Advanced Forecast and Warning Criteria for Tornadoes and Severe Thunderstorms in the Northeastern United States

 
Table of Contents

 Introduction 
 

Project 1
Project 2 
Project 3 
Project 4 
  
 
References 


Appendix A: Acronyms 
 

 Introduction


Over the past two decades there has been significant improvement in the NWS' ability to warn the public of impending tornadoes and severe thunderstorms (Crowther and Halmstad 1992). The driving forces behind this improvement were advances in technology and in atmospheric understanding. The WSR-88D has been the primary technological tool responsible for improved severe weather warnings with advanced computer and communication systems providing faster and more accurate delivery of warning messages. The second key to improved severe weather warnings has been a better understanding of thunderstorm morphology. A series of research projects that began in 1990 (LaPenta and Maglaras 1993, Maglaras and LaPenta 1997) has produced a better understanding of the atmospheric processes that drive the development and evolution of severe storms in the northeastern United States. This research included two COMET partners projects that studied the 29 May 1995 tornado outbreak (#NA37WD0018-01) and the 15 July 1995 intense derecho (#NA67WD0097). Knowledge gained has been shared with the staffs of NWSFO ALY as well as other NWS Offices in the Northeast, and has already paid substantial dividends. These research projects were instrumental in the issuance of accurate and timely warnings during the tornado outbreaks of 3 July 1997 and 31 May 1998, and the derechos of 7 September 1998 and 26-27 September 1998. Advances in the forecast and warning processes have been impressive, but important areas remain to be investigated.


While almost all the tornadoes on 3 July 1997 and 31 May 1998 occurred while tornado warnings were in effect, better refined WSR-88D warning guidelines and the integration of other operational available data sets into the warning process, might have reduced the number of tornado warnings during which only straight line wind damage occurred. Modern data archival capabilities provide a unique opportunity to re-examine severe weather events, and to develop criteria for differentiating tornadic storms and non-tornadic severe storms in an environment conducive to tornadoes. In the first project, a case by case study of tornadic events in the Northeast will yield guidelines for issuing tornado warnings in this part of the country. While guidelines will likely be based primarily on WSR-88D output, high resolution satellite and lightning data will also be examined to determine if the integration of multiple real-time data sources will produce increased warning accuracy.


Severe thunderstorms occur across a wide spectrum of intensities varying from tornadic storms and large scale, long-lived convective systems producing widespread damage, to short-lived, small scale pulse storms that produce very localized damage and marginally severe hail. Although tornadoes, derechos and other organized convective systems account for a large percentage of severe storm related deaths, injuries, and damage; forecaster experience suggests pulse severe thunderstorms are often more difficult to warn for due to their short life cycle and isolated nature. NWS severe thunderstorm warning verification scores improved steadily during the 1980s and early 1990s (Crowther and Halmstad 1992). Warning criteria established for the WSR-74C and WSR-74S radars (Lemon 1980) contributed to this improvement. Many of these techniques are not directly transferable to the WSR-88D. In addition, other data sources such as satellite imagery and lightning data have not been fully incorporated into the warning decision making process. Forecast verification is usually higher for large, well-organized events than for pulse type events. Further improving warning verification scores may be contingent on being better able to identify and warn for these pulse storms. Severe thunderstorm warning criteria for pulse severe thunderstorms in the northeastern United States will be established in order to improve overall severe weather warning verification scores and to increase the warning lead time for these storms. Again, warning criteria will likely be based primarily on WSR-88D data, but satellite and lightning data will be examined to see if integration of multiple data sources will result improved warning accuracy.


The first two proposed projects deal with the severe thunderstorm and tornado warning process. Warnings are short lead time products usually issued less than an hour prior to an event. Two additional studies will focus on the severe weather forecast process, that is, improving longer range (12 to 24 hours) identification of severe weather types as well as temporal and spatial resolution of severe weather threats. The third project will focus on comparing tornado and derecho events in the Northeast in order to differentiate the atmospheric conditions that produce these devastating storms and to develop forecast guidelines for predicting them up to 24 hours in advance. The differences may be complex as tornadoes have been observed within derechos and large bow echoes have been observed during predominantly tornadic events. In a final project, researchers will attempt to examine the feasibility of better identifying locations especially prone to tornadoes and extreme wind events once forecasters have identified a general severe weather threat. Important physiographic features significantly affect atmospheric processes. It is hypothesized that under different atmospheric flow regimes, these physiographic influences play a role in focusing the most intense convective storms. While this study will focus on eastern New York and western New England, it has important implications for many parts of the country. If the hypothesized relationship is established, it will identify the operational benefits of similar studies at other locations throughout the country.

