Late in my teaching career, in 1985-86, I spent the academic year on sabbatical at the European Centre for Medium-Range Weather Forecasts in Reading, England. I worked with ECMWF scientists, particularly Tony Hollingsworth, in evaluating the performance of the operational prediction model (then the T-106 model) in identifying and predicting African easterly waves and the occasional hurricanes they spawn in the Atlantic. Despite the deficiencies of the observational network over Africa, the model did a surprisingly good job.
This project consumed most of my time, but I took time each day for a solitary, unofficial activity: I scrutinized the daily forecasts and other material plastered on the high walls of the operations center, making a personal evaluation of how well the model predicted extratropical weather systems. My conclusion was that on most occasions the model did a highly realistic job of simulating the development and evolution of midlatitude synoptic systems, just as it did for the tropical systems we were investigating. Out of this judgment grew the conviction that output from the model could provide a rich source of data for diagnostic studies of synoptic systems.
Before my sabbatical, my research focused on the problem of explosive marine cyclogenesis. It now seemed to me that high-resolution model output afforded the opportunity to identify and quantify the mechanisms involved in rapid cyclogenesis in a way and with a precision not hitherto possible. About the same time, Lou Uccellini and Dan Keyser expressed a similar view in an article in the Bulletin of the American Meteorological Society. However, I had no immediate way of introducing model data into my NSF-sponsored research.
On an earlier visit to the University of Washington, Rick Anthes (then an NCAR division director) had suggested that I extend a research project--a diagnosis of a case of rapid cyclogenesis in the eastern Pacific, using conventional synoptic tools--by teaming up with his junior colleague Bill Kuo to apply the Pennnsylvania State University/NCAR Mesoscale Model. My sabbatical kept me from acting on this invitation immediately, but now that I was back home, imbued with the faith that model data could open new horizons in synoptic research, it was natural to proceed with Rick's suggestion. Thus a collaboration began that has lasted to the present.
As an example of the benefits that can flow from such collaboration--some of them straightforward and anticipated, others unforeseen--let me highlight what I feel to have been the main accomplishments of this partnership.
Joint research. Ten coauthored papers have been published on a variety of topics related to marine cyclogenesis, including the role of surface energy fluxes, condensation heating, symmetric instability, and potential vorticity and the documentation of the three-dimensional airflow in representative cases. NCAR support scientists have provided valuable assistance; Simon Low-Nam, a coauthor on some of the papers, especially deserves credit. Two visiting scientists from China, Yubao Liu and Kun Gao, also participated in the research and coauthored papers.
Ph.D. training. UW Ph.D. candidates Warren Blier, Mark Stoelinga, and Jordan Powers became NCAR graduate research assistants in the Advanced Study Program, under the helpful hand of John Firor. Not only did residency at NCAR provide them with ready access to vital facilities, it afforded the opportunity for interaction with NCAR scientists having a wide range of expertise, particularly their supervisors, Rich Rotunno, Chris Davis, and Bill Kuo.
Postdoctoral appointments. Postdocs Georg Grell and Jim Bresch took part in the collaborative research, with financial support from the Office of Naval Research and NSF, respectively. Grell developed a convective parameterization scheme akin to, but simpler than, the well-known Arakawa- Schubert scheme. The new scheme has received considerable use by the modeling community. After his postdoctoral stint, Grell remained for a time at NCAR, where, working with Jimy Dudhia, he made important contributions to the development and documentation of the Mesoscale Model, version 5 (MM5). Bresch conducted fruitful research on polar lows in collaboration with UW scientists. He established ties with NCAR and NOAA scientists, academic investigators at the Universities of Iowa and Wyoming, and operational forecasters at the National Weather Service (NWS) office in Pendleton, Oregon.
Meteorological practice. The creation of a hospitable environment at the University of Washington for the MM4 and MM5 excited the interest of my colleague Cliff Mass and his students. Jim Steenburgh and Brian Colle, two of Cliff's students who are now professors at the University of Utah and State University of New York at Stony Brook, respectively, adapted the MM5 for local use as part of their Ph.D. research, aided by NCAR's Wei Wang. The local adaptation has continued under the efforts of UW staff members David Ovens, Mark Albright, and others. Now, the model is run twice daily in a nested grid configuration, providing high-resolution forecasts for the Pacific Northwest and Washington State.
At UW, the MM5 has become the centerpiece of an environmental modeling system. MM5 operational forecasts are now driving a distributed hydrological model that predicts streamflow for eight watersheds in western Washington. State and local agencies are also using MM 5 output for air quality modeling. The resources required for this exceed, of course, those available to a single university. Faced with this dilemma, Mass organized and now directs a consortium that supports the operation financially, and the consotium members derive forecasts and other model ouputs that they need. The other consortium members are the NWS, U.S. Navy, Puget Sound Clean Air Agency, U.S. Forest Service, U.S. Environmental Protection Agency, Oregon State Department of Forestry, and Washington State Departments of Ecology, Natural Resources, and Transportation.
Though my subject is the relationship between the University of Washington and NCAR, I cannot stress too strongly that the development and use of the MM4 and MM5 have encompassed a much wider community. The model was originally developed at Penn State by Rick Anthes and Tom Warner. Much of the further development has taken place under the outstanding leadership of Bill Kuo, but Penn State continues to be a major contributor through the efforts of Nelson Seaman and Dave Stauffer, among others.
The spread of the MM5 in recent years to an ever-widening circle (currently about 600 users from 200 institutions worldwide) has been a remarkable phenomenon. With only a small, dedicated staff to serve external users and an annual workshop and tutorial to provide instruction and promote interest and cohesion, NCAR has served as the catalyst for interactions that have had a huge impact on mesoscale modeling. In doing so, it has provided a striking example of the mutual benefits that can derive from cooperative efforts.