Retrogressive thaw slumps (RTS) are a prevalent type of mass wasting feature in permafrost regions that can mobilize massive amounts of soil carbon and nutrients as warming collapses ice-rich permafrost terrain. Undisturbed tundra vegetation is abruptly transported downslope, causing newly exposed patches of collapsed organic material and bare ground for plant colonization, which may simultaneously follow a combination of primary and secondary successional trajectories. Collapsed RTS terrain creates a heterogeneous mosaic of various fine-scale microtopographic units that influence the redistribution of water, nutrients, and carbon throughout the thaw slump, potentially influencing the spatial patterns of vegetation growth and composition following disturbance.
However, due to varying patterns and rates of RTS expansion throughout the Arctic, the underlying controls on RTS morphological change and the resulting patterns of vegetation succession remain unclear. This dissertation aims to model the historical patterns and rates of RTS expansion, which will be used to advance our predictive understanding of the patterns, controls, and implications of vegetation succession across northern Alaska. I modeled RTS expansion by estimating the rates of surface deformation/collapse through a combination of 70+ years of historic image analysis (i.e., RTS area change) and associated LiDAR observations (i.e., RTS volumetric change).
These data were used to train a machine-learning (ML) model to predict both RTS area and volumetric change using topographic and climatic variables. To determine the impacts of RTS on vegetation succession, I used another ML model to map fine-scale distributions of dominant species and plant functional types with UAS-hyperspectral imagery. Because such airborne hyperspectral datasets are extremely limited and important for discriminating between vegetation types, I expand this research across northwestern Alaska, by simulating airborne hyperspectral with multispectral data. Specifically, I investigate how fine-scale vegetation patterns respond to RTS change at sites with varying morphology and climate conditions by mapping plant functional types and species with simulated hyperspectral imagery and ML models.
This research provides new insights into the various local to regional-scale factors that influence patterns of vegetation succession in response to permafrost disturbances across permafrost landscapes in Alaska.
