A Spectrum of Tropical Cyclone-Tornado Producing Cells: Anticipation, Identification, and Lead-time Warnings*

Scott M. Spratt NWSO Melbourne, Florida

NEXRAD/OSF Tropical Weather Workshop January 7-9, 1997 Norman, Oklahoma

Abstract

     A spectrum of tropical cyclone-tornadic cells which occurred across Florida during Gordon (94), Erin (95), Opal (95) and Josephine (96) will be examined. The presentation will use WSR-88D radar data and WATADS algorithm output to compare and contrast the horizontal diameter and vertical depth of detected circulation couplets and illustrate cell persistence and intensity through trend tables. Based upon the cases examined, a manual recognition and warning technique will be introduced and a new radar scan strategy (volume coverage pattern; VCP) will be proposed to increase detection efficiency of shallow, small-diameter mesocyclones.

    Rapid changes of the local wind field which allow the environment to become more conductive to tornado development must be assessed continually through examination of WSR-88D VWP data. Once the environment becomes favorable, relatively isolated, high reflectivity cells which display persistence should be deemed "suspicious." Values of shear across circulation couplets were found to correlate better with tornado development than values of rotational velocity. Share of .015 s-1 were consistent with times of tornado touchdown, while values of .010 s-1 preceded touchdown by 5 to 15 min and may represent a lead-time warning threshold. Prior to tornado times, the diameter of the low-level velocity couplets often decreased to 1 n mi or less, resulting in the increased shear. A new VCP which samples additional low-level elevations (in lieu of several upper slices) is currently being proposed to allow for increased resolution within the portion of the atmospheric column where tropical mesocyclones are most pronounced.


*Portions of this presentation were excerpted from A WSR-88D Assessment of Tropical Cyclone Outer Rainband Tornadoes (Spratt et al., published in Weather and Forecasting), Radar Detected Mesocyclones within Tropical Cyclone Gordon (1994) (Spratt and Sharp) and A Spectrum of Mesocyclone Circulations Observed within Tropical Cyclone Rainbands (Spratt and Hodanish). The later two papers were presented at the 21st NWA Annual meeting (Cocoa Beach, Florida, December 1996).  Also, see a companion paper written by Andy Nash at NWS Tampa Bay titled Examination of Tornadic Cells Within a Rainband Associated with Tropical Storm Josephine (1996).


Introduction

    The oral presentation associated with this topic showcased numerous WSR-88D and WATADS products which helped define a spectrum of tornadic cells within several recent tropical cyclone events (Fig. 1). The following pages address a suggested forecast process which, when used in real-time, should allow for tornado warnings in advance of many outer tropical cyclone rainband events.

    In order for meteorologists to provide timely weather warnings for TC outer rainband tornadoes, a thorough, yet prompt examination of WSR-88D data is essential. Given the small dimensions of most TC-tornado producing cells and the often subtleness of their reflectivity signatures as severe indicators, the decision to issue severe weather warnings can prove to be more arduous than during situations involving more traditional mesocyclones/supercells. Furthermore, since TC-outer rainband cells are less likely to produce either large hail or damaging straight-line winds, the effectiveness of issuing severe thunderstorm warnings in lieu of tornado warnings is questionable, even for marginal situations, and is discouraged. To promote an aggressive radar approach, operators must be able to recognize potential tornadic cells early in their life cycle while compensating for radar sampling limitations. The following sections will addresss these issues and present a systematic approach for identifying potentially tornadic cells within TC outer rainbands.

Compensating for WSR-88D sampling limitations

     Although traditional severe weather indicators were occasionally observed in reflectivity products during several of the Gordon and Allison tornado events (e.g. inflow notches, pendants, or hook-like echoes), most often the features were more subtle. Even when detected, they were usually at or below cloud base or within 1-km of the surface. Assuming standard atmospheric refraction of the 0.5 elevation (lowest) beam, detection of such features beyond 55-km of the radar site becomes unlikely. Fortunately, mesocyclonic rotation often extended to slightly higher heights. Cyclonic shear, or rotational velocities of weak mesocyclone strength (Figs. 2a and 2b ), were observed to a height of nearly 3.5 km within all the studied cases. This height equates to detection distances of approximately 160, 100, and 65-km for the 0.5o, 1.5o, and 2.4o elevation slices, respectively. At ranges where only the beam from one or two elevations intersect a cell below 3.5 km, the vertical continuity threshold for warning decisions should be relaxed since the average updraft itself is only 3.5 km (Weisman and McCaul 1995). A cell possessing weak cyclonic shear or rotation at low/mid-levels and high reflectivity relative to surrounding cells should be immediately categorized as suspect, especially if persistence is noted or a history of wind damage exists (plates A and B).

