1. Project Objectives and Accomplishments
During the final year of this COMET Cooperative Project, emphasis was placed on finishing research work and incorporating knowledge gained into National Weather Service (NWS) operations. Four presentations were made at the American Meteorological Society's 20th Conference on Severe Local Storms and two University at Albany/SUNY (UA) graduates students completed Master's theses based on this research. As part of ongoing office training, a WEB-based training module was created and incorporated into NWS Albany's spring severe weather drill. It will soon be made available to other NWS offices in the Northeast. It is available at: http://web.nws.cestm.albany.edu/comettraing/cometmain.htm (Please note: requires MS Internet Explorer to view). In addition, the project Web page (http://web.nws.cestm.albany.edu/comettraingnet/COMETMAIN.HTM) was updated as new information became available.
Project 1: Using WSR-88D, satellite and lightning data to establish guidelines for differentiating tornadic and non-tornadic thunderstorms.
In Project I we examined WSR-88d archive level IV data for 86 tornadoes in the northeastern United States. Data for these cases were obtained from National Weather Service Forecast Offices around the Northeast. The 86 tornadoes were classified according to the structural characteristics of the thunderstorms producing the tornadoes. Understanding the types of convective storms that produce tornadoes may be useful knowledge in the warning process. Supercells produced about half of the 86 tornadoes examined. More than a third of the tornadoes developed with bowing lines or cells, with most (two thirds) of these tornadoes forming on the bulging portion of the bow. Appendages or hooks were observed in less than half the tornadoes, and boundary interactions appeared to have played a role in tornadogenesis in about one fifth of the storms. Gate to gate rotating velocity couplets of varying intensities were identified with a majority of the tornadoes. However, more than a third (mostly F0 and F1) were associated with non-rotating wind maxima. A number of tornadoes produced no discernable radar signature due to time and spatial sampling limitations of the radar.
Analysis of the tornado data has shown that the WSR-88D-calculated gate-to-gate shear (S) of rotating velocity couplets associated with many tornadoes is useful in identifying tornadic storms. The effects of beam spreading with range in both the actual shear calculation and the ability of the radar to resolve storm features required that the observed shear values be adjusted for range. The adjusted shear values, which provide evidence of the strength of small scale (2 km or less), low-level mesocyclones were compared to the larger scale (2 km to 15 km), midlevel mesocyclone strength in order to assess a storm's tornadic potential. In addition, 34 non-tornadic mesocyclones were examined for comparison with the tornado cases. Plots of large scale (mesocyclone) versus small scale (gate-to-gate shear) circulation were created for the tornadic and non-tornadic cases for the entire data set and for a subset of supercell storms. Using the combination of the gate-to-gate shear and rotational velocity of the mesocyclone (Vm) showed skill in identifying tornadoes, especially in the supercell cases. It should be noted that lead time was not factored in. The statistical correlation between tornadic intensity (F scale) and S, normalized for range, was 0.542. The correlation between tornado intensity and the gate-to-gate rotational velocity (Vr) used in the calculation of S, was similar (0.533). (If Vr was range normalized the correlation increased slightly to 0.561.) Using Vr instead of S would yield similar results and is operationally easier to use. Nomograms using both S and Vr have been provided to forecasters.
Mean characteristics (VIL, storm motion, shear, mesocyclone rotational velocity etc.) of the tornadic storms were determined for the time preceding and just after tornado touchdown when a complete data set was available (20 to 30 cases). Work has also been completed on the collection and analysis of cloud-to-ground (CG) lightning data as part of this project to improve tornado warning criteria in the Northeast.
A computer program was developed to isolate single tornadic thunderstorms and monitor storm CG lightning characteristics (i.e. percent positive, frequency) in relation to tornado touchdown time. Tornado touchdown time is defined as the reported time the tornado touches the ground as noted in Storm Data. The reported touchdown times can be in error for many reasons. Archived WSR-88D data as available was used to verify tornado touchdown times. The single tornadic cells were isolated by using the lat/lon of the tornado touchdown as a reference. The results show a weak relative CG maximum (as high as 80 min-1) at tornado touchdown amidst decreasing frequency, and an increase in percent positive CG (3% to as high as 10%) centered on tornado touchdown. The results are relatively more robust in the stronger (F2/F3) tornadic storms.
