SECTION 1: PROJECT OBJECTIVES AND ACCOMPLISHMENTS
1.1 Summary of Overall Project Objectives and Responsibilities of Participants
To improve the operational prediction and subsequent forecasts and warnings of mesoscale lake-effect snow events downwind of the Great Lakes and to expand the interaction and exchange of knowledge between personnel at the National Weather Service Forecast Offices in Buffalo and Binghamton and faculty and students at SUNY Oswego, SUNY Brockport, and Cornell University. University researchers were to develop a method to run the Penn State/NCAR Mesoscale Model (MM5) on a grid suitable for the simulation of lake-effect snowstorms, and NWS personnel were to evaluate the MM5 forecast products as part of the operational framework.
1.2 Description of Research/Development Activities and Accomplishments
A. During Year 1 of the project, an automatic method was developed to initialize MM5 using the most recent ETA GEMPAK file, run the model out to 24 hours, and generate hourly output in GEMPAK format for display (using GARP) at the NWS offices and universities.
B. During Year 2, the modeling system was ported to the dual processor Linux workstations at the NWS offices and run out to 36 hours in realtime. NWS forecasters are still using MM5 output along with the regular NCEP products. Hourly loops of MM5 output were put on the NWS Buffalo Web site.
C. Radar data and MM5 output have been archived for about a dozen lake-effect snowstorms. Analysis of the results has been carried out by faculty and students at the universities and at Oswego High School under another project.
D. In Spring 2001, MM5 Version 3 was set up at NWS Binghamton and at SUNY Oswego where it is now being run in realtime on a 54-km large grid and an 18-km nested grid. The new nested grid covers a domain about four times larger than the grid now in use at the NWS offices. MM5V3 is initialized using ETA212 GRIB files that include realistic water temperature over the Great Lakes.
E. For the Montague Storm of 11-12 January 1997, MM5V2 produced a good simulation to the east of Lake Ontario, but it moved the Lake Erie band southward prematurely. The simulation was rerun using a 36 km large grid and 12 km nested grid. The higher resolution run produced a more focused and intense band down-wind of Lake Ontario, but no improvement was noted with the lake Erie band. A run with all lakes removed except Lake Ontario showed that only about two-thirds as much precipitation fell downwind of Lake Ontario during the last 18 hours of the simulation. Apparently, ambient moisture was sufficient to produce heavy snowfall during the first 18 hours, but elimination of moisture fluxes from Lake Superior and Lake Huron reduced precipitation significantly during the latter period. Recently, the Montague storm was rerun using MM5V3 on a single 20-km grid covering the same area as the 60-km large domain. The purpose was to determine whether the interface boundary of the nested grid upwind of Lake Erie might have resulted in erroneous veering of the winds with time over Lake Erie that might explain the southward drift of the model band. However, the differences in the winds and position of the Lake Erie band were virtually identical in the single-grid and nested-grid MM5V3 runs. We conclude that the interface upwind of Lake Erie had no effect on the erroneous location of the simulated Lake Erie band. In another simulation, MM5V3 was run on the 54 km large grid and the expanded 18 km grid whose boundaries were far removed from Lake Erie. Still, there was no improvement in the simulation of the Lake Erie band. We suspect that the ETA initial analysis may have been responsible for the misplacement.
F. A method was devised to generate hourly surface predictions for a specified location from the MM5 model output. Forecasters at Buffalo NWS used this system during Year 3 in support of the Marine Weather program.
G. Researchers at SUNY Brockport and Oswego carried out a study of the March 1999 Blizzard, which produced more than 24 inches of snow in about 12 hours. They plan to simulate this event with MM5V3 using a recently acquired dual-processor Linux workstation to examine the contribution of Lake Ontario, if any, in enhancing the blizzard-related snowfall in the Rochester, NY area.
H. Josh Watson, NWS Eastern Region SSD, is working on a method to port the MM5 output into AWIPS for integration with other high-resolution data sets.
I. Jeff Waldstreicher carried out a study of the ocean-effect snowstorm near Boston using MM5V2. He presented results at two workshops and on the NWS Binghamton Local Applied STudy (BLAST) Web page.
