More methane : making the case for explicitly modelling microtopography in permafrost landscapes

Abstract:

The northern permafrost zone is a vast store of carbon and projected to be a significant source of emissions under warming. However, within this expanse processes on the scale of meters arising from repeating patterns of elevation difference and ground ice compound to create differences in soil moisture and temperature on similar scales. This in turn can result in increased anaerobic decomposition and methane release from warmer wetter areas, and feedbacks causing localised rapid degradation of permafrost. Global models used for future climate projections do not resolve this heterogeneity and simply model a 'grid-cell average', leading to greater uncertainty in the permafrost carbon feedback, and limiting the ability to validate models against in-situ observations. In addition, most long-term observations of permafrost are of limited spatial resolution and do not encompass local variability, hampering efforts to accurately model these localised processes. However, at a handful of field sites the local variability has been well quantified and 'tiling' approaches to modelling microtopography have been tested at these sites. Even so, all such methods necessarily rely on simplifications, and the impact of a fully coupled physical and biogeochemical scheme has not yet been quantified.

Here, a simple two-tile approach is presented which uses the redistribution of snow, lateral flows of moisture and surface ponding to model the effect of landscape heterogeneity within the Joint UK Land Environment Simulator (JULES). Site simulations are set up in locations where microtopographically resolved data is available, covering different permafrost landscape types including polygonal tundra (Samoylov, Siberia), and palsa mire (Iskoras, Norway). The model approach is validated against observations and the thermal and hydrological effects of lateral flows quantified. The total methane emissions are shown to be larger than simulated by a standard gridbox, suggesting that standard Earth System Models may be underestimating methane emissions from the Arctic and confirming the need to explicitly model microtopography. A possible development pathway towards full integration within the land surface model and a global simulation is proposed.