The next part of this project isolated single tornadic thunderstorms and monitored the spatial distribution of CG flashes. All CG flashes were plotted relative to storm motion on a map, binned in two minute periods. These periods were then overlayed to see any apparent trends in CG activity during the life of the storm. The results showed a large case-to-case variability. CG lightning was concentrated to the left of the storm track, and there are vault and rear flank downdraft signatures seen in some of the cases. The vault is defined as the rain-free region under a strong updraft. The rear flank downdraft is defined as a region of dry air subsiding on the back side of and wrapping around a mesocyclone (strong rotation in a thunderstorm). A classic supercell signature (as seen in Moller et al (1994)) was seen in the 31 May 1998 Mechanicville, NY, tornadic thunderstorm.
In terms of operational value, the results showed no robust precursor signatures in the CG lightning patterns. Further work will include writing a short paper summarizing the results of this work.
Project 2: Developing severe thunderstorm warning criteria for pulse thunderstorms in the northeastern United States.
Storm Data, and NWS severe weather warnings from 1994 to 1998 were reviewed to identify pulse thunderstorm days. Criteria for pulse days were established. Pulse days were those days with severe weather events, scattered and non-sequential in nature, with any one event affecting no more than 2 contiguous counties. Events were short lived with a cell life cycle of an hour or less, with severe weather during a 30 minute or less period. Events used in the derecho (Project 3) and tornado (Project 1) projects, squall lines, supercells, storms that produced mesocyclones or tornadoes, and cases where convection was organized along fronts were eliminated. Applying this criteria resulted in 235 event days with over 800 storms. The data set was further reduced by eliminating days without WSR-88D archive II data, and was further checked against the OSF radar database to eliminate those events that did not satisfy the criteria. This reduced the number of days available for study to 94, with many of these days producing multiple storms. Using a random selection process, archive II data was ordered. The data is being analyzed using Watads 10.2 software.
Work continues to develop criteria that will allow the warning meteorologist to distinguish between pulse thunderstorms that may become severe and those that will not. Due to the short lived nature of these events, many parameters were examined with an eye toward maximizing lead time. The data analysis has just about been completed. Analysis of storm cross sections indicates some potentially useful differences between storms that become severe and those that do not. Severe pulse storms generally develop relatively higher reflectivities, at higher altitudes, earlier in the life of the storm when compared to non-severe storms. The results of this project will be formalized in a manuscript in the next few months.
Project 3: Comparing the synoptic and mesoscale environments that produce significant tornado outbreaks and derechos.
(a) Tornado cases:
Three upper-level flow patterns conducive to severe weather emerged from the 11 tornado outbreak cases studied. Two southwesterly flow regimes, one with a prominent anticyclone aloft in the south-central US, the other absent of the anticyclone, and a northwesterly flow regime were observed. The strongest upper-level jet occurred in the conjunction with a strong anticyclone over the south-central US. The tornadoes spawned by the southwesterly flow cases with the prominent anticyclone usually occurred on the cyclonic shear side of the upper-level jet embedded in a deep trough. Both composites exhibited an anomalous cutoff low pressure system at the surface and strong southwesterly flow (~ 30 m s-1) at 850 hPa.
The upper-level trough in the northwesterly flow cases is located farther poleward than in the two southwesterly cases, as much of the eastern US is dominated by an anticyclone centered over the Ohio Valley. As expected, the northwesterly flow cases indicate that the tornado outbreak occurs on the anticyclonic shear side of a jet streak. Unlike in the southwesterly flow cases, no closed cyclonic circulation is seen in the 850 hPa height field. However, by storm onset, there is a deep 850 hPa trough located over the tornado outbreak region, with a slightly weaker lower-level jet at 20 ms-1.
The last case did not fit into any of the previous flow regimes. This event is an outlier in that it did not fit the upper-air environmental conditions of the previous cases. The 500 hPa field bears some resemblance to the southwesterly flow cases with the high amplitude ridge, except that the ridge is centered farther east over the lower Mississippi River Valley. The 200 hPa jet streak is well removed from the tornado outbreak region. A cutoff low evident in the 850 height field develops 48 h before the tornadic event and moves Northeast directly over the region with a jet maximum of ~25 m s-1. The mean sea-level pressure field for this case shows that the deepening of the low pressure system does not occur until the time of onset. Leading up to the time of the event, there is weak southwesterly flow that changes to predominantly southerly flow as the base of the low pressure system moves directly over the outbreak region.
