Models applied to simulate the impact of climate change on vegetation dynamics generally face the trade-off between computational expenses (computation time and memory) and modelled detail. Models used for simulations of large areas (e.g. continental) often abstract processes entailing spatial linkages, e.g. species migration, and have too coarse resolutions to depict microsite heterogeneity. Regional to local models, on the other hand, are more detailed, but their computational expenses prevent applications on larger scales. For manageable and accurate simulations of vegetation dynamics on large scales, small-scale dynamics need to be integrated with large-scale applications in a balanced way. Several methods have been proposed and applied to expedite the integration of scales. However, each method has different advantages and drawbacks and the applicability of a method also strongly depends on the initial model and on the research question.
Here we present a conceptual framework for a further step integrating the scales in simulations with spatially explicit, time- and space-discrete models simulating vegetation dynamics under climate change. In such models, grid cells with similar environmental drivers and species compositions often entail repetitive calculations. Our method strives to reduce this redundancy and aims to disentangle repetitive calculations from processes specific to single cells. The proposed method is based on a dynamic two-layer classification (D2C) concept, in which the majority of processes is simulated in representative cells constituting the coarse layer, and only processes which might lead to changes specific to a single cell are simulated on the original grid, i.e. the fine layer. This new concept is a further step to enable the simulation of more detailed small-scale dynamics on a larger scale. We provide an example applying the D2C concept with the forest-landscape model TreeMig and shortly discuss its advantages and limitations.