     One method of compensating for the WSR-88D's decreased skill in sampling small features at greater distances is to examine a mesocyclone strength nomogram based upon the detected circulations core diameter. For example, at a range of 160-km, the strength of a given mesocyclone with a rotational velocity of 15 m s-1 and diameter of 6.5 km would be classified as minimal, while the same feature with a contracted diameter of 1.85 km would be classified as strong, suggestive of a tornado warning.

     Beam broadening at increased ranges also increases the potential for sampled inbound/outbound velocities to be unrepresentative. In these circumstances, relatively high SW values (i.e. >7.5 m s-1) may add confidence to the determination of marginally severe cells (plates C and D).

     Due to the small, shallow nature of most TC-mesocyclones, it is not surprising that the current WSR-88D M algorithm often results in under-detection. It should be noted that reducing the threshold pattern vector (TPV) adaptable parameter within the M algorithm (to allow for the identification of smaller 2-D circulations) did not dramatically improve its performance during the Barefoot Bay mesocyclone/F2 tornado. Therefore, until a more robust mesocyclone algorithm is employed, manual mesocyclone recognition must be stressed.

     Additionally, the relatively long periodicity of WSR-88D data refresh (5-6 minutes) could be detrimental when attempting to assess parameter trends within cells capable of producing tornadoes of similar or lesser duration. One way to shorten the period for data refresh in the lower elevations would be to manually restart the volume scan once the beam reaches a sufficient elevation (e.g. 4.3o ). Although this procedure would preclude algorithm processing of volumetric derived products, a manual re-examination of the lower elevations may outweigh the usefulness of such products during fast-breaking events.

Operational recommendations

a) Environmental assessment

    As TC outer rainbands begin to affect a particular area, frequent assessment of the local environment becomes necessary to determine whether the profile of vertical shear is becoming more favorable for tornadoes (Fig. 3). The most useful tool for this purpose is the WSR-88D VWP product. Interactive workstations (i.e. SHARP) can also be utilized to assess specific changes to SRH. In addition, greater thermodynamic destabilization through the intrusion of dry air aloft or diabatic warming of the near-surface layer can further increase updraft speeds and the potential for tornadic mesocyclones. This is more likely to occur during post landfall "remnant" cases as was observed over the mid-Atlantic region in association with TC Beryl (Vescio et al. 1995) or with extra-tropical hybrid cases, rather than during pre-landfall events.

b) Rainband assessment

     Climatologically, low-level shear is maximized within the northeast quadrant of tropical cyclone circulations. Within this region, local enhancements to shear and vorticity occur along convergent boundaries. As a result, organized rainbands inherently possess a higher degree of concentrated vorticity than less organized adjacent regions. Additionally, the smaller radius of curvature of some rainbands may further act to locally concentrate convergence and vorticity. In order to exploit the available vorticity for mesocyclogenesis (and perhaps tornadogenesis), improved lift is needed to overcome low buoyancy. Such lift, and subsequent cell intensification may result from sustained updrafts produced via frictional convergence as rainbands move onshore within highly sheared environments, and from localized convergence along and near intersecting or merging boundaries. Boundary interactions are more likely to occur with rainbands which rotate outward far from the cyclone center or with TC remnants as opposed to rainbands located within the highly organized structure of the inner bands of a mature tropical cyclone. Both frictional convergence and intersecting boundaries also appear to have been factors during the Gordon and Allison tornadoes.