Overall, southwesterly flow regimes are more frequent than other upper-level flow patterns for producing severe tornadic outbreaks in the Northeast. The southwesterly flow cases have stronger upper-level winds and have favorable profiles for warm-air advection. Typically, cooler air from the Atlantic Ocean is transported by the southerly flow off the New England coast and converges with the southwesterly warm, moist air that is being advected into the Northeast. Northwesterly flow cases are characterized by a prominent high pressure system just to the south of the event region, that allowing warm, moist air to be advected into the Northeast.
(b) Derecho cases:
The 10 derecho events were separated into two composites based on the path of the hourly wind damage reports and the radar imagery. These composites, northwest and west, each contain five derecho events. The northwest derecho cases display strong west to northwesterly flow over the Great Lakes and into the Northeast. The derechos occur on the anticyclonic shear side of the jet with a 40 m s-1 jet maximum. At 850 hPa, a confluence zone is observed in the Midwest prior to the onset of the derechos. Downstream of this confluence zone, a strong low-level jet is observed over the Northeast. This jet is responsible for advecting warmer air from the southwest into the Northeast. At the surface, strong southwesterly flow develops prior to the event. The southwesterly flow intensifies ahead of an approaching frontal trough until event onset.
The west derecho cases exhibit westerly zonal flow over much of the northern half of the US at 500 hPa, with a trough located much farther north than in the northwest derecho grouping. Derecho onset occurs beneath the equatorward-entrance region of the 200 hPa jet. The 200 hPa jets in both of the derecho cases are about 20 m s-1 weaker than in the tornado cases. The 850 hPa heights are very similar to the northwest composite in that there is confluence in the 850 hPa heights from the Great Lakes eastward to New England at the onset time of the events. There is strong westerly flow over the Northeast by the time of the events with warm air being advected into the area. The southwesterly flow at the surface, while not as prominent as in the northwest flow composite, is still relatively strong over the Northeast by onset time.
Major differences between the two derecho groupings are noted with stronger 850 hPa jets and surface winds seen for the northwest derecho cases. Likewise, the majority of the northwest derechos covered a larger area and persisted longer than the west derecho events. The position of the upper-level jet is also different for each case. The northwest derecho cases are positioned under the anticyclonic shear side of the jet, while the west derecho cases are positioned between the core and the equatorward entrance region of the 200 hPa jet. Both cases exhibit an east-west oriented axis of high qe air embedded within strong westerly flow at 850 hPa. This qe ridge over the Northeast coincides with the strong low-level jet and surface winds at the time of onset for both cases as warmer air is advected into the area beneath an equatorward jet-entrance region at 200 hPa. Both of the derecho composites were similar in that the storms occurred near the leading edge of 700 hPa ascent and high relative humidity. In contrast, the tornado cases tended to occur more directly beneath the area of 700 hPa ascent and high relative humidity.
Project 4: Identifying the relationship between flow regimes and specific areas of tornado genesis and significant damaging wind storms in eastern New York and western New England.
There have been numerous studies of the climatology of severe weather and synoptic set ups associated with severe weather in certain areas of the United States. The Great Plains, in particular, has been an area where many advances have been made in understanding how severe thunderstorms develop. However, there has not been a lot of research done on the development of severe weather, particularly tornadoes, in the northeast United States. This study attempted to fill the gap somewhat. Tornadoes can and do occur in the Northeast, although not with the frequency that they occur in the Plains. It is important to understand how the local terrain can enhance or inhibit the development of severe weather and/or tornadoes in this area. There are some differences in major synoptic features that tend to be associated with a tornado event in the Northeast as opposed to a tornado event in the Great Plains. It is important for local forecasters to understand these differences so that they can properly anticipate a potential severe weather or tornado day in the Northeast.
In an effort to assess the possible role of terrain on Northeast US severe weather occurrences, all reported severe weather occurrences in Storm Data for the period 1955-1998 were tabulated. The data were stratified based on the 700 hPa flow direction as derived from Albany, NY, radiosonde observations. The 700 hPa level was chosen because it is high enough to be representative of the large-scale flow and low enough to feel the influence of the underlying terrain. The study area, consisting of upstate eastern New York and western New England, was divided into 0.5 degree latitude/longitude boxes. For each box the average number of reported events per severe weather day with northwest flow was compared to the average number of reported events with southwest flow. After correction for population bias, the results indicate that the areas south (north) of the Mohawk River Valley are more likely to experience severe weather when the 700 hPa flow is northwesterly (southwesterly). The second part of this study focused primarily on the large-scale synoptic setups that area associated with tornado events in the region. A goal of this part of the project was to identify any differences in synoptic setups which are favorable for a Northeast tornado event as opposed to tornado events in other parts of the country. This was done by creating schematics based upon composites of Northeast tornado events, and comparing them to previous work which discussed likely synoptic setups for severe weather in other areas of the country. A major finding of this part of the study was that there are indeed differences in the upper- and lower-level flow patterns associated with tornadoes in the Northeast when compared to previous studies. There tends to be less directional shear between 850 hPa and 200 hPa for a Northeast tornado event than for a Great Plains tornado event. The directional shear tends to be concentrated from the surface to 850 hPa, and there tends to be more speed shear between 850 hPa and 200 hPa. More specifically, the 850 hPa low-level jet is not oriented north-south as it would be in a typical Great Plains tornado event. Closer to the surface, the local topography of upstate New York and western New England often helps to enhance the directional wind shear in the lowest levels of the atmosphere.
