Predictability of Flows over Complex Terrain

Abstract

We review the relevance of predictability theory to flows in local domains that are modified by boundary conditions produced by topography, surface physical forcing, or imposition of numerical lateral boundary constraints. There is an apparent inconsistency between the results of classical predictability theory and limited domain experiments. We attempt to reconcile the discrepancy by analysis of error growth and chaos. The analysis suggests that boundary conditions provide the dominant flow contribution for many limited domain predictions, and this motivates review of error spread from uncertain boundary specification. Case summaries include free flows of the troposphere, moist convection, topographically bounded local flows, and flows above other complex surfaces. A tentative conclusion is that boundary uncertainty does not provoke chaotic responses so easily as do small imperfections of initial data, even for dynamical systems that are chaotic in global or periodic domains. One exception to this appears to be the case that the boundary uncertainty possesses a spatial scale that is not much shorter than the perimeter bounding the simulation domain.