Pursuing model errors from column processes to large-scale outcomes

Brian Mapes

University of Miami


This talk will try to synthesize work on 2 very different approaches to identifying model errors and strategizing parameterization development. The first involves study of local processes (here tropical convection) in Eulerian columns of the atmosphere, observed and simulated. The other looks at large-scale systematic forecast errors, in this case in an older (Reanalysis-like) NCEP global model's forecasts of the North American Monsoon (NAM), seeking clues about process problems.

For a pervasive process like moist convection, small-scale and large-scale model metrics must be coordinated, since it affects everything from the tropical mean state and zonal jets, to planetary-scale waves, down to local impacts like the desert rainstorms in the NAM's 'dirty ridge' flow regime. Can our little NCEP-University Climate Process Team contribute to substantive progress on such a hard problem?