Integrated modeling and observations of ice wedge degradation and thermokarst pool expansion

Abstract:

Ice wedge melting and thermokarst pool expansion have accelerated in Arctic landscapes over the past several decades. Recent studies indicate that positive feedbacks on permafrost degradation, attributable to the presence of thermokarst pools, may cause pan-Arctic permafrost thaw to progress at rates that far exceed predictions from current Earth system models. However, the extent to which these feedbacks accelerate permafrost thaw at large spatial scales is difficult to quantify, due the highly uneven and local response of ice wedges to climate change. To address this gap, we have paired fine-scale, physics-intensive, numerical simulations of the ground thermal regime in polygonal terrain with analysis of enormous, high-resolution remote sensing datasets. Results from our numerical experiments indicate that two attributes of ice wedge polygons—rim height and trough width—play key roles in determining the strength of positive feedbacks on permafrost thaw. To map these two attributes at landscape to pan-Arctic scales, we have developed two machine learning based tools. The first, operating on high-resolution laser altimetry data, defines ice wedge polygon boundaries and measures microtopographic relief. The second, designed to operate on high-resolution satellite imagery available from 2008-present, measures the areas of thermokarst pools and tracks their growth over time. Together, the results from our modeling and remote sensing analyses provide rich context for identifying which Arctic landscapes are most vulnerable to rapid permafrost thaw, thus potentially altering carbon emission pathways. Additionally, the maps produced through our machine learning workflows represent an extensive calibration-validation dataset for the next generation of Earth system models, as thermokarst processes are increasingly incorporated into estimates of pan-Arctic permafrost degradation.