It was concluded that there are three likely setups for a tornado event in the Northeast: warm air advection into a 500 hPa shortwave ridge, large amplitude cold trough at 500 hPa, and northwest flow at 500 hPa. In both the warm air advection and the cold trough case, the area in which the tornado occurs is under southwesterly flow at 500 hPa. In the warm air advection case, the upper level jet is downstream, and the region is positioned under the equatorward entrance region of the jet, while in the cold trough case, the jet is upstream and the area in which the tornado occurs is under the poleward exit region of the upper-level jet. In all the cases except for northwesterly flow events, there is ample low-level moisture over the region. In all cases, there is a maximum of upward motion over the area in which the tornado occurred, as well as cyclonic vorticity advection at 500 hPa over the area. These results agree quite well with the findings of Johns and Dorr (1996) in their study of strong tornado episodes in the northeast United States.
The case studies of each of these
three types of events showed that while the composites are useful for giving
the forecaster a general idea of a favorable synoptic setup for tornadoes, the
forecaster must be aware that a tornado day can look significantly different
from the composite or average. For instance, the tornado event of 3 July 1996,
while a cold trough tornado event, exhibited many differences from the cold
trough composite. Most obviously, the tornado was an early morning event, while
most of the tornadoes in the composite occurred during the late afternoon hours.
In general, the 500 hPa pattern for individual tornado events tends to look
similar to the composite, although the amplitude and position of the trough
can differ slightly. In all event types, the position of the 200 hPa jet can
deviate significantly from the average or composite jet. The position of the
surface features may look different from the composite, but in general the local
features in the region of the event look similar (e.g., cold front approaching
from west, warm front just north of area, etc.).
In the second part of the project, future work would include expanding the study to include a longer time period of tornado events. With a longer time period and more tornado events, the composites could be stratified by number of tornadoes to see if the strongest tornado events tend to be associated with one particular type of synoptic setup.
2. Summary of University/ NWS/AWS/Navy Exchanges:
3. Presentations and Publications:
(a) The following Master's Theses have been completed by UA students based on
research with this COMET project.
Honikman, S. F., 2001: Forecasting synoptic and mesoscale environments for tornadoes and derechos in the northeast United States. Master's Thesis completed at the University at Albany/SUNY. (Available at the University at Albany Library)
Wasula, A. C., 2001: Northeast severe weather distribution as a function of flow regime. Master's Thesis completed at the University at Albany/SUNY. (Available at the University at Albany Library)
(b) The following presentations
were made at the first Northeast Regional Operational Workshop in Albany, NY
in September 1999:
LaPenta, K. D., H. W. Johnson, G. J. Maglaras, J. S. Quinlan, and T. J. Galarneau, 1999: Improving tornado warnings in the northeast United States. Preprints, Northeast Regional Operational Workshop, Albany, NY, 48.
LaPenta, K. D., J. S. Quinlan, and A. C. Cacciola, 1999. Terrain and low-level boundary influences on tornado genesis during three major northeastern tornado outbreaks. Preprints, Northeast Regional Operational Workshop, Albany, NY, 49.
(c) The following presentation was
made at the National Weather Service Eastern Region Warning and Coordination
Meteorologist Meeting in Albany, NY in November 1999:
LaPenta, K. D., H. W. Johnson, G. J. Maglaras, J. S. Quinlan, and T. J. Galarneau, 1999: Improving tornado warnings in the northeast United States. National Weather Service Eastern Region Warning and Coordination Meteorologist Meeting, Albany, NY.