Hazardous weather is a part of life in the Northeast. In 1998, there were four presidential disaster declarations for New York State alone. It is the mission of the NWS to warn the public of impending weather hazards. Combining the resources of the NWS and UA has been shown to be an effective mechanism for researching forecast problems and for integrating new knowledge into the operational environment.

 

Project 1: Using WSR-88D, satellite and lightning data to establish guidelines for differentiating tornadic thunderstorms and non-tornadic severe thunderstorms in a tornadic environment.
 

Research during the past decade has greatly increased our knowledge of severe thunderstorms. This research included a COMET partners project that studied the 29 May 1995 eastern New York and western New England tornado outbreak. Knowledge gained through this research was instrumental in issuing accurate and timely warnings during the tornado outbreaks of 3 July 1997 and 31 May 1998. On 31 May 1998, 142 severe thunderstorm and tornado warnings were issued for New York State. Tornadoes, damaging winds, or large hail were reported in 117 of these warnings. Table 1 summarizes warning verification scores in New York State on 31 May 1998. The probability of detection (POD) equals the number of warned events divided by the total number of events. The POD varies from 0 to 1, with a value of 1 indicating that all severe weather occurred with warnings in effect. The false alarm ratio (FAR) is equal to the number of unverified warnings divided by the total number of warnings and indicates the tendency to overwarn. A FAR of 0 indicates all warnings occurred with severe weather, while an FAR of 1 indicates no warnings were verified. The critical success index (CSI) combines the POD and FAR into an index that evaluates overall warning performance with values ranging from 1 (perfect score) down to 0. It should be noted that a severe thunderstorm event (large hail or damaging wind) verifies a tornado warning, while a tornado verifies a severe thunderstorm warning. On 31 May 1998, NWS offices serving New York recorded an outstanding POD of 0.91, a FAR 0.18 and a CSI of 0.76 for the event. These statistics can be compared to overall severe local storm verification statistics for the Eastern Region of the NWS from 1 January 1996 - 31 August 1998 (Table 2). Given the potential tornadic environment, forecasters issued a significant number of tornado warnings (51). Tornadoes or straight line wind damage occurred with all 51 tornado warnings, although only 13 tornado warnings were verified by confirmed tornadoes.


A better understanding of WSR-88D signatures and a better utilization of other available data might have reduced the number of tornado warnings during which only straight line wind damage occurred. Modern data archival capabilities provide a unique opportunity to re-examine severe weather events, and to develop criteria for differentiating tornadic storms and non-tornadic severe storms in an environment conducive to tornadoes. A case by case study of tornadic events in the Northeast will yield guidelines for issuing tornado warnings in this part of the country. Preliminary work by the researchers indicates that the shear of the rotational velocity (V_r) may be a useful tool in tornado warning decisions, however the number of cases examined so far is small. During 1997 and 1998 several major tornado outbreaks have significantly increased the number of tornado events available for study. In order to further expand the data base, radar data will be examined from the radar sites listed in Table 3. In addition to V_r, other radar parameters including temporal trends in V_r, weak echo regions, bounded weak echo regions, reflectivity pendants, mesocylcones and alphanumeric data on storm structure will be studied. The deployment of the Advanced Weather Interactive Processing System (AWIPS) will allow greater integration of radar data with other available data. This study will examine high resolution multi-channel satellite data and cloud to ground lightning data for each case in order to ascertain whether integration of multiple data sources can increase warning lead time and accuracy (Seimon 1993). Satellite imagery may reveal information about storm structure (cloud top divergence, cloud top temperature structure, etc.) and mesoscale flow structure. In addition, satellite derived atmospheric temperature and moisture profiles, and wind estimates, will be examined, as they may reveal key information about the state of the atmosphere that will assist meteorologists in warning decisions. The utility of other derived satellite estimates such as precipitable water and Convective Available Potential Energy (CAPE) will also be evaluated. Previous studies relating lightning and lightning rates to severe local storms have produced conflicting results. While research by Maier and Krider (1982) as well as Kane (1991) found some relationship, other studies (including McGorman et al. 1989) showed little connection. Researchers will first identify tornado cases to be studied and then data for these events will be collected. Primarily archive IV radar data will be used since archive II data is not available for many cases. Archive II data will be used where available to supplement archive IV data and to evaluate the new WSR-88D Build 10 TVS (tornado vortex signature) algorithm. Subjective and statistical methods will be used to create guidelines and radar techniques for issuing severe thunderstorm and tornado warnings within a tornadic environment.

