IMPROVING OPERATIONAL FORECASTING OF LAKE-EFFECT
SNOWSTORMS IN THE EASTERN GREAT LAKES REGION
OVERVIEW
The regions to the lee of Lake Erie and Lake Ontario experience some of the most severe
snowfall events in North America (Niziol et al., 1995). Lake-effect snows are mesoscale,
convective precipitation events that develop downwind of the Great Lakes during the late fall and
winter. These storms typically span the mesogamma (2-20 km) to mesobeta (20-200 km) size
range as defined by Orlanski (1975). Lake-effect snowbands can produce extreme snowfalls,
with accumulations exceeding 150 cm (60 in) during a quasi-stationary, multiday event (Niziol
1989; Sykes 1966).
Across upstate New York, lake-effect snowstorms frequently (e.g., several times a year)
produce very heavy snowfall that severely impacts more than 4.5 million people who live
downwind of the lakes. The National Weather Service offices in Buffalo and Binghamton have
forecast and warning responsibility for these areas. The localized nature of these storms presents
a unique challenge to operational meteorologists to provide detailed forecasts of the location and
intensity of snowband development with sufficient lead time to carry out the basic mission of
NWS which is to protect life and property.
The inability of the larger-scale numerical models to adequately simulate lake-effect snow is
discussed by Niziol et al. (1995). As a result, NWS offices have over the years developed a
plethora of local operational forecast techniques, several of which are described in Niziol et al.
(1995). In addition, rapid increases in computer resources and the availability of high-resolution
datasets have made it possible to run mesoscale models which can simulate mesoscale circulations
such as lake-effect snowbands, and produce model forecasts in a timely manner.
Researchers at SUNY Oswego and SUNY Brockport participated in 1992 with forecasters at
NWS Buffalo in a COMET Partners Project to improve short-range prediction of lake-effect
snowstorm behavior, and to increase cooperation between operational forecasters and the
university community (Ballentine, et al., 1993). A mesoscale model was tested for 8 lake-effect
snow events using initial data from the nested grid model (NGM). A protocol was established to
run the model operationally on a 486 microcomputer at NWS Buffalo. The model domain covered
the Lake Ontario region with a grid having 10 km resolution. For the Player version of the
model, it took about 3 hours to complete a 24-hour simulation. In each case, the model produced
a snowband circulation, and in most cases, the location of the model snowband was near the
initial location of the observed band. However, in several cases, the movement of the model
snowband was not in agreement with observations.
The model used in the Partners Project was run over flat terrain. It used simple boundary
layer and cloud physics packages, and a very crude form of synoptic-scale forcing. A single
sounding (usually Buffalo) was used to initialize the model. Prognostic variables were assumed
to be horizontally homogeneous initially. Changes in the geostrophic wind were prescribed at
Page 2.
every grid point based on projections by the NGM interpolated to Buffalo. Not surprisingly, the
model was unable to predict changes in snowband location and intensity for cases in which the
synoptic-scale fields changed significantly during the period of interest. Nevertheless, the ability
of the model to simulate snowband development and to predict snowfall rates fairly accurately
was encouraging to both forecasters and researchers.
We wish to expand this research in the form of a cooperative effort among the State Univer-
sity of New York Colleges at Oswego and Brockport and Cornell University with the Weather
Service Forecast Offices in Buffalo and Binghamton to improve the operational prediction of
lake-effect snowstorms. We propose to develop a procedure for using the workstation version of
the Penn State/ NCAR Mesoscale Modeling System (mm5ws) (Haagenson,l994) to make real-
time predictions of the development and behavior of lake-effect snowstorms downwind of the
eastern Great Lakes, and to provide model output in the form of GEMPAK-generated metafiles to
forecasters at the National Weather Service offices in Buffalo and Binghamton.
L. PROJECT OBJECTIVES
1.1 Our long-term goals are:
* To improve the operational prediction and subsequent forecasts and warnings of mesoscale
lake-effect snow events downwind of the eastern Great Lakes.
* To expand the interaction and exchange of knowledge between operational meteorologists at
the National Weather Service Forecast Offices in Buffalo and Binghamton and researchers at
SUNY Oswego, SUNY Brockport, and Cornell University.
