Relevance of sub-grid-scale land-use effects for mesoscale models

The dependence of the performance of mesoscale-alpha and mesoscale-beta models on model resolution and parameterisation of sub-grid-scale land-use effects is investigated for a meteorological situation that is strongly influenced by local effects. Two mesoscale models of different horizontal resolution and model physics are applied to the same meteorological situation, and the results are compared with observations. Sensitivity studies are performed to estimate the effect of resolution (4 km or 5 km versus 18 km), parameterisation of sub-grid-scale land-use effects and initial state on the results of both models. Results show that the accuracy improves with resolution when simple parameterisations of sub-grid-scale land-use effects are applied. The inclusion of sub-grid-scale land-use effects with application of a flux aggregation method provides the best solution, which is nearly independent of the resolutions used. The use of parameter averaging to account for sub-grid-scale land-use effects was worse than just using one (main) land-use per grid cell. Initial conditions were also found to have a significant impact on the results.