The Influence of Artificial and Physical Factors upon Predictability Estimates Using a Complex Limited-Area Model

Abstract

Recently, optimistic reports have appeared indicating that mesoscale circulations are more predictable than synoptic scale circulations. These have been based on studies using limited-area meso-α-scale forecast models. Warnings have also appeared suggesting that these results are party artifacts of the experimental and model designs, particularly strong diffusion and an ?error sweeping effect? of lateral boundaries. We demonstrate that an additionally important effect of the lateral boundaries is to restrict the scales at which errors can grow: if the domain is sufficiently large, forecast differences grow with time, but only at large scales. Our results show a strong sensitivity to synoptic situation and selection of an initial perturbation. Experiments with and without topography reveal that predictability is enhanced due to systematic topographic forcing. Detailed scale analysis of forecast differences and comparison with global model results indicate that the predictability using limited-area mesoscale models is not fundamentally different from that using global synoptic scale models.