     Operationally, both advecting rainbands and the development of bands in-situ necessitate a greater degree of scrutiny than adjacent ill organized regions of stratiform rain with embedded convection. Observations indicated that small mesocyclonic circulations were generated more often within persistent band-like features than within transient, less persistent features. Organized bands should be closely evaluated for mesocyclones to develop in series, or in banded families, as was also seen with both Gordon (plates E and F) and Allison (plates G and H).

c.) Storm-scale assessment

     Although maximum reflectivity values associated with tropical tornadic cells may not necessarily be impressive, often they will be greater and more persistent relative to adjacent cells. Likewise, ET, VIL, and SW values may be greater and more persistent. Fundamental to the early recognition of suspect cells is the interrogation of the most intense cells within the rainband, regardless of their actual intensity values. A scan by scan examination of suspect cells should be performed by comparing the evolution of reflectivity and SRM velocity products for trends which favor intensification. Also, when assessing SRM products, the radar operator should be cognizant of the average motion vector which is being subtracted from the radial velocities since an erroneous motion can visually mask rotational signatures. Indications of persistent, organized shear or weak rotation within
SRM products should be taken seriously, even in the absence of traditional severe weather reflectivity indicators. Likewise, persistent cells which exhibit relatively high reflectivity (compared to surrounding echoes) and remain isolated should be monitored closely for early indications of circulation.

     Post analyses of the Gordon and Allison radar data indicated a close correlation between the time of tornado touchdown and the time of contraction of the low-level mesocyclone. Preliminary results based upon this limited study show that mesocyclones exhibiting shear values of .010 s-1 or greater should be considered prime candidates for tornadogenesis. As the shear approaches .015 s-1, the likelihood of tornado formation appears to increase significantly (plates I-L). Likewise, Grant and Prentice (1996) in a study of 16 non-tropical low-topped mesocyclones, found the magnitude of maximum shear doubled in a 24-min period prior to tornado time, reaching a peak of .015 s-1 at the time of touchdown. Given these findings, the prompt issuance of a tornado warning is strongly suggested at times when environmental conditions are favorable for tornadoes and small diameter mesocyclones with moderate to large shear are detected. Nonetheless, it is imperative that radar operators understand the relationship concerning the variability of both the core diameter and rotational velocity on the shear and how well each can be sampled at varying ranges from the radar. Understanding this relationship also places great importance on operator skill for properly discerning the mesocyclone core (i.e. perceiving the idealized Rankine-combined vortex) for diameter, rotational velocity, and shear assessments.

d.) Future requirements

     Improved forecasts and warnings of TC outer rainband tornadoes will likely occur as additional radar case studies are examined and model simulations evolve. Incorporated improvements at the local office level have already lead to successful tornado detections and warnings during the tornadic outbreak associated with Tropical Storm Josephine (1996) over central Florida. Base data archival from the WSR-88D during TC events should be given a high priority so that new and improved algorithms can be pursued and adaptable parameter optimization can continue. The use of radar-based trend tables must be expanded and exploited to help identify potentially severe cells from the multitude of cells upon the radar display.

     Finally, a new volume coverage pattern (VCP) should be developed to allow additional low-level scanning. For example, a VCP with added tilts between the current 0.5o, 1.5o, and 2.4o tilts would provide valuable vertical information for shallow phenomena. To compensate for the additional slices and keep the data refresh rate reasonable, several upper elevation tilts could be sacrificed. This proposed scan strategy would allow for a better assessment of phenomena at lower heights where TC-mesocyclones are most likely to be detected and would also likely improve algorithm detection efficiencies.

References

Weisman, M.L. and E.W. McCaul, Jr., 1995: Simulations of shallow supercells in landfalling hurricane environments. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 428-430.

Vescio, D.M., S.J. Weiss, and F.P. Ostby, 1995: Tornadoes associated with Tropical Storm Beryl. Preprints, 21st AMS Conf. on Hurricanes and Tropical Meteorology, Miami, Fl, Amer. Meteor. Soc., 469-471.

Grant, B. and R. Prentice, 1996: Mesocylone characteristics of mini supercell thunderstorms. Preprints, 15th Conf. on Weather Analysis and Forecasting, Norfolk, VA, Amer. Meteor. Soc., 362-365.