Table 1. New York State verification statistics for 31 May 1998. WAR indicates total
warnings, tornado warnings (TOR) plus severe thunderstorm warnings.

NWS office	WAR	WAR verified	WAR missed	TOR	TOR verified by tornadoes	TOR verified by severe storms	missed tornadoes
KALY	37	33	04	27	05	22	00
KBGM	47	36	11	15	07	08	01*
KBTV	08	01	07	00	00	00	00
KBUF	39	38	01	02	01	01	00
KOKX	11	09	02	07	00	07	00
TOTAL	142	117	25	51	13	38	01*

*severe thunderstorm warning in effect at time of tornado


Table 2. Eastern Region of the NWS severe local storms verification statistics
for 1996, 1997 and 1998 through 31
August (Statistics from NWS Office of
Meteorology).

Project 2: Developing severe thunderstorm warning criteria for pulse severe thunderstorms.


Although tornadoes, derechos and other organized convective systems account for a large percentage of severe storm related deaths, injuries, and damage; forecaster experience suggests pulse severe thunderstorms are often more difficult to warn for due to their short life cycle and isolated nature. Forecast verification is usually higher for large, well-organized convective events than for pulse type storms. Table 4 is a comparison of verification statistics for ALY between major and minor events during 1998. Four major severe weather events produced a combined CSI of .77 while the CSI for the remaining 21 minor events was .52. There was a much higher rate of false alarms during minor events (FAR = .41) than during the 4 major events (FAR=.21). Further improving overall warning verification scores may be contingent on being better able to identify and warn for pulse storms.


NWS severe thunderstorm warning verification scores improved steadily during the 1980s and early 1990s Crowther and Halmstad 1992). Warning criteria (Lemon 1980) established for the WSR-74C and WSR-74S radars contributed to this improvement. Many of these techniques are not directly transferable to the WSR-88D. The WSR-88D employs a fully automated scan strategy that provides complete volumetric sampling every five to six minutes. The older radars were manually controlled, but this did allow more frequent sampling of a particular cell of interest. In addition, the WSR-88D provides a set of new products which were not available on most of the older radars. Preliminary studies (Blaes et al. 1998 and Amburn and Wolf 1997) have begun to examine the use of some of these new parameters in severe storm identification.


Researchers will use data from a large number of pulse type events across the Northeast from 1994 to the present. WSR-88D archive level II and level IV data will be examined for identified cases in order to evaluate WSR-88D products (VIL, VIL density (Blaes, et al. 1998), CR, RCS, VCS, MESO, CELL TRENDS etc.) and their relationship to pulse severe thunderstorms. In addition, high resolution multi-channel satellite data and cloud to ground lightning data for each case will be examined in order to ascertain whether integration of multiple data sources can increase warning lead time and accuracy. Satellite imagery may reveal information about storm structure (cloud top divergence, cloud top temperature structure, etc.) and mesoscale flow structure. Satellite derived atmospheric temperature and moisture profiles, wind estimates, as well as derived satellite products such as precipitable water and CAPE will also be evaluated to determine weather they can provide information useful to the warning process. Guidelines will be created for issuing Severe Thunderstorm Warnings for pulse severe thunderstorms in the northeastern United States. These guidelines will then be tested in order to verify any improvements in overall severe weather warning verification scores and warning lead times. Validation tests will be conducted using both historical data and operationally during actual severe weather events. Other NWS offices in the northeastern United States will be recruited to assist in operational evaluations.


