5
Concluding Thoughts
Many workshop participants agreed that progress in model development is being impeded by the slow pace of improvement of the parameterizations of physical processes. Workshop participants noted a disturbing tendency away from a rigorous course of development and testing of new parameterizations and toward tuning existing parameterizations or using new but inadequately tested ones. The participants identified several likely contributors to this situation:
-
Widely available, easily run models and the current funding and academic environments may be turning both graduate students and their faculty advisors toward fast-turnaround research in numerical simulation and away from the traditional but much slower path of theory and observation.
-
Bright young scholars best suited to tackling scientific problems may incorrectly perceive that the atmospheric and oceanic sciences are an applied field whose goal is merely to improve weather and climate forecasts.
-
Progress in parameterization, which often requires interactions across traditional disciplinary boundaries, could now be inhibited by the compartmentalization of educational, research, and funding institutions.
-
The rigidity of long-existing models and the lack of efforts to remove inferior or flawed physical representations hinder progress by preventing opportunities for new, fresh thinking.
These trends could cloud the future of the fields of atmospheric science, climate, and oceanography. Models are widely used, perhaps much more so than could have been anticipated at the dawning of the computer age some 40 years ago. The growing user community includes many nonspecialists who are not prepared to assess or question the reliability of model predictions. As computers become larger and allow even finer model resolution, some local model forecasts will become increasingly stochastic. It was not clear to the workshop participants that the atmosphere, climate, and oceanography communities are prepared or even constituted to make the required adjustments, refinements, and improvements in model parameterizations that this will require. It is particularly troubling that if our unresponsiveness to the parameterization challenge is indeed culturally driven, its implications carry an aspect of inevitability.
Atmospheric scientists and oceanographers understand both the practical and scientific importance of the parameterization problem. But we also need to bestow it academic dignity by acknowledging it as a difficult and important problem in mathematical physics. Indeed, it involves, among other things, turbulence, which is commonly referred to as the outstanding unsolved problem in classical physics. The parameterization problemwhich could be defined as the identification and understanding of the physics of unresolved processes and the compact, optimal representation of this physics in numerical modelsdeserves serious attention from the best and brightest.
This report summarizes the discussions of the workshop participants on parameterization approaches and frameworks and on specific targets for parameterization improvement. Among the key issues identified by the participants were the following:
-
An important field of parameterization science has emerged over the past 40 years as a result of the computer revolution. Workshop participants believe that our educational, research, and funding institutions need to recognize, accommodate, and foster this new field.
-
The modeling community needs to encourage, support, and pursue alternative computational approaches (e.g., multi-scale computing) and better computational methods.
-
More extensive and rigorous comparisons of models with observations and field experiments designed to support such comparisons are needed.
-
The cultural issues thought by the workshop participants to be limiting progress in model development might not be self-correcting; they could require the institutional adjustments that are occasionally but necessarily made as society and the atmospheric and oceanic sciences respond to changing conditions.