J. Mike Jurewicz continued studies of Finger Lakes snowstorms. Results were presented at three workshops.
K. Students at Cornell University conducted numerical experiments with MM5in an attempt to better understand lake-effect snowfall in the Finger Lakes region. Predicted vertical velocity and relative humidity were found to be sensitive to initial ambient moisture and evaporation from the lakes. Upward vertical motion increased as lower tropospheric temperatures decreased.
1.3 Discuss any Major Changes to the Scope of Work
There were no fundamental changes in the scope of work. However, several additional sub-projects were undertaken as outlined above. Also, when we started, there was no plan to install MM5 at the NWS offices.
1.4 Improvement of Understanding and/or Development of Tools or Software
Researchers at the universities and forecasters at the NWS offices gained a better understanding of the capabilities and limitations of a mesoscale model and the convenience of GEMPAK/GARP for displaying model output.
SECTION 2: SUMMARY OF UNIVERSITY/NWS/AFWA/Navy EXCHANGES
2.1 Seminars or Training Workshops
During Year 1, university researchers presented seminars at NWS Buffalo and NWS Binghamton on the application of MM5 and the use of GEMPAK/GARP. Jeff Waldstreicher and Ed Mahoney held training seminars on interpretation of MM5 output at their offices, and other NWS personnel made presentations at their offices on case studies related to this project. Frequent communication between university researchers and Jeff Waldstreicher and Ed Mahoney continued during Year 2 and Year 3.
2.2 Participation of Forecasters in University Classes
Ed Mahoney assisted Prof. Ballentine and others by making BUFKIT available to university students. Jeff Waldstreicher provided numerous data sets to university faculty and students working on case studies. Jeff Waldstreicher has been a guest lecturer for the Advanced Forecasting course at Cornell for the last two years.
2.3 Joint Preparation of Publications
There was cooperation in the preparation of several conference presentations and workshop presentations as indicated in Sec. 3.2 and 3.3.
2.4 Participation of Students at the Forecast Offices
SUNY Oswego students Steve Nesbitt and Brooke Taber worked as interns at NWS Binghamton and Mike Coniglio worked as an intern at NWS Buffalo. SUNY Brockport students Debby Slocum and Marty Snyder worked as interns at NWS Binghamton.
SECTION 3: PRESENTATIONS AND PUBLICATIONS
Ballentine, R.J., A.J. Stamm, E. Chermack, G. Byrd, and D. Schleede, 1998: A mesoscale model simulation of the 4?5 January 1995 lake?effect snowstorm. Weather and Forecasting: Vol. 13, No. 4, pp. 893-920
3.2 Conference Presentations
Ballentine, R. J., A. Stamm, G. Byrd, J. Maliekal, D. Schleede, J. S. Waldstreicher, and E. Mahoney, 1998: The Montague Snowburst of 11?12 January 1997: Mesoscale Model Simulations. Preprints, 16th AMS Conference on Weather Analysis and Forecasting. Phoenix, AZ, 305-307.
Ballentine, R. J., A. Stamm, G. Byrd, J. Maliekal, D. Schleede, J. S. Waldstreicher, and E. Mahoney, 1997: The Montague Snowburst of 11?12 January 1997: Mesoscale Model Simulations. Postprints, 7th NCAR/PSU MM5 Users Workshop. Boulder, CO.
Maliekal, J. A., S. M. Rochette, J. D. Caughel, R. J. Ballentine, and A. Stamm, 2001: The Genesee Valley Blizzard of 4 March 1999: An Analysis of Forcing Mechanisms. Preprints, Symposium on Precipitation Prediction: Extreme Events and Mitigation, Albuquerque, NM, January 14-18, American Meteorological Society, 22-23.
Padavona, D. J., 1999: The May 31, 1998 Violent Tornado Outbreak: A Synoptic and Mesoscale Analysis. 24th Northeastern Storm Conference (Saratoga Springs, NY, March 12?14, 1999).