Table 4. A comparison of preliminary 1998 severe storm warning verification scores (from ALY) for large scale organized convective events (major) and for more isolated,
local (minor) events. Major events included the squall line of 29 May, the 31 May
tornado outbreak, the derechos of 6-7 and 26-27 September.

Project 3.: Comparing the synoptic and mesoscale environments that produce significant tornado outbreaks and derechos.
 


Johns and Hart (1993) have shown that differentiating between predominantly tornadic events and events that produce long-lived derechos is difficult. During the last decade a significant number of tornadoes and derechos have affected the northeastern United States. Tornado and derecho events will be compared in order to establish forecast guidelines for differentiating the atmospheric conditions that spawn these devastating weather events. Since 1989 there have been 4 tornado events in ALY's county warning area (CWA) that meet the criteria identified by Grazulis (1993) for a significant tornado event (Table 5). During the same period, at least 4 derechos (Table 5) have affected the ALY CWA (Bosart et al. 1996, 1998). In order to significantly increase the number of events to be investigated, the geographical area of study will be expanded to include areas from the Great Lakes and Ohio Valley eastward to New England and the mid Atlantic States. A climatological study of derecho-producing mesoscale convective systems (Bentley and Mote 1998) may be useful in expanding the number of cases. Data employed will include surface and upper air analyses, satellite data and gridded data sets from numerical models including model soundings. A large quantity of archived meteorological data is available on station since 1989.


The differences between the two types of events are complex (Johns and Hart 1993) as tornadoes have been observed within derechos and large bow echoes have been observed during predominantly tornadic events. These differences are apparent in the 31 May 1998 event. During the early morning hours of that day a derecho moved east from Michigan into the eastern Great Lakes. The derecho weakened as it entered the western Adirondacks around 1600 UTC (Fig. 1a). During the next several hours, thunderstorms, and some supercells, developed across the Northeast, with a number of tornadoes touching down. Fig. 1b shows supercells embedded within a large area of convection across central and eastern New York at 2007 UTC. Fifteen minutes later the supercell over southeastern Saratoga County produced an F3 tornado. Several hours later another line of thunderstorms moved across eastern New York with a significant bow echo along the line south of Albany at 2321 UTC (Fig. 1c). An F2 tornado developed along the leading edge of the bow echo.

While some success has been achieved with alerting the public to the threat of tornadic events 12-24 hours in advance, less emphasis has been placed on identifying potential derecho events. This study will better enable forecasters to identify the synoptic scale conditions (12 to 24 hours in advance) that produce derechos in the northeastern United States and to distinguish between potential tornadic and derecho events. Researchers will to focus on areas within a certain synoptic regime that are conducive to tornado and derecho development. For example, we will examine the role of confluent jet-entrance region dynamics on the periphery of a continental anticyclone. Ascent in the equatorward entrance region of this jet perhaps can be tied to deep warm air advection (south flow at low-levels, west to northwest flow aloft) with very high _e air masses (_e above 350 K). In addition, the role of physiographic features (Fig. 2) on derecho evolution will be analyzed. In particular, investigators will:

• examine the role of mountains in channeling high _e air poleward beneath a west to northwest flow aloft.
• examine how lee-side troughing east of the mountains beneath a west to northwest flow aloft acts to focus new storm development and its impact on the derecho environment.
• examine upslope effects (Adirondacks, Catskills, Berkshires and Green Mountains) on convective initiation, organization and focusing beneath a west to northwest flow in derecho environments.
• the role of thermal boundaries created by differential heating over the cool waters of the Great Lakes versus the heated land as a focus for convective initiation and evolution.

