Close Window

Univ. of North Dakota: "Evaluation of the local analysis and prediction system for the northern high plains"

Final report


The work of the outreach program was to implement the Local Analysis and Prediction System (LAPS) of the NOAA Forecast Systems Laboratory for use across the Northern Plains. Investigation of the LAPS under several difficult to forecast situations was the primary learning activity for the students involved in the project. Modification of the program was made when the prediction component of LAPS was not available to the project. LAPS was replaced in the project by the NCAR/Penn State mesoscale model (MM5) in order to provide the numerical weather prediction components of the project.

The University of North Dakota (UND) provided the necessary implementation platforms to execute the MM5 and the required data sets for model initialization. The National Weather Service (NWS), through primary assistance from the Great Falls, Montana office, provided case study selection and guidance on key forecast issues related to each case.

Overview of Activities

As this project was an undergraduate student project, the first order of business was the preparation of students to participate in the work. A selection process was established to identify able candidates and to determine their level of commitment to completing the project. This involved interested students performing an extensive literature search of meso-scale weather analysis and prediction to acquaint themselves with the general topic. Formal and informal discussion groups were held outside of class to stimulate student involvement in the project. This activity was performed to broaden the number of students involved in the project and to provide a method of selecting a final student(s) to become the compensated student(s) in the project. Mr. Ryan Zerr (senior - atmospheric sciences) was selected through this evaluation process in April 1996. Mr. Zerr was selected based upon his exceptional academic record and strong desire to pursue scientific research in atmospheric science. However during the summer of 1996, Mr. Zerr elected to graduate during the summer semester and was no longer participating with the COMET project. It was decided that in place of Mr. Zerr there would be three students involved with the project: Mr. Bryan McQuade, Mr. Morgan James, and Ms. Kristi Schueler. All three students were atmospheric sciences majors. Mr. James was also entering graduate studies in mathematics.

During the summer months student work on the project was suspended; however, faculty activity was continued intermittently during the summer. This latter work involved MM5 simulations and construction of analysis data sets. The information compiled during the summer by faculty was used during the regular academic year by the student investigators.

University Results

Scientific Investigations

The first six months of the project was directed towards identification of case studies to be investigated, preparation of meso-scale model support to the case studies, and the preparation of students to assist in the project. The identification of case studies was completed by the NWS personnel located in Great Falls, Billings, and Glasgow, Montana. Four cases were identified which provided a range of synoptic/mesosynoptic events. The cases and dates for the study were

Simulation studies for the four cases used the NCAR/Penn State meso-scale prediction model - MM5. A 60-kilometer coarse domain was established centered on east-central Montana. This domain extended westward into eastern Washington, eastward into eastern North Dakota, northward into central Saskatchewan, and southward into southern Wyoming. Centered within this coarse domain was a 20-kilometer domain. This finer domain was the primary focus domain for all cases except the tornado case of June 25, 1996. For the latter, a sub-I 0 kilometer domain was used to attempt to better resolve terrain induced storm environments.

Thirty-two simulation studies were conducted for the four cases. For the first twenty cases the model was initialized using MRF data (both 2.50 x 2.50 and 2.50 x 5.00 grids are use based upon availability). The tornado case study and freezing rain studies were later evaluated using the RUC (remapped to NCEP grid 212) as initialization. Model physics was altered between simulations; however, the basic physics used were

Comparison of model analyses/forecasts were made with NCEP ETA, RUC and NGM output. The high wind and fog case studies yielded no significant improvement (the fog case actually was degraded from the ETA) to the NCEP models. In the case of the freezing rain (November 1995) and tornado (June 1996) cases, improvement was found to occur with the use of the meso-scale model. The finer resolution of the MM5 provided better timing and positioning of the tornadic activity in the vicinity of Great Falls. However, best results were found for the freezing rain case.

