Close Window

Saint Louis Univ.: "Analysis and forecasting of tornadoes associated with tropical cyclones"

Final Report


The main objective has been to study the mean characteristics of mesocyclones producing tornadoes in landfalling tropical cyclones (TC). About twenty-five TC mesocyclones in Tropical Cyclones Frances (1998), Earl (1998), and Floyd (1999) that produced tornadoes and two in Floyd (1999) that did not produce tornadoes have been studied. These characteristics have been divided into kinematic and thermodynamic features.


An accomplishment has been to examine the different characteristics as revealed by Doppler Radar and thermodynamic data of the tornadic and non-tornadic mesocyclones spawned by TCs.

Based on our study the following conclusions are drawn: 1) Tornadic mesocyclones did not seem to exhibit characteristics in the Radar imagery distinct from those which are not tornado producing. It is possible that there are genuine differences but because of the current five-minute volume scan during which mesocyclones may be evolving the possible differences are perhaps diluted. It is also likely that there are some distinct characteristics but were not captured by the Radar because of the range. 2) Mesocyclones formed over the coastal waters were strong and long-lived compared to those formed over land, 3) Low-level thermal boundaries such as fronts or surface troughs play a crucial role in the intensification of mesocyclones. Some mesocyclones produced tornadoes only after crossing such boundaries.

A PowerPoint presentation entitled, Some diagnostic aspects of mesocyclones associated with tropical cyclones will be available online shortly.

Experience gained

Mesocyclones of the various tropical cyclones seem to belong to a wide variety. Some are small both in terms of horizontal and vertical extents. Some are short-lived while others are long-lived. A few of them are typified by large Convective Available Potential Energy (CAPE) while others by small amounts. All of them appear to be identified by large storm relative helicity (SRH). The helicity amounts vary slightly but non-uniformly depