To tile or not to tile?

Soils and landscapes vary within centimeters to decameters, which is not captured by state-of-the-art land-surface models that operate on kilometer scale. This leads to potential mismatches when simulating the exchange of energy, water and gasses between land and atmosphere, which are summarized under the term “aggregation error”, and is a major source of uncertainty. To overcome this issue and account for subgrid-scale heterogeneity so-called tiling approaches are used, which separate grid cells internally into different tiles that interact with each other. Although this is a valid approach, it remains unclear, if and to what extend tiling reduces the aggregation error and consequently, if tiling is sufficient to account for subgrid-scale heterogeneity.

Permafrost soils are especially heterogeneous and the aggregation error when simulating permafrost landscapes is especially problematic, because it can make the differences between frozen and unfrozen, as well as waterlogged and unsaturated areas. This affects the presence of permafrost itself, the build of soil ice and resulting frost heave, and determines pond locations as well as the duration and thickness of the seasonal snow cover, which all together influence vegetation and thus ecosystem dynamics.

To address the sufficiency of tiling at permafrost landscapes, we apply the two-dimensional pedon-scale soil model DynSoM at a non-sorted circle site. We run DynSoM with four different horizontal resolutions: (i) with an explicit resolution of 10cm, (ii) with three tiles, representing center, rim, and interface area, (iii) with two tiles, representing center and rim, and (iv) without tiling, representing a typical state-of-the-art land surface model. By comparing mean simulations, we assess the benefits, but also the shortcomings and limitations of the different tiling set-ups, and discuss implications for tiling within kilometer-scale land-surface models.