, thus increasing permafrost thawing. Our study focuses on how vegetation type differs in areas that have been impacted by thawing permafrost at Stordalen Mire located in Abisko, Sweden. To estimate change in vegetation communities, field-based measurements combined with remotely sensed image data was used. 75 randomized square-meter plots were measured for vegetation composition and classified into one of five site-types, each representing a different stage of permafrost degradation. New high-resolution imagery (1 cm) was collected using Unmanned Aerial Vehicles (UAV) providing insight into the spatial patterning, characterizations, and changes of these communities. The UAV imagery was georectified using high precision GPS points collected across the mire. The imagery was then examined using a neural network analysis to estimate cover type across the mire. This 2015 cover type classification was then compared to previous UAV imagery taken on July 2014 to analyze changes in vegetation distribution as an indication of permafrost thaw. Hummock sites represent intact permafrost and have lost 21.5% coverage since 2014, while tall gramminoid sites, which indicate fully thawed sites, have increased coverage by 12.1%. A discriminate function analysis showed that site types can be differentiated based on species composition, thus showing that vegetation differs significantly across the thaw gradient. Using average flux rates of CH₄ from each cover type reported previously, the percent of CH₄ emitted over the mire was estimated for 2014 and 2015. Comparing both estimates, CH₄ emissions increased with a flux change of 5604.5 g CH₄/day. Our estimates of vegetation change may be used to parameterize simulation models and create future scenarios of how the vegetation cover will change in response to climate change. Data from this study will also help to explain how the ecology of the subarctic peatlands, now a carbon sink, may be on its way to changing into a source of carbon.