1.2 Our Short-term Objectives are:
* To develop and test the mm5 modeling system (i.e., initialization programs, mesoscale model,
and post-processing programs) with the intention of freezing the modeling system for
operational use by the start of the 1996-97 lake-effect snow season.
* To conduct joint studies of the structure and behavior of lake-effect snowbands including
verification and evaluation of the mm5 modeling system using WSR-88D and GOES-8 data as
well as detailed snowfall reports from the Upstate New York Snow Spotter Network. For
example, we will compare WSR-88D data and GOES-8 imagery with model predictions of
snowband intensity, location and movement.
+ To archive model input data and raw model output for some of the more interesting lake-
effect snow events. Model output will be used in case study analysis to learn more about the
structure and dynamics of mesoscale snowband circulations, and to improve our under-
standing of the role of larger-scale environmental factors, including lake-lake interactions, in
Page 3.
snowband formation and behavior. We will also archive WSR-88D data, GOES-8 imagery,
and snowfall reports for comparison with model predictions of snowband location, move-
ment, and precipitation rate. The model input data from all of the events will be archived so
that we can run experiments to test the sensitivity of model results to changes in grid
resolution, grid geometry, model physics, lake temperature, and the distribution of ice cover.
* To hold a series ofjoint seminars between the NWS and University community to presenr
results of cooperative lake-effect studies and to discuss and review the progress of this
project. These meetings will take place when students are available, and they will be
scheduled to coincide with critical phases of this project. A larger meeting similar to the
1993 Symposium on Central New York Weather Forecast;ng, or a regional workshop similar
to the 4th United States/Canada Workshop on Great Lakes Operational Meteorology
(September 1995) may also be held.
* To provide training sessions given by university staff for NWS personnel on using GEMPAK
and Ntrans to analyze and display gridded mesoscale model data, and on interpreting mm5
model output. This training will facilitate the transition of operational forecasters to an
AWIPS-like environment by giving them hands-on experience using a Hewlett-Packard work
station, and a common software platform to display gridded model data. GEMPAK and
Ntrans can also be used to display gridded data currently available from operational models
such as the NGM and ETA (early and mesoscale versions), global spectral (AVN and MRF).
and the RUC model.
* To provide instruction for university students and faculty regarding the use of WSR-88D
products, GOES-X imagery, and operational forecast and warning techniques. This
instruction will be given by NWS forecasters during visits to SUNY Brockport, SUNY
Oswego, or Cornell, and when small groups of students are invited to NWS Buffalo and/or
NWS Binghamton
1.3 Potential Benefits of the Operational Use of Mesoscale Models
Existing operational forecast models are not capable of explicitly predicting mesoscale lake-
effect snow events. Mesoscale models have been under nearly continuous development for
research applications at various universities during the past 25 years. These models have been
successful in simulating a variety of mesoscale circulations. It is anticipated that the mm5 model
results integrated with new high-resolution remote sensor technologies, such as the WSR-88D and
GOES-8, will lead to:
* increased lead time in predicting the onset of lake-effect snow events;
* improved prediction of the change in location of the snowbands;
* improved resolution in predicting the areas impacted by lake-effect snow;
Page 4.
* improved prediction of snowband intensity and duration which will lead to greater accuracy in
predicting snowfall amounts.
These improvements will help NWS forecasters to provide more accurate information to
critical agencies such as highway and transportation departments, school systems, as well as local
commerce and industry.
2. PROPOSED EFFORT
For this project we propose to use the workstation version of the Penn StateiNCAR
Mesoscale Model (mm5ws) to simulate lake-effect snowstorms and other mesoscale circulations
in the eastern Great Lakes region. Our primary objective is to develop a method to provide mm.5
predictions of lake-effect snowband development, movement, and precipitation rate to NWS
forecasters at Buffalo and Binghamton. Since we anticipate that the workstations at NWS Buffalo
and Binghamton will not be available to run mm5 (at least for the next few years), we propose to
run the model on the Hewlett-Packard 9000 Model 735/100 workstation 'monsoon.weather.
brockport.edu' at SUNY Brockport. When 'monsoon' is not available, or when network
problems prevent access to 'monsoon', we will run mm5 on 'blizzard.oswego.edu' , the HP
Model 715/75 workstation at SUNY Oswego. In consultation with NWS forecasters, scientists at
SUNY Brockport and SUNY Oswego will develop user-friendly UNIX scripts to execute the
entire modeling system. Model output in the form of GEMPAK-generated metafiles will be
transferred electronically to NWS Buffalo and Binghamton using channels such as Internet.
Forecasters will view output using Ntrans which is a graphical user interface for GEMPAK.
The advantages of using mm5 include:
* up-to-date options used in the initialization package to provide accurate initial fields over and
upwind of the Great Lakes;
* accurate representation of advection owing to state-of-the-art numerics;
* flexibility of grid nesting configurations allowing us arrive at the optimal balance between
grid resolution and model runtime;
* realistic terrain to account for orographic effects including precipitation enhancement;
+ sophisticated model physics.
2.1 Modeling Protocol
An overriding concern is to provide mm5 output to NWS forecasters in a timely manner. Our
goal is to have results from a 36-hour simulation available to forecasters (in graphical form)
Page 5.
within 12 hours of the time of the initial data. For example, if a forecaster at either NWS Buffalo
or NWS Binghamton calls SUNY Brockport or SUNY Oswego on a Tuesday evening to request a
36-hour simulation (from 002 Wednesday to 122 Thursday) of a potential lake-effect snowstorm,
the person on duty will activate the procedure to:
* acquire the 002 upper-air and surface data files (available from Unidata and other sources) by
approximately 0130 Z;
* run programs to create initial conditions (in sigma coordinates) over the mm5 nested grid;
* run mm5 on 'monsoon' or 'blizzard' out to 36 hours;
* convert mm5 output to GEMPAK format and produce a set of metafiles (graphics products);
* transfer electronically the GEMPAK metafiles to NWS Buffalo and NWS Binghamton.
We will automate this procedure as much as possible to minimize the workload of the
operator on duty, and to make the end-products available to forecasters in a timely manner
(before 122 Wednesday in the example above). Faculty involved in the project and students at
SUNY Brockport and SUNY Oswego will be trained in implementing this procedure. During
periods when lake-effect snow is favored, we will maintain 24-hour coverage (including week-
ends) by assigning trained student operators to shifts at times when faculty participants are not on
campus. During these periods, a faculty member or the Research Associate will be on-call at
night and on weekends.
Since snowfall downwind of Lake Ontario and Lake Erie has the maximum societal impact,
we want to provide forecasters with the option of previewing hourly model forecasts in graphical
form (for example, snowfall rate and vertical velocity on the nested grid) midway through the
model run. This 'early look' will enable forecasters to integrate the latest mm5 predictions into
the cycle of forecasts and advisories when the situation warrants. For example, a forecaster who
makes a request for a 36-hour model run on a Tuesday evening will be able to view loops of
selected intermediate forecast products (verifying out to at least 18Z Wednesday) by about 071,
Wednesday. The complete set of GEMPAK metafiles (verifying out to 122 Thursday) will be
available to NWS forecasters by 122 Wednesday.
A sample grid configuration is shown in Figure 1. In our final configuration, the large grid
will extend far enough upwind to ensure that any feature which could affect the eastern Great
Lakes region during a 36 hour period is included in the initial data. The nested grid will cover
most of New York State, Lake Ontario, and enough of Lake Erie to simulate the formation of
snowbands that affect western New York. Since the nested grid will also cover most of Georgian
Bay and the eastern portion of Lake Huron, it will be possible to simulate the formation of snow
bands that develop near the Bruce Peninsula which separates these bodies of water. Satellite
photographs indicate that bands which form near the Bruce Peninsula interact with snowbands
over Lake Ontario (Niziol, 1995).
The vertical grid structure is chosen to ensure that there is good resolution in the lowest 3 km
of the atmosphere (at least 8 model layers) and adequate resolution throughout the troposphere.
Page 6.
We have used 23 layers for most of our preliminary experiments. By reducing the number of
vertical layers, we can decrease the 'runtime" of the model. We will conduct sensitivity
experiments to determine the optimal number and location of model layers, and the horizontal
resolution, which will permit completion of the 36-hour simulation before the next cycle of
upper-air observations. We anticipate that the eventual 'optimized' configuration will consist of
an outer grid with a horizontal resolution of approximately 36-45 km, and an inner grid with a
resolution of 12-15 km. Note that mm5 requires a 3:1 outer grid-to-inner grid ratio.
