Explicitly modelling microtopography in permafrost landscapes in a land-surface model (JULES vn5.4_microtopography)

Abstract. Microtopography can be a key driver of heterogeneity in the ground thermal and hydrological regime of permafrost landscapes. In turn, this heterogeneity can influence plant communities, methane fluxes and the initiation of abrupt thaw processes. Here we have implemented a two-tile representation of microtopography in JULES (the Joint UK Land Environment Simulator), where tiles are representative of repeating patterns of elevation difference. We evaluate the model against available spatially resolved observations at four sites, gauge the importance of explicitly representing microtopography for modelling methane emissions and quantify the relative importance of model processes and the model’s sensitivity its parameters. Tiles are coupled by lateral flows of water, heat and redistribution of snow. A surface water store is added to represent ponding. The model is parametrised using characteristic dimensions of landscape features at sites. Simulations are performed of two Siberian polygon sites, Samoylov and Kytalyk, and two Scandinavian palsa sites, Stordalen and Iskoras. The model represents the observed differences between greater snow depth in hollows vs raised areas well. The model also improves soil moisture for hollows vs the non-tiled configuration (‘standard JULES’) though the raised tile remains drier than observed. For the two palsa sites, it is found that drainage needs to be impeded from the lower tile, representing the non-permafrost mire, to achieve the observed soil saturation. This demonstrates the need for the landscape-scale drainage to be correctly modelled. Causes of moisture heterogeneity between tiles are decreased runoff from the low tile, differences in snowmelt, and high to low-tile water flow. Unsaturated flows between tiles are negligible, suggesting the adequacy of simpler water-table based models of lateral flow in wetland environments. The modelled differences in snow depths and soil moistures between tiles result in the lower tile soil temperatures being warmer for palsa sites. When comparing the soil temperatures for July at 20 cm depth, the difference in temperature between tiles, or ‘temperature splitting’, is smaller than observed (3.2 vs 5.5 °C). The mean temperature of the two tiles remains approximately unchanged (+0.4 °C) vs standard JULES, and lower than observations. Polygons display small (0.2 °C) to zero temperature splitting, in agreement with observations. Consequently, methane fluxes are near identical (+0 to 9 %) to those for standard JULES for polygons, though can be greater than standard JULES for palsa sites (+10 to 49 %). Through a sensitivity analysis we identify the parameters resulting in the greatest uncertainty in modelled temperature. We find that at the sites tested, varying the parameters can result in the modelled July temperature splitting being at most 0.9 or 3 °C larger than observed for palsa or polygon sites respectively. Varying the palsa elevation between 0.5 and 3 m has little effect on modelled soil temperatures, showing that having only two tiles can still be a valid representation of sites with a large variability of palsa elevations. Lateral conductive fluxes, while small, reduce the temperature splitting by ~1 °C, and correspond to the order of observed lateral degradation rates in peat plateau regions, indicating possible application in an area-based thaw model.