Surface aggregation and subgrid‐scale climate

Land‐surface representations used by the climate modelling community characterize locally heterogeneous surfaces as larger, homogeneous units. This loss of detail in land‐surface properties has prompted research into methods by which subgrid‐scale properties may be included within global climate models (GCMs). Coupling a regional model within a GCM represents one possible approach to resolving subgrid‐scale heterogeneity, but this method is computationally demanding. For use at global scales, two aggregation schemes have been proposed in the literature. The first assigns to each GCM grid‐cell the area‐weighted values of structural and physiological parameters typical of each subgrid surface type; the second computes separate energy balances for each surface type within the cell, then averages their responses based on areal coverage. Neither of these aggregation schemes, however, account for subgrid surface‐atmosphere interaction, surface geography, or advective effects. This study examines climate sensitivity to methods of surface aggregation by comparing climates simulated using a three‐dimensional, mesoscale atmosphere‐land‐surface model with those produced by parameter‐based (‘averaged geography’) and energy‐balance‐based (‘mosaic geography’) aggregation schemes. Simulations are carried out for a 240 kmx240 km area using three land‐surface configurations. Each domain contains a small area of bare soil, along with varying amounts of grass and coniferous trees. Model comparisons indicate that the averaged geography scheme tends to overestimate the latent and sensible heat flux from the model domain, and to yield surface temperatures that are cooler (by as much as 1.8°C) than control‐simulated values. Mosaic geography results are more similar to control climates, but tend to underestimate surface sensible heat flux and to overestimate the latent heat flux. Surface temperatures produced by mosaic runs are warmer than control values by up to 0.8°C. Overall, the mosaic aggregation scheme appears superior to the averaged geography method in its ability to capture the domain‐averaged climatic effects of subgrid heterogeneity. However, relatively large differences between the climates simulated by mosaic and control experiments for the individual grass, coniferous tree, and bare soil areas suggest that studies of subgrid climatic forcing and response are best performed with regional‐scale or coupled regional‐global climate models.