2.2 Initial Data Programs
One of the most important parts of this project is to port the NCAR initial data programs
(Datagrid, Rawins, and Interp) to our HP workstations. Workstation versions of these programs,
which were written for the CRAY supercomputer, have not yet been officially released by the
Mesoscale and Microscale Meteorology Division of NCAR. Other researchers have ported these
programs to IBM, DEC, and SGI workstations. Don Schleede at SUNY Brockport is converting
the Fortran code for the initial data programs to HP Fortran. The programs will be modified,
where necessary, to apply to operational forecasting of lake-effect snowstorms. For example, an
efficient method will be included to input the latest information concerning lake temperature and
the distribution of ice cover on all of the Great Lakes. We will search through several data bases
for Great Lakes water temperature, snow cover data, and lake ice cover. We expect that some of
this data will be available on a daily basis from the Great Lakes Environmental Research Labora
tory (GLERL) at the University of Michigan and from the NOAA Ice Center at Suitland, MD.
Using the program Datagrid, we compute a 'first guess' field for the objective analysis on
constant pressure surfaces. In a research mode, scientists typically use data from the archive of
NGM analyses available at NCAR. We receive NGM and ETA model analyses and forecasts
from Unidata with backup capability directly from NMC. Since the initial analyses are not
available until about 4 hours after the observation time, we will use the 12-hour forecast from the
previous NGM or ETA model run as input to our version of Datagrid.
We complete the objective analysis using the program Rawins together with the most recent
surface and upper-air observations to 'correct' the first guess supplied by Datagrid. We will test
the effectiveness of using the Cressman objective analysis option versus the multi-quadric analysis
option (Nuss and Titley, 1994) for several lake-effect snowstorm case studies. The latter analysis
scheme may be better suited to the lake-effect snow forecast problem because of the scarcity of
upper-air observations over Canada. The program Interp is used to interpolate the analysis on
pressure coordinates (from Rawins) to the sigma levels of the model.
2.3 Model Physics
The high-resolution scheme of Blackadar (Zhang and Anthes, 1982) is used to compute
surface fluxes, the height of the planetary boundary layer, and turbulent transports of heat,
moisture and momentum within the PBL. A surface energy budget is used to predict the
Page 7.
the ground temperature. We are testing a version of the Blackadar scheme which has been
modified to treat the case of the exceptionally large air-water temperature differences which we
sometimes encounter over the Great Lakes in winter.
The explicit moisture scheme is used to predict precipitation, calculate cloud water content,
and to account for the release of latent heat. In the simple ice scheme (Dudhia, 1992), clouds are
composed entirely of liquid droplets and precipitation is all rain if the temperature is warmer than
O"C. Clouds consist entirely of ice crystals and precipitation is all snow if the temperature is
below O"C.
2.4 Post-processing of mm5 Output
Since a program has been written to convert mm5 output files to GEMPAK format (Schleede,
et al., 1995), we can run any GEMPAK program to produce graphical displays of mm5 output.
The versatility of GEMPAK allows for a wealth of displays including multiple analyses of model
output on horizontal or vertical cross sections, model-generated soundings, and looping of any
field. Difference fields of model output minus observed data may also be displayed.
2.5 Transfer of Graphics Piles to NWS
At the present time, both NWS Buffalo and NWS Binghamton have 56-KB lines for Internet
communication. NWS Eastern Region Headquarters is also in the process of establishing a Wide
Area Network (WAN). Plans for this WAN, which will consist of a frame relay network with an
initial bandwidth of 56 KB with burst rates of 128 KB, also include an Internet Gateway. Since
there will be times when the network is slow, we will establish two lists of products for electronic
transfer to Buffalo and Binghamton. The 'long list', consisting of the complete set of GEMPAK
metafiles ordered by NWS, will be sent when network communication is normal. A 'short' list
of the most essential graphics products (e.g., the loop of hourly precipitation on the nested grid)
will be sent when the network is slow or subject to frequent outages. NWS forecasters, in
consultation with GEMPAK programmers, will make the final decision as to which products are
on which list. In the event of prolonged network failure, we will use telephone data lines (or
facsimile) to transmit the graphics.