Considering the MM5 forecast was executed between 24 and 48 hours before the onset of freezing rain, the model did a favorable job of showing the potential for freezing rain close to the actual time of occurrence. Although the model failed to predict the occurrence of light precipitation that fell as light snow early in the simulation, the model did provide appropriate information on warm air advection moving across the frontal boundary that resulted in the freezing rain conditions. The main problem the MM5 model had was keeping the temperatures near 850-mb excessively cool and pushing the surface warm front through the area too quickly. Otherwise, the model proved that it has good potential for forecasting freezing rain near the time of occurrence and at a smaller scale.

Benefits to the University

This project has provided a vehicle to promote greater activity between the university and the NWS offices in Montana. While not utilized to the greatest benefit (see below), the opportunity for students to discuss synoptic meteorology and operational importance of the case studies selected provided students first-hand contact with non-university professionals in meteorology. This exposure outside the university was important to the three final students involved in the project. All are now presently employed in the operational sector of meteorology and two are professional weather forecasters.

Two of the cases studied in this project have been published on the UND Regional Weather Information Centers world-wide web site: "Case Study: Evaluation of the Effectiveness of Operational Use of the PSU/MM5 Mesoscale Model for an Isolated Severe Thunderstorm" and "MM5 Analysis of a Freezing Rain Episode in Glasgow Montana." The first case study was also published (and presented) in the Proceedings of the Second Workshop on Northern Plains Convective Storms (20-21 March 1997). This workshop was co-hosted by the Eastern Dakota WFO and UND.

The cases studied under this project will provide valuable case studies for reference in future university courses and research. Efforts are underway to incorporate the freezing rain study into the university's senior-level synoptic meteorology course beginning the fall semester 1998. This case study provides a representative look at the development of a major winter weather hazard that has impact on many sectors of applied meteorology for the Northern Plains. The development of this case by Mr. McQuade was done with the insight that all materials could be presented in an instructional environment, not just the research environment.

The presence of this collaborative project has also interested a greater number of UND students into exploring summer internships and cooperative education experiences not only with the Great Falls WFO, but with other WFOs in the region. While a cooperative education program has long existed between the NWS and UND, this program emphasized the role investigative analyses play within operational meteorology. Some students realized the role of a NWS meteorologist is more that forecasting and this increased their interest in the NWS as a career possibility.

This project has resulted in a higher level of extracurricular activity for students interested in mesoscale meteorology and numerical weather prediction. These activities were not funded through this project even though the result of the activity was critical to the selection of a student to be compensated by the project. This is important, as there are no formal undergraduate courses at the University of North Dakota in these areas. In addition, discussions with the Eastern North Dakota (FGF) NWS office have occurred as a result of this project and could result in future research efforts. Planning is presently underway to provide access to operational UND MM5 products by the beginning of 1999.

Difficulties Encountered

Four significant problems were encountered during this project. The first was administrative in nature and related to difficulty in negotiating language within the COMET award, which was acceptable to the University of North Dakota. Per university policy, the project was not permitted to begin until the contract was signed. This administratively forced a two-month delay in scheduling a January start of the project.

The second problem dealt with the lack of availability of the original mesoscale prediction model within the NOAA/FSL LAPS. This model, CSU-RAMS, was anticipated as the prediction component of LAPS. However, the university was notified in late October 1995 by FSL officials that this model would not be available and that another model would have to be located and interfaced with LAPS. An application was made to NCAR for permission to use MM5. This approval was received in mid-November and the process of transitioning to MM5 began immediately. By April 1996, MM5 was installed and running on the UND Cray J90. Unfortunately, the ability to interface this model to LAPS has not been as successful at UND. The process of interfacing LAPS analysis software to MM5 data initialization modules is ongoing at the university at this time. Fortunately, an extensive library of in-house analysis routines has served as an excellent replacement for LAPS to this point.

The third problem was logistical and related to coordination of efforts between UND and the NWS. While inexpensive means of communication existed at both agencies, neither group was effective in using this tool to maintain coordination. Fault for not exerting more emphasis on this coordination must reside with UND; however, some student contact with the WFOs was met with indifference that blunted their motivation.

