Comments on "preliminary tests of multiscale modeling with a two-dimensional framework: Sensitivity to coupling methods"

In summary, in the test used in JA05, the original SP approach (GS99; G01; Grabowski 2004) performs substantially better than concluded by JA05. The key is the coupling between mesoscale and convective dynamics, which (for unclear reasons) was not captured in JA05's test of the original SP approach. The 3D results presented here (Figs. 4 and 5) suggest that the original approach is attractive for mesoscale models with a horizontal grid length in the range of 20-50 km (e.g., in regional climate models). In such a case, small-scale convective dynamics (albeit in 2D) are captured by the SP model, and 3D mesoscale dynamics are captured by the mesoscale model. The approach developed in JA05, on the other hand, seems better suited for large-scale models with a horizontal grid length of a few hundred kilometers, as in contemporary global climate models. In such a case, the mesoscale and convective dynamics have to be treated by SP models, and coherent propagation of cloud and mesoscale structures across the large-scale model grid is only possible in the approach proposed by JA05. In a nutshell, the test used in JA05 seems a poor choice for illustrating their approach. It would be more appropriate to use the idealized 2D Walker circulation test (Grabowski et al. 2000) as applied in G01.