2.6 Model Verification and Evaluation
The mm5 predictions on the nested grid will provide the first opportunity to use high-
resolution model output for operational forecasting for many NWS forecasters. We expect that
forecasters will learn to make good use of the mm5 output after learning the strengths and
weaknesses of the model as applied to lake-effect snow simulation. Forecasters at Buffalo and
Binghamton will evaluate the model performance by assembling case studies including regional
surface analyses, snowfall reports, satellite photographs, and WSR-88D data from Binghamton,
Buffalo, and West Leyden. Of particular interest is determining the ability of the model to
predict the onset time, shift of location, and precipitation rate from snowbands downwind of
Page 8.
Lakes Erie and Ontario. In addition, the ability of the model to predict the height of the capping
inversion, the low-level wind field in the vicinity of the snowband, and the distribution of
temperature and humidity downwind of the lakes will be evaluated. Once the modeling system is
'frozen', we expect that forecasters will learn how to adjust their forecasts based on a knowledge
of model biases and systematic errors.
3. PROJECT SCHEDULE
We propose a 3-year project beginning 1 January 1996. During the first half of Year 1, we
will concentrate on testing various grid configurations, minimizing model runtime, and stream
lining the operational procedure. We will develop efficient scripts to automate model setup and
post-processing, During this period, we will train faculty and students in operating the modeling
system. We will simulate a few lake-effect snow events during the first two months using a
preliminary version of the model. This will enable us to find out right away whether there are
any serious problems with initialization, post-processing of model output, and transmission of the
forecast products to forecasters at the times they when need it most. We expect to get some early
feed-back from forecasters on the benefits and limitations of using mm5 predictions in an
operational environment. We plan to hold the first joint seminar in late April 1996 to discuss use
of the modeling system, and to review lake-effect events during the 1995-96 winter season.
During Summer 1996, we will do case study analyses and model evaluation for these events. We
will also carry out sensitivity studies with mm5 to determine whether a change in the input data,
model physics, or grid configuration leads to a significant improvement in model predictions.
The final version of the modeling system will be ready before the beginning of the 1996-97
lakeeffect snow season. We expect to hold the second joint seminar in October 1996 to explain
any and all changes that have been made in the modeling system and post-processing procedure.
Work on case study analysis and model evaluation will also be presented at this seminar.
We expect that at least 12 significant lake-effect snowstorms will need to be simulated in an
operational environment before we can make a valid assessment of the usefulness of the mm5
modeling system. The events should cover a wide range of synoptic-scale conditions and storm
intensities. The likelihood of having this many significant events to simulate will be much
greater if the project covers two full lake-effect snow seasons. Therefore, we propose a three-
year project so that we will be able to run mm5 operationally during the winters of 1996-97 and
1997-98. The project schedule is outlined in the table below:
Project Task Year 1 Year 2 Year 3 Develop and test mm5|*********** | | | Run mm5 operationally| ****|***** ****|***************| Case study analysis| **********|***************|***************| gempak/mm5 training|****************|***************|***************| NWS/university fcst| **********|***************|***************| technique development
Page 9.
Much time will be needed for case study analysis, model evaluation, and the writing of
journal articles. Most of this work will be done during Summer 1997 and during all of 1998.
The GEMPAK training and the instruction in using mm5 predictions will be an ongoing task
that can begin soon after the project starts. On several occasions, faculty and staff from SUNY
Brockport and/or SUNY Oswego will travel to NWS Buffalo and NWS Binghamton on days
when a group forecasters will be available to receive training. Some of the GEMPAK and mm5
training will take place at the colleges when NWS forecasters can make trips to SUNY Brockport
or SUNY Oswego.
Trips to NWS Buffalo and NWS Binghamton for demonstrations of GOES-8 and WSR-88D
applications, and explanation of NWS forecast and warning techniques will be scheduled at times
when most students are on campus, and when NWS personnel are available. It may be possible
to combine one of these meetings with the October 1996 seminar on the final modeling system.