The Mohawk, and other, River Valleys may also contribute to channeling of low-level flow. On 31 May 1998 (as on 29 May 1995), tornadogenesis occurred as a supercell crossed the Hudson Valley. An F3 tornado touched down in southern Saratoga County. An organized line of thunderstorms developed west of the supercell and gradually caught up to it. Tornadogenesis occurred when the supercell moved into the more favorable shear and thermodynamic environment in the Hudson Valley, and as the outflow from the line of storms to the west reached the supercell, increasing low-level vorticity in its vicinity. In this case, the outflow from the line of thunderstorms moved down the Mohawk Valley. Fig. 3 shows the WSR-88D base reflectivity and base velocity at 2012 UTC on 31 May 1998. The supercell that spawned the Saratoga County F3 tornado was located at s, with a line of thunderstorms to the west (q). In Fig. 3b, m marks the mesocylcone circulation associated with the supercell and o indicates the location of the outflow ahead of the line of thunderstorms. The KENX radar is located about 25 n mi south-southwest of the supercell (s). Although storms crossing the Hudson Valley appear to encounter a more favorable environment for intensification and tornadogenesis, tornado development is not restricted to this area. If conditions are favorable, tornadogenesis will occur across the higher terrain east and west of the valley (10 July 1989, Seimon and Fitzjarrald 1994 ).

Albany is located near the confluence of the Hudson and Mohawk Rivers. Forecasters have noted some tendency for tornadoes and significant severe weather outbreaks to occur further to the south of Albany as the atmospheric flow (and storm motion) becomes more north of west. As the flow (and storm motion) becomes south of west, tornadogenesis and significant severe weather seems to be favored north and east of Albany. It is hypothesized that physiographic influences play a role in focusing the most intense convective storms based on the direction of the atmospheric flow. The actual direction of storm motion will be considered since it is related to the atmospheric flow. When organized convective systems are present, storm motion is determined by both the general atmospheric flow and generation of new convective elements within the system (propagation, Chappell 1986). As a result, convective systems may move in a direction somewhat different from the atmospheric flow.


This research project will:

• compare the distribution and timing of severe weather as a function of the direction and the strength of the prevailing synoptic scale flow and its orientation to principal terrain features.
• compare the distribution and timing of severe weather as a function of the direction of storm motion and its orientation to principal terrain features.
• test the hypotheses that with a southwest wind flow, tornadoes and significant severe weather outbreaks are favored to the north and east of Albany; and with a west to northwest flow, tornadoes and significant severe weather outbreaks are favored to the east and southeast of Albany.
• study the role of deep north-south river valleys such as the Hudson Valley, the west-northwest to east-southeast oriented Mohawk Valley, and the surrounding mountains on the structure and life cycle of the convection.
• test the hypothesis that rain-cooled outflow air that is channeled down the Mohawk Valley when storms move from the west-northwest might be effective at initiating new convection when it meets the channeled south flow moving up the Hudson Valley with possible impact on severe weather east of the valley.

If a relationship exists between atmospheric flow or storm motion and the distribution of tornadoes and significant severe storms, it would allow forecasters to identify locations especially prone to tornadoes and extreme wind events once forecasters have identified a general severe weather threat for eastern New York and western New England. While this study will focus on eastern New York and western New England, it has important implications for other parts of the country. If the hypothesized relationships exist, it will identify the operational benefits of similar studies at other locations.

 References


Amburn, S. and P. Wolf, 1997: VIL density as a hail indicator. Wea. Forecasting,
12, 473-478.


Bentley, M. L. and T. L. Mote, 1998: A climatology of derecho producing mesoscale
convective systems in the central and eastern United States, 1986-1995. Part I.
Temporal and Spatial Distribution. Bull. Amer. Meteor. Soc., 79, 2527-2540.


Blaes, J. L., C. S. Cerniglia and M. A. Caropolo, 1998: VIL density as an indicator of
hail across eastern New York and western New England. Eastern Region
Technical Attachment, No. 98-8. National Weather Service, NOAA, U. S.
Department of Commerce, pp.17.


Bosart, L. F., W. E. Bracken, A. Seimon, J. S. Quinlan, K. D. LaPenta and J. W. Cannon, 1996: The northwesterly flow extreme derecho event of 15 July 1995 across New York and New England. Oral presentation at the Twenty First National Weather Association Annual Meeting, 1-6 December 1996, Cocoa Beach, Florida.

_____, W. E. Bracken, A. Seimon, J. S. Cannon, K. D. LaPenta, and J. S. Quinlan, 1998: Large-scale conditions associated with the northwesterly flow intense derecho events of 14-15 July 1995 in the northeastern United States. Preprints, 18th Conf. On Severe Local Storms, Minneapolis, MN, 503-506.