Waldstreicher, J. S., E. A. Mahoney, R. J. Ballentine, D. Schleede, J. Maliekal, and S. J. Colucci, 1998: Operational Use of a Mesoscale Model for Prediction of Lake Effect Snow in Upstate New York. Preprints, 16th AMS Conference on Weather Analysis and Forecasting. Phoenix, AZ, 393-396.
Watson, J. S., M. L. Jurewicz, R. J. Ballentine, S. J. Colucci, and J. S. Waldstreicher, 1998: High Resolution Simulations of Finger Lakes Snow Bands. Preprints, 16th AMS Conference on Weather Analysis and Forecasting. Phoenix, AZ, 308-310.
3.3 Other Presentations
Ballentine, R, A. Stamm, J. Maliekal, G. Byrd, D. Schleede, J. Waldstreicher, and S. McLaughlin, 1997: Numerical simulation of the Montague lake?effect snowstorm of 11?12 January 1997, Seventh PSU/NCAR Mesoscale Model Users' Workshop, July 1997, Boulder, CO.
Ballentine, R.J. and B. Taber, 1998: A synoptic overview of the North Country ice storm of January 1998. Presented at QUEST '98, SUNY Oswego, April 23, 1998 and the 7th U.S./Canada Workshop on Great Lakes Operational Meteorology (August 19?21, 1998 Buffalo, NY).
Coniglio, M. and R. Ballentine, 1997: 'The potential for More Efficient Real?time Modeling of Lake?effect Snow: A Case Study of the 12?13 November 1996 Central New York Snowstorm'. Honors Independent Study paper presented at QUEST 1997, SUNY Oswego, 23 April 1997.
Demetriades, N., 1997: 'Simulation of the 9?10 December 1995 Buffalo Snowstorm using the Penn State/NCAR Mesoscale Model (MM5)'. Honors Independent Study paper presented at QUEST 1997, SUNY Oswego, April 23, 1997.
Jurewicz, M.L., 2000: "A Summary of Research on Finger Lakes Precipitation Bands in Central New York."9th U.S./Canada Great Lakes Workshop on Operational Meteorology (Toronto, CA, October 25-27, 2000), and the 2000 WFO BGM Winter Weather Workshop (November 30, 2000).
Jurewicz, M. L, 1998: An analysis of an unprecedented large hail event on January 9, 1998 in central New York. 7th U.S./Canada Workshop on Great Lakes Operational Meteorology (August 19?21, 1998 Buffalo, NY). Hibbert, W.C. and E. Mahoney, 1998: Analysis of MM5 model data with Vis?5D: The Record Montague Snowburst. 7th U.S./Canada Workshop on Great Lakes Operational Meteorology (August 19?21, 1998, Buffalo, NY).
Mahoney, E. A., 2000: "Internet Visualization of new forecast products at WFO Buffalo, 9th U.S./Canada Great Lakes Workshop on Operational Meteorology (Toronto, CA, October 25-27, 2000).
Nesbitt, S., 1997: 'The Montague Storm: Synoptic Overview and Operational MM5 Analysis." Honors Independent Study paper presented at QUEST 1997, SUNY Oswego, April 23, 1997.
Niziol, Thomas, A., 1999: "WSR-88D Analysis of the June 2nd 1998 SevereWeather Outbreak in Western New York." 8th Great Lakes OperationalMeteorology Workshop (Ann Arbor, MI, August 18-20, 1999).
Niziol, Thomas, A., 2000: "Hydrological and Meteorological Aspects of the June 26th, 1998 Flash Flood on Cattaraugus Creek" 9th Great Lakes Operational Meteorology Workshop (Toronto, Canada, October 25-27,2000).
Schleede, D., 1998: Development of a stand?alone, Linux?based MM5 modeling system for the National Weather Service. Ice Cream Social at 7th MM5 Users Workshop, June 1998, Boulder, CO.
Schleede, D., R. Ballentine, G. Byrd, J. Maliekal, and T. Niziol, 1997: Development and use of MM5 in an operational setting. Seventh PSU/NCAR Mesoscale Model Users' Workshop, July 1997, Boulder.