The final problem resulted from the chaos generated during late winter blizzards and spring flooding from early March through May 1997. While the contract period was complete, except for final report production, the evacuation of the university and local community placed all normal activities on hold. Students required to submit a final project report were displaced from the area and unable to communicate their summary findings. In some situations the report findings were lost to flood conditions.

National Weather Service Results

Scientific Investigations

A trip was made to the University of North Dakota to discuss running the MM5 on a number of synoptic events. Data was provided to the university to aid in the investigation as the model was run on each case. The university was left alone to parameterize the different storms, given the input from the NWS.

Benefits to the NWS

A number of benefits were realized by the NWS as a result of the University of North Dakota's running of the MM5 in this investigation.

It has added to the interest level of research in the office. Forecasters have used other software packages to write technical papers, and other software packages to display model data, but have not had the opportunity to post analyze a weather event by manipulating a meso-scale model.

The COMET project has kept the channels open between the university and the NWS. An additional summer student has been hired since the project was started. This project has shown that a resource of high quality students exists at our university, and these students would make good hires at the NWS.

The greatest benefit is that it has heightened the interest in using output from a mesoscale model. This office has procured computing power so that the MM5 might be run operationally in the office. The value of a meso-scale model has been realized by using the results of the meso-ETA and the MM5. Forecasters have seen how a high-resolution model might give the additional information necessary in a difficult forecast, especially when complex terrain is considered.

Difficulties Encountered

The main difficulty encountered was the distance between the university and the NWS office. This created other problems, but was highlighted by lack of communication. This was not a large problem, however, as the university was running and adjusting the model. Upon completion of the investigation, the results were published on the UND home page.


University of North Dakota

Considering the problems encountered during the project the experience by both students and faculty would have to be considered positive. Certainly, the exposure to advanced methodologies by undergraduate students was a positive experience. Their academic experience and the availability of quality data limited the detail to which they could investigate the various case studies. Regarding the former, the faculty experience gained by this project points to the need for future COMET projects to involve graduate students rather than undergraduates. In many situations the level of knowledge required to grasp the numerical weather prediction concepts involved was beyond that expected of an undergraduate. The fact that the students did so well reflects on the extraordinary intellect they posses and should not be expected universally; however, the amount of faculty time required to equip the students with additional knowledge and skills well exceeded the time budgeted for the project. The involvement of graduate students would have been more commensurate with the budgeted time and research expectations.

The availability of quality data reflects on changes needed within the University of North Dakota to manage its weather data resources better. While data from NWS sources existed at UND, attempts to use the data collected at UND from NWS sources proved less than reliable. Acquisition of these data from external sources was not possible due to budget constraints.

And while the analysis of the case studies proved to be valuable to university students and faculty, the lack of routine communications between NWS and UND has limited the benefit to the NWS. Had the physical distance between the two agencies been less intimidating there would have no doubt been better coordination. This capability for the principles to meet face-to-face is particularly important when the student researchers are undergraduates. However, the only possible means of providing face-to-face meetings would have been through airline travel. With the number of participants and associated costs involved this was not even considered. As UND has recently developed a graduate atmospheric sciences distance education program, future visits may be better facilitated through the use of video conferencing. This latter possibility, i.e. the use of Internet conferencing was submitted as a follow-on proposal to COMET, but rejected.

Two NWS personnel must be recognized for providing ongoing support and information to the UND effort. Mr. David Bernhardt, GTF-NWS, and Mr. Greg Gust, GGW-NWS, provided important assistance in the form of analysis insight, supporting data, and synoptic overview of their respective case study topics.

National Weather Service

Considering the lack of coordination and communication during this project, it was generally a positive experience, after all. It underscored the importance of using mesoscale data sets and models to prepare a better forecast. The lack of good surface and model data has long plagued forecasters in Montana. The results of this investigation have reinforced forecasters perceived need for better data sources and model output. It has shown that the high resolution possible through models such as the MM5 can assist the forecaster in giving more definition to timing and location of convective and wintertime events.