Bracken, W. E., L. F. Bosart, A. Seimon, K. D. LaPenta, J. S. Quinlan and J. W.
Cannon, 1998: Supercells and tornadogenesis over complex terrain: the Great
Barrington (Massachusetts) Memorial Day (1995) tornado. Submitted for review
to Mon. Wea. Review, July 1998.


Chappell, C. F., 1986: Quasi-stationary convective events. Mesoscale Meteorology
and Forecasting, Ed. P. S. Ray, Amer. Meteor. Soc., 289-310.


Crowther, H. G. and J. T. Halmstad 1993: Severe local storm warning verification:
1992. NOAA Tech. Memo. NWS NSSFC-37, 30 pp.


Grazulis, T. P., 1993: Significant tornadoes 1680-1991. The Tornado Project of
Environmental Films, St. Johnsbury, VT, 1326 pp.


Johns, R. and J. A. Hart, 1993: Differentiating between types of severe thunderstorm
outbreaks: A preliminary investigation. Preprints, 17^th Conference on Severe
Local Storms, Saint Louis, MO, Amer. Meteor. Soc., 46-50.


_____ and W. D. Hirt, 1997: Derechos: widespread convectively induced
windstorms. Wea. Forecasting, 2, 32-49.


Kane, R. J., 1991: Correlating lightning to severe local storms in the Northeastern
United States. Wea. Forecasting, 6, 3-12.


LaPenta, K. D. 1995: Forecasting tornadic versus non-tornadic severe thunderstorms in New York State. Eastern Region Technical Attachment, NO. 95-4A, National
Weather Service, NOAA, U.S. Department of Commerce, 15 pp.


_____ and G. J. Maglaras, 1993: New York State Tornadoes. Eastern Region
Technical Attachment, NO. 93-12A, National Weather Service, NOAA, U.S.
Department of Commerce, 13 pp.


Lemon, L. R., 1980: Severe thunderstorm radar identification techniques and warning
criteria: A preliminary report. NOAA Tech. Memo. NWS NWSSFC-1 [NTIS Accession NO. PB273049], 60 pp.


Maglaras, G.J. and K. D. LaPenta, 1997: Development of a forecast equation to predict the severity of thunderstorm events in New York State. Nat. Wea. Digest, 21-3,
3-9.


MacGorman, D. R., D. W. Burgess, V. Mazur, W. D. Rust, W. L. Taylor and B. C.
Johnson, 1989: Lightning rates relative to tornadic storm evolution on 22 May
1981. J. Atmos. Sci., 46, 221-250.


Maier, M. W., and E. P. Krider, 1982: A comparative study of cloud-to-ground lightning
characteristics in Florida and Oklahoma thunderstorms. Preprints: Twelfth
Conference on Severe Local Storms, San Antonio, Amer. Meteor. Soc., 334-337.


Seimon, A., 1993: Anomalous cloud-to-ground lightning in an F5-tornado-producing
supercell thunderstorm on 28 August 1990. Bull. Amer. Meteor. Soc., 74, 189-204.


_____, and D. Fitzjarrald, 1994: Topographic influences on mesocylcone evolution
and storm structure in an extreme supercell thunderstorm over rough terrain. Preprints, 6^th Conference on Mesoscale Processes, Portland, OR, Amer. Meteor. Soc., pp 513-514.

Appendix A: Acronyms

ALY - Albany, NY, office of the National Weather Service
AWIPS - Advanced Weather Interactive Processing System
BWER - bounded weak echo region
CAPE - convective available potential energy
CELL TRENDS - WSR-88D individual cell trend algorithm output
CR - composite reflectivity
CSI - critical success index
CWA - county warning area
FAR - false alarm ratio
MESO - WSR-88D Mesocyclone identification algorithm
NWS - National Weather Service
NWSFO - National Weather Service Forecast Office
POD - probability of detection
RCS - reflectivity cross section
SOO - science and operations officer
TVS - tornado vortex signature
UA - The University at Albany/SUNY
VCS - velocity cross section
VIL - vertically integrated liquid water content
V_r - shear of the rotational velocity
WER - weak echo region