Steiger, S. 1999: 'Analysis of the 31 May 1998 tornado outbreak in Upstate New York'. Honors Independent Study paper presented at QUEST 1999 at SUNY Oswego.
Thomas, T. 1999: 'An in?depth study of the 7 September 1998 derecho in New York State. Independent Study paper presented at QUEST 1999 at SUNY Oswego.
Waldstreicher, J. S., 2000: A Foot of Snow from a 3000 Foot Thick Cloud: The Ocean Effect Snowstorm of 14 January 1999. Binghamton Local Applied Study(BLAST) #5 (Web-based Publication Series available via the WFO BGM homepage - http://www.nws.noaa.gov/er/bgm/blast.html).
Waldstreicher, J. S., 1999-2000: "A Foot of Snow from a 3000 Foot Thick Cloud: The Ocean Effect Snowstorm of 14 January 1999." 25th Northeast Storm Conference (Saratoga, NY, March 10-12, 2000); the 9th U.S./Canada Great Lakes Workshop on Operational Meteorology (Toronto, CA, October 25-27, 2000); and the 2nd Northeast Regional Operational Workshop (Albany, NY, November 7-8, 2000).
Waldstreicher, J. S., 1999: The Labor Day 1998 Derecho. Binghamton Local Applied STudy (BLAST) #3. Publication Series available via the WFO BGM homepage - http://www.nws.noaa.gov/er/bgm/blast.html).
Waldstreicher, J. S., 1999-2000: "The Labor Day 1998 Derecho" Jeff Waldstreicher presented at the 24th Northeast Storm Conference (Saratoga Springs, NY, March 11-13, 1999), the 9th U.S./Canada Great Lakes Workshop on Operational Meteorology - Invited Presentation (Toronto, October 25-27, 2000), and a special meeting of the Central New York chapter of the American Meteorological Society (Syracuse, NY, November 22, 1999). Jeff was also interviewed about the event and my research by Granada Television Inc. for their Series Eye of the Storm (March 4, 2000).
Waldstreicher, J. S., 1999: "Operational Use of a Mesoscale Model at a National Weather Service Forecast Office." Jeff Waldstreicher - invited presentation at the 1999 Southern New England Winter Weather Workshop (Taunton, MA, December 11, 1999).
Waldstreicher, J. S., 1997-1998: "Operational Use of a Mesoscale Model for Prediction of Lake Effect Snow in Upstate New York." 6th AMS Conference on Weather Analysis and Forecasting (Phoenix, AZ, January11-16, 1998), and the 6th Great Lakes Operational Meteorology Workshop (Toronto, Canada, September 17-19, 1997).
SECTION 4: SUMMARY OF BENEFITS AND PROBLEMS ENCOUNTERED
4.1 Benefits to the University
A. During Year 1, Ed Mahoney provided BUFKIT to the universities and gave seminars on using BUFKIT at SUNY Oswego and SUNY Brockport.
B. Ed Mahoney and Jeff Waldstreicher helped the universities acquire BUFKIT profiles by helping them install the MODSND program.
C. Ed Mahoney helped SUNY Oswego set up Linux and GEMPAK/GARP on a cluster of inexpensive computers in the meteorology lab. These computers have made in much easier for students to display weather data.
D. Jeff Waldstreicher assisted students doing independent study projects at the universities by providing data for case studies. He also discussed cases with some of the students. Cases include the 31 May 1998 tornado outbreak, the 7 September 1998 derecho in central New York, the January 1998 North Country ice storm, the lake-effect snowstorm of 23-24 December 1999.
E. Ed Mahoney provided SUNY Oswego with a program to access .gif images from a Web site. Oswego used this program to collect radar images automatically from the WIXT-TV Doppler radar (with permission from the station). Since the radar, located near Syracuse, is closer to the Lake Ontario snow bands, we have increased our ability to monitor lake-effect snow events and to evaluate MM5 predictions of snow band behavior.
F. Universities students learned about the operational constraints at NWS and about the latest techniques for forecasting lake-effect snow.
