Newsroom |
|
|||
|
||||
Home Contact Newsroom Reports & Documents Links |
Akio Arakawa
UCLA - Department of Atmospheric and Oceanic Sciences, Los Angeles, CA
Time: 3:30 PM
Location: Goddard Space Flight Center, Greenbelt, MD
Bldg. 33, Room H114
The Cumulus Parameterization Problem: Past Confusions, Current Frustrations, and Future Excitements
Although the importance of cumulus-convective processes in the atmosphere has long been well recognized, the progress in representing them in weather and climate models has been very slow in the past, party due to confusion in understanding the parameterizability of cumulus convection. In my opinion, controversies due to such confusion have been essentially dissipated. The principal closure in all surviving cumulus parameterization schemes can be interpreted as adjustment, which is a negative feedback against large-scale destabilization.
There exists a clear trend, however, from diagnostic closures including instantaneous adjustment to prognostic closures including relaxed, delayed or triggered adjustment. This transition is analogous to that of the dynamics in NWP models from the quasi-geostrophic equations to the primitive equations, which took place in 1960s. In the present case of cumulus parameterization, however, we are not going back to known equations. In particular, there is no well-established theory on the transient behavior of cumulus convection comparable to the geostrophic adjustment theory. In addition to a number of uncertainties on cloud and associated processes, there are even conceptual problems in conventional model physics, such as artificial separation of processes and artificial separation of scales.
Cumulus convection is inherently anisotropic and penetrative, and therefore one-dimensional parcel method is still useful as a starting point. The trend in the conventional parameterization approach is, however, squeezing all complexities associated with cloud systems into single-column model physics. This is an extremely challenging task even when it is possible. At present, most modelers are frustrated because they cannot delay improvement of weather and climate predictions until all relevant scientific problems are solved. In particular, the link between the nature and highly parameterized conventional models are so remote that observational verifications of the models are not as constructive as they should be. In this respect, the recently proposed super-parameterization and its extension to a quasi-3D multi-scale modeling framework (MMF) may well lead to future excitements, providing a new framework for model physics that can link modeling and observations more closely and can better coordinate our currently diversified modeling efforts.
Print-friendly page (opens in a new window)
Click here for photos from the seminar
|
|