NWS forecasters will travel to the universities to present seminars to meteorology students and
faculty on forecast technique development. In some cases, these trips will coincide with meetings
between forecasters and researchers to discuss the progress of the modeling
effort.
4. COOPERATIVE PARTICIPATION
The relationship between project tasks has been described in Section 2. A broad outline of
the roles of the universities and the National Weather Service is summarized below.
4.1 Role of the Universities
The universities will provide expertise in numerical modeling, data management, GEMPAK
programming, and UNIX systems administration for the purpose of developing and testing a
workstation version of the Penn StatelNCAR mesoscale modeling system (rxm5ws). University
participants will run mm5 on a workstation at SUNY Brockport or SUNY Oswego, at the request
of forecasters from NWS Buffalo and/or NWS Binghamton, during periods of significant lake-
effect snow potential, and transmit mm5 output in the form of GEMPAK-generated metafiles
electronically to NWS offices in a timely manner. Faculty and students at the universities will
participate in model evaluation and in case study development for selected lake-effect snow
events. University faculty, as well as the Research Associate, will train NWS forecasters at
NWS offices in Buffalo and Binghamton and on the college campuses in using GEMPAK and
Ntrans and in interpreting mm5 model output.
4.2 Role of the National Weather Service
Forecasters at WSFO Buffalo and Binghamton will evaluate the usefulness of mm5 predic
tions in making lake-effect snow forecasts and in issuing lake-snow advisories. Forecasters will
attempt to identify biases and systematic errors in mm5 predictions and report these problems to
Page 10.
university researchers. Forecasters will make recommendations as to which sets of graphical
products are most useful, and which products should be available at critical rimes in the
operational forecast cycle.
4.3 Contributions of the University
1) SUNY Oswego has approved a partial waiver of indirect costs in support of this project. For
each year, SUNY Oswego will waive $1912 from the full indirect cost of $3725. SUNY Brock-
port will waive $3115 of overhead in Year 1, $3380 of overhead in Year 2, and $2704 of
overhead in Year 3.
2) The time of 5 undergraduate meteorology majors. The students from SUNY Oswego and
SUNY Brockport will be trained to activate the modeling system when faculty are unavailable.
Students from Oswego, Brockport, and Cornell will assist faculty in lake-effect snowstorm case
study development and in evaluation of the numerical model.
3) The time of 3 faculty members. Dr. Ballentine will design the mm5 grid configuration, test
model physics options, supervise student operators at SUNY Oswego, participate in case study
development, and train NWS forecasters in using GEMPAK and interpreting mm5 output. Dr.
Maliekal will donate his time, at no cost during Year 1, to supervise student operators at
Brockport. He will supervise students involved in model evaluation and case study analysis
during Year 2 and Year 3. Dr. Colucci will supervise a student at Cornell working on lake-effect
snow in the Finger Lakes region during Year 1, and he will supervise students working on model
evaluation and case study analysis during Year 2 and Year 3. Dr. Colucci will donate his time at
no cost.
4) The time of the Research Associate. Donald Schleede will convert the NCAR initial data
programs to workstation format, solve any problems related to acquisition and preparation of
model initial data, write a UNIX script to automate model operation, write a script to generate a
set of GEMPAK metafiles from model output, and optimize the mm5 code to minimize model
runtime.
5) University workstations and software development. The modeling system will be run on the
HP workstation at SUNY Brockport with the HP workstation at SUNY Oswego as a backup. A
program to convert mm5 output to GEMPAK format, and a program to make it possible to use
NWS data to initialize the mm5 model will be donated by the Research Associate.
4.4 Contributions of the National Weather Service
1) Time of WSFO forecasters to display graphical forecast products using Ntrans and to attempt
to integrate these mm5 predictions into the regular cycle of operational forecasts and advisories.
page 11.
2) Time of WSFO meteorologists to evaluate the usefulness of the mm5 predictions, and to
discuss operational problems in using mm5 predictions with university researchers.
3) Time of WSFO meteorologists to write papers and make conference presentations on lake-
effect snow case studies and on the usefulness and limitations of short-term mm5 predictions.