(d) The following papers were presented
at the 20th Conference on Severe Local Storms in Orlando, FL (September 2000):
Galarneau, T. J., S. F. Honikman, A. C. Cacciola, L. F. Bosart, K. D. LaPenta, J. S. Quinlan, and G. Wiley, 2000. Lightning in tornadic thunderstorms over the northeastern United States. Preprints, 20th Conference on Severe Local Storms, Orlando, FL, 108-109.
Cacciola, A. C., L. F. Bosart, S. F. Honikman, T. J. Galarneau, K. D. LaPenta, and J. S. Quinlan, 2000: Northeast severe weather distribution as a function of flow regime. Preprints, 20th Conference on Severe Local Storms, Orlando, FL, 453-456.
Honikman, S. F., A. C. Cacciola, T. J. Galarneau, L. F. Bosart, and K. D. LaPenta, 2000: Forecasting synoptic and mesoscale environments for tornadoes and derechos in the northeastern United States. Preprints, 20th Conference on Severe Local Storms, Orlando, FL, 509-512.
LaPenta, K. D., G. J. Maglaras, J. S. Quinlan, H. W. Johnson, L. F. Bosart, and T. J. Galarneau, 2000: Radar observations of northeastern United States Tornadoes. Preprints, 20th Conference on Severe Local Storms, Orlando, FL, 356-359.
(e) The following presentations were made at the 2nd Northeast Regional Operational Workshop at Albany, NY (November 2000):
Cerniglia, C. S., and W. R. Snyder, 2000: Development of severe thunderstorm warning criteria for pulse severe thunderstorms in the northeastern United States. Second Northeast Regional Operational Workshop, Albany, NY.
Galarneau, T. J., S. F. Honikman, A. C. Cacciola, L. F. Bosart, K. D. LaPenta, J. S. Quinlan, and G. Wiley, 2000: Lightning in tornadic thunderstorms over the northeastern United States. Second Northeast Regional Operational Workshop, Albany, NY.
Cacciola, A. C., L. F. Bosart, S. F. Honikman, T. J. Galarneau, K. D. LaPenta, and J. S. Quinlan, 2000: Northeast severe weather distribution as a function of flow regime. Second Northeast Regional Operational Workshop, Albany, NY.
Honikman, S. F., A. C. Cacciola, T. J. Galarneau, L. F. Bosart, and K. D. LaPenta, 2000: Forecasting synoptic and mesoscale environments for tornadoes and derechos in the northeastern United States. Second Northeast Regional Operational Workshop, Albany, NY.
LaPenta, K. D., G. J. Maglaras, J. S. Quinlan, H. W. Johnson, L. F. Bosart, and T. J. Galarneau, 2000. Radar observations of northeastern United States Tornadoes. Second Northeast Regional Operational Workshop, Albany, NY.
(f) The following presentation was made at the Northeast Storms Conference at Saratoga Springs, NY in March 2001:
Wasula, A. C., 2001: Northeast severe weather distribution as a function of flow regime. Northeast Storm Conference, Lyndon State University, Saratoga Springs, NY.
(g) The following presentations were made at the Weather at the Summit Conference at Sugarbush, VT in March 2001:
Wasula, A. C., 2001: Northeast severe weather distribution as a function of flow regime Weather at the Summit Conference, Sugarbush, VT.
Honikman, S. F., 2001: Forecasting synoptic and mesoscale environments for tornadoes and derechos in the northeastern United States. Weather at the Summit Conference, Sugarbush, VT.
4. Summary of Benefits and Problems Encountered
4.2 NWS Researcher
Research on severe thunderstorms and tornadoes began at the NWS Albany office in 1989. This COMET project represents another step in the ongoing effort to increase our ability to forecast and warn for severe weather. The partnership with the UA has provided expertise and resources necessary to continue toward our goal of providing the public with the best forecast and warning service possible. Specific research results and forecast techniques developed are presented in Section 1 of this report.
In performing this research, NWS participants gained valuable experience. Forecasters have spent a great deal of time examining large quantities of WSR-88D radar data. This work in itself has provided a valuable training exercise for the forecasters involved and will increase the staff's ability to detect and warn for severe weather.
During the past year the availability of NWS personnel has been limited to some extent by competing priorities (very active weather, implementation of AWIPS Interactive Computer Worded Forecast System, hosting NROW etc.). The Management of NWS Albany has done a good job of securing extra work time for several NWS researchers.
The availability of WSR-88D archive data was a problem. A significant amount of archive IV data for the tornado cases in Project I were not available. Still, more than 80 tornadoes were available for this study. Some of the data received was incomplete or is made up of more limited archive III data.