4.1.1 Problems Encountered at the University
A. During Year 1, we had to run MM5 on our HP workstation that was also deployed as our Unidata server. The solution came when the NWS offices acquired their own Linux workstations and began running MM5 at their offices. Since then, SUNY Oswego has acquired two linux-based PC's to run MM5 and the LDM.
B. Our Research Associate, Don Schleede, who deserves credit for getting MM5 running and for converting MM5 output to gempak format, left SUNY Brockport at the end of Year 1. Since then, he has helped when his time allows. He was able to install MM5 on the dual processor workstations at NWS.
C. SUNY Brockport is ready to install MM5V3 on a dual Pentium III workstation, but there was a delay in shipping the computer to Brockport. The computer finally arrived in July 2001.
4.2 Benefits to the NWS
A. At NWS Binghamton and Buffalo, the MM5 output has proven to be one of the most valuable tools NWS meteorologists have for lake effect snow forecasts and warnings. MM5 has given NWS forecasters insight on snowband evolution and interaction with subsynoptic-scale features that was often not evident in other models with coarser resolution. Forecasters' understanding of lake effect snow morphology has increased, as has confidence in their forecasts. As a result, the use of this mesoscale model has directly lead to improved products and services during lake effect snow events. In particular, lead times for lake-effect snow warnings have increased, and forecasts during these events now have much greater detail regarding the location, duration, and intensity of expected heavy snowfall. MM5 is regularly cited within forecast discussions as being a primary the factor for decisions made lake effect during lake effect events. In addition, calls have occasionally been received from television forecasters inquiring about MM5 predictions, or to discuss the simulations after we began posting data to the Internet at http://tgsv5.nws.noaa.gov/er/buf/mm5/main.html.
B. At NWS Buffalo, the MM5 output has been used to improve marine forecasts over Lakes Erie and Ontario. Ed Mahoney also developed a method to generate text files of MM5 hourly surface predictions output for various over-water and on-land forecast points.
C. It was discovered during the first year of the COMET project that Finger Lakes snow events (a specific type of lake effect snow event where snowbands developed to the lee of the Finger Lakes in central New York) was a particular challenge for forecasters at NWS Binghamton. These events were studied in-depth through a collaboration of BGM forecasters and Cornell and SUNY-Oswego faculty and students. In addition, special higher resolution MM5 simulations (with an inner grid resolution of 2.5 km) were made for some of these events. A conceptual model and series of forecast techniques were developed for these events. As a result, forecasts for these type of lake effect snow events have dramatically improved.
D. Several studies were conducted examining MM5 performance during severe convection and flood events. For some of these cases, MM5 sensitivity studies were conducted by running simulations with different parameterization schemes utilized. These studies gave forecasters insight into the strength and weakness of MM5 during convective events, as well as a better appreciation and understanding of the importance of various parameterization schemes (especially the cumulus convection schemes).
4.2.1 Problems Encountered at the NWS
A. During Year 1, we had a problem with timeliness in receiving MM5 output. At that time MM5 was being run on a 75-MHz HP workstation at SUNY Oswego. Results from the 24-hour simulations were not available at NWS offices until about eight hours after the model run was started. As a result, only the last 12 hours of the simulation were of any use to forecasters. The problem was reduced considerably during Year 2 when the NWS offices acquired dual processor Linux workstations. After MM5 was set up at the Binghamton and Buffalo offices, model output from the 36-hour simulations was available to forecasters within three hours of the start time.
B. During Year 1, we had access to Great Lakes water temperature using a file from Unidata that was updated daily. By Year 2, the NMC computer that generated the file was taken off line and the water temperature was unavailable from Unidata. As a solution, monthly mean climatological values were used during Years 2 and 3. More realistic Great Lakes water temperatures are now available with the GRIB files used to initialize MM5V3 at SUNY Oswego.
C. After installing MM5V3 on computers at NWS Binghamton and NWS Buffalo during Spring 2001, we found that it took between 40 and 45 minutes to download the ETA212 GRIB files that are used to initialize MM5V3. These files are being downloaded in less than 5 minutes at SUNY Oswego. So far, MM5V3 has not been set up for realtime use at the NWS offices.
D. We have not been able to add ice-cover data to the MM5 initialization.