Consequences of Permafrost Thaw on Arctic-Boreal Soil Moisture Dynamics: Lessons from Long-Term In Situ Monitoring, Alaska, USA

Abstract:

In arctic and boreal regions, the presence of permafrost acts as an important control on soil hydrology by preventing vertical drainage of soil water from the active layer. This effect produces hydrological conditions that, in turn, are strong controls on ecosystem carbon exchange. As climate warming deepens the permafrost table, hydrological conditions are expected to shift in response; however, these dynamics are difficult to predict at the regional level due to the dependence of soil moisture on complex interactions among local environmental drivers. Through a synthesis of data from long-term in situ monitoring sites across Alaska, we are seeking to: 1) characterize long-term trends in arctic-boreal soil moisture; 2) evaluate how these trends vary across space and time; and 3) understand how permafrost thaw is changing soil moisture, both in magnitude and in sensitivity to environmental drivers. To address these questions, we compiled and synthesized measurements of soil moisture from ~175 sites across Alaska, along with related soil and meteorological measurements and ancillary information from metadata and remotely sensed datasets. Preliminary trend analyses suggest that while wetting trends, drying trends, and stable conditions occur across the region, drying trends are more prominent than wetting trends, and the magnitude of both drying trends and wetting trends is increasing over time. Drying trends are particularly dominant in the discontinuous and isolated permafrost regions of Alaska, while in the continuous permafrost region, shallow soil horizons appear to be drying as deeper horizons become wetter. In general, drying trends appear to be associated with decreases in precipitation and with well-drained landscapes (i.e., higher elevation and slope), while wetting trends are associated with warming air temperatures, higher probability of shallow permafrost (i.e., permafrost within 1 m of ground surface), and poorly drained terrain. Machine learning tools will be used to synthesize the effect of environmental drivers and their interactions on soil moisture trends.