4) Time of WSFO meteorologists to host visits by university students and faculty at NWS offices
and to travel to the universities to present seminars on using WSR-88D and GOES-8, and to
discuss developments in specialized forecast and warning techniques.
5) Time of WSFO meteorologists to archive WSR-88D data, GOES-8 imagery, and reports from
the New York State Snow Spotters Network to provide 'ground truth' for
precipitation.
5. PROJECT PARTICIPANTS
5.1 Project Participants
Jeff Waldstreicher
Jeff Waldstreicher has been Science and Operations Officer at the National Weather Service in
Binghamton, NY since March 1993. Previous assignments include: Scientific Services Division,
NWS Eastern Region Headquarters, Bohemia, NY, 1989-1993; National Weather Service
Forecast Office, Boston MA, 1987-1989; and the Techniques Development Laboratory - Local
Applications Branch, National Weather Service Headquarters, Silver Spring, MD, 1984-85. He
has written several articles on winter and severe weather forecast techniques, including editing
and co-authoring a Cpaper series on forecasting winter weather in the eastern United States for
Weather and Forecasting. During the winter of 1994-95, he lead a COMET Partners Project
with the State University of New York College of Environmental Science and Forestry in
Syracuse, NY entitled '.1 Preliminary Investigation of WSR-88D Data for Winter Hydrometeor-
ological Events in Upstate New York.
Edward Mahonev
Ed Mahoney arrived as the Science and Operations Officer of the National Weather Service in
Buffalo during this past year. Graduating from SUNY Albany and the University of Oklahoma,
he started his professional career as a Wing Weather Officer in the Air Weather Service and was
eventually responsible for the deployment of the WSR-74C weather radar at selected Air Force
installations. In the late 1980's he was the Operations Test Director for the Air Force Test and
Evaluation Center (AFOTEC) of NEXRAD that assessed the 'fly-off' between UNISYS and
Raytheon Doppler weather radar designs. Ed transferred to the National Weather Service where
he helped design and implement the WSR-88D Doppler Radar Hotline. He has published several
articles in Federal Meteorological Handbook No. 11, and he served as Editor and contributing
author to the WSR-88D Operational Support Facility's Tales from the Hotline.
Page 12.
Thomas Niziol
Tom Niziol has been employed at the National Weather Service in Buffalo, NY for the past
15 years, where he is currently a Lead Forecaster and the lake-effect snow focal point. He has
published numerous articles in professional journals about the characteristics and the operational
forecasting of lake-effect snow. In addition, he has served as the National Weather Service
principal investigator for various research projects on the eastern Great Lakes.
Robert Ballentine
Robert Ballentine is an Associate Professor of Earth Sciences at SUNY Oswego. IIe has over
20 years of experience designing, testing and running numerical models. He has written articles
on New England coastal frontogenesis and lake-effect snowband simulation. He is currently co-
PI . and project director at SUNY Oswego, for a 3-year NSF grant to study the dynamics and
structure of winter storms in the Great Lakes region.
Steven Colucci is an Associate Professor of Atmospheric Sciences at Cornell University. He
has written articles on synoptic-scale phenomena, and on the evaluation of numerical models. He
is currently PI of a 3-year NSF project to study the utility of eta-model forecast ensembles. Prof.
Colucci has more than 15 years experience supervising undergraduate research projects.
Jose Maliekal
Dr. Maliekal is an Assistant Professor of Earth Sciences at SUNY Brockport. He is currently
the PI for a National Science Foundation grant on tropical climatology. He has experience in
supervising undergraduate research assistants.
Donald Schleede
Don Schleede has several years experience in systems administration and scientific program-
ming. He has made conference presentations on the use of computer methods in meteorology and
on using GEMPAK to display mm5 output. He was a technical consultant at Unidata's Northeast
Regional Workshop at SUNY Brockport in August 1993. He is currently the Research Associate
for two National Science Foundation research grants at SUNY Brockport.
5.2 References cited in the text
Ballentine, R.,T.Niziol, and G.Byrd, 1993: A COMET Outreach Partners Project to Develop
Cooperation Between the National Weather Service Forecast Office at Buffalo New York
and the Meteorology Programs at the State University of New York (SUNY) Colleges at
Oswego and Brockport. Final Report to the COMET Outreach Program Office.
Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment
using a mesoscale two-dimensional model. J. Atmos.Sci., 46, 3077-3107.
Page 13.
Haagenson,P.L.,J.Dudhia,G.Grell and D.Stauffer, 1994: The Penn Stare/NCAR mesoscale model
(mm5) source code documentation. NCAR Technical Note, NCAR/TN-392 +STR, 200pp.
Niziol,T.A.,1989: Some synoptic and mesoscale interactions in a lake-effect snowstorm.
Postscripts, 2nd National Winter Weather Workshop, Raleigh, NC, NOAA, 260-269.
Niziol,T.A., W.R.Snyder, and J.S.Waldstreicher, 1995: Winter weather forecasting throughout
the Eastern United States. Part IV: Lake-effect Snow. Wea. Forecasring, 10, 61-77.
Nuss, W.A. and D.W. Titley, 1994: Use of multi-quadric interpolation for meteorological
objective analysis. Mon. Wea. Rev., 122, 1611-1631.
Orlanski, L., 1975: A rational subdivision of scales for atmospheric processes. Bull. Amer.
Mereor. Sec., 56, 527-530.
Schleede,D. ,J.Case,L.Keshishian,R.Ballentine,G.Byrd 1995: Processing of mm5 output files
into an interactive graphical environment called GEMPAK. Fifth PSU/NCAR Mesoscale
Model Users Workshop, Boulder, CO.
Sykes, R., 1967: The blizzard of '66 in central New York: A legend in its time. Weathenvise,
19, 241-247.
Zhang, D.L. and R.A. Anthes,1982: A high-resolution model of the planetary boundary layer
Sensitivity tests and comparison with SESAME-79 data. J. Appl. Meteor.,
21, 1594-1609.
5.3 Other References Related to this Proposal
Ballentine, R., 1980: A numerical investigation of New England coastal frontogenesis. Mon.
Wea. Rev., 108, 1479-1497.
Ballentine, R., 1982: Numerical simulation of land breeze-induced snowbands along the western
shore of Lake Michigan. Mon. Wea. Rev., 110, 1544-1553.
Ballentine,R. ,E.Chermack,A. Stamm,D.Frank,M. Thomas,and G.Beck,1992: Preliminary
numerical simulations of the 31 January 1991 lake-effect snowstorm. Proceedings: 49th
Eastern Snow Conference, Oswego, NY.
Ballenline,R.,G.Byrd, and T.Niziol, 1993: An operational forecast model for lake-effect snow
storms. 12th AMS Conference on Weather Analysis and Forecasting, Vienna, VA.
Ballentine,R. ,A.Stamm,D.Schleede, and G.Byrd, 1994: Numerical simulation of the 45 January
1994 East Coast snowstorm. IOth AMS Conf. on Numerical Weather Prediction, Seattle.
Page 14.
Ballentine,R.,D.Schleede.L.Keshishian,G. and J.Case, 1995: Simulation of lake effect
snowstorms using mm5. Fifth PSU/NCAR Mesoscale Model Users Workshop, Boulder,CO.
Byrd,G., Bikos,D., Schleede,D., and Ballentine, 1995: Numerical investigation of lake-lake
interaction. 13th AMS Conf. on Weather Analysis and Forecasting. Dallas.
Nouck,R.E., J.S.Waldstreicher, J.M.Hassett, and P.F.Blotman, 1995: Preliminary investigation
of WSR-88D data for winter hydrometeorological events in upstate New York. Proceedings,
52nd Eastern Snow Conference. Toronto, Ontario, Canada. (In press).
Maglaras, G.J., J.S.Waldstreicher, P.J.Kocin, A.F.Gigi, and R.A.Marine, 1995: Winter weather
forecasting throughout the Eastern United States. Part 1: An Overview. Wea. Forecasting.,
10, 5-20.
Niziol, T.A., 1987: Operational forecasting of lake-effect snow in western and central New
York. Wea. Forecasting, 1, 311-321.
Niziol, T.A. and S.F.McLaughlin, 1992: The use of high resolution data from the Nested Grid
Model for the prediction of lake-effect snows. Postprints. 3rd NWS Winter Workshop,
Portland, OR, NOAA/NWS, 179192.