Close Window

University of Oklahoma: "Evaluation of numerical stormscale forecasts at an Air Force Base weather station"

Final Report


The first half of the project was quite successful, with an active partnership between the Tinker AFB (TAFB) weather officers and OU faculty and students. As enumerated in the semi-annual and annual reports, many new forecast products were developed, eventually responding to about 95% of the products requested by Tinker in the original proposal. A web site was set up ( that was used by TAFB forecasters. Notable among the new products were forecast reflectivity fields, 3D winter precipitation type displays, icing potential, meteorgrams, hail potential and other more standard fields. New forecast verification methods became a major part of our efforts in order to validate these new forecast products. Two students received their M.S. degrees on this work, Capt. Chris Stock, an AFIT student, and Eric Kemp. Stock's work focused on comparing several icing potential forecast methods used by aviation weather forecasters. Eric Kemp continued further with the comparison of newer, more sophisticated icing algorithms and studied various clear air turbulence forecasting procedures as well.

By 1999 there was a complete turnover among the TAFB weather officers and the interactions between OU and Tinker personnel gradually ended. We did support two additional MS students, Capt. Robert Edwards and Jennifer Ritterling. These students have both nearly completed their MS theses, although both have left OU, Capt. Edwards to Offut AFB when his two years were completed, and Ms. Ritterling accepted full-time employment at WeatherBank in Edmond, OK. They are slowly completing their work in their spare time and should be done by December 2001.

Capt. Edwards is performing a study on the relative importance of mesoscale data sets on the performance of the ARPS forecast model. Unique features of this research include the testing of the value of NEXRAD radar products such as radial velocities and reflectivity, the high resolution of the forecasts, and new products such as CAPE that will be evaluated in the overall assessment. These sensitivity tests will help determine what observing systems are most important for storm-scale modeling.

Ms. Ritterling is examining the important issue of the relative usefulness between a set of ensemble forecasts vs. one high-resolution forecast. Both forecast approaches have merit but given limited computing power for operational forecasts, choices will have to be made. An important component of this research is to decide on the proper way to compare forecasts run at different resolutions (the ensemble runs are done at lower resolution in order to use only the same computer time as the single, high-resolution run).

In summary, once the TAFB collaboration ended, we chose to pursue two important contemporary research problems that would not only benefit AFWA numerical forecasting progress, but would be of interest to all NWP researchers and organizations.


The primary achievements of the collaborative efforts are mentioned in the first section, with more details provided in the 6-month and annual reports. Credit should be given to Capt. Sock, who developed the above web site and maintained close ties with TAFB personnel. The useful of the ARPS products, however, was marginal to TAFB forecasters because our timelines, domains and forecast lengths did not match up with their operational needs. This was primarily an issue of computing power limits at CAPS, resulting in forecasts too short and with domains too small to be useful. Efforts were underway to solve these problems when the personnel turnovers began.


Stock, Christopher M., 1998: Intercomparison of Icing Aviation Impact Variable Forecasts Produced during Real-Time Mesoscale Numerical Weather Prediction. MS Thesis, University of Oklahoma, Norman, OK. 187 pp.

Kemp, Eric M.,1999: Comparative Assessments of Mesoscale Aircraft Icing and Turbulence Forecasts from the Advanced Regional Prediction System. MS Thesis, University of Oklahoma, Norman, OK. 138 pp.

Kemp, E. M., C. M. Stock, F. H. Carr, and M. B. Schott: Storm-scale numerical weather prediction for commercial and military aviation: Part II: Development and assessment of aviation impact variables. Preprints, 8th Conf. On Aviation, Range and Aerospace Meteorology. Amer. Meteor. Soc., Dallas, TX, 164-167.

Two more MS theses are in progress. I will also note that Eric Kemp has moved on to become an employee of CAPS where he is the focal point for all of CAPS verification efforts. He has written several articles on this topic since receiving his MS, and although these were not supported by COMET, the training that led to them was. The P.I. (FC) also co-authored a review article on mesoscale verification issues with Chris Davis that was published in B.A.M.S.


All the forecast products developed during this product have become part of the standard product suite of the ARPS model. Similarly, the verification procedures and software developed and used during this research are now being used by CAPS. As mentioned above, Eric Kemp has become a valued employee of CAPS. Similarly, Capt. Chris Stock developed into an outstanding research support scientist and is at OFFUT AFB. Capt. Bob Edwards, who is still working on his thesis, is also benefiting the AFWA at Offut AFB. Jennifer Ritterling's talents led to her being recruited for a full-time meteorologist position at WeatherBank, but she is also completing her thesis later this year. The production of four good students for the benefit of AFWA and the meteorological profession is a good result. The research results so far have greatly benefited CAPS work with the FAA and American Airlines. The theses of Edwards and Ritterling have the potential to be converted into refereed papers.

The primary problems have been enumerated above. There is really no fault to assign, since personnel turnover at Air Force bases is a fact of life. A final contributing factor that discouraged ARPS products, even if the domain and forecast timing problems had been solved, was that AFWA, during this project, made a major decision to adapt the MM5 model for their operational forecasting guidance. Thus there was little motivation for forecasters to learn how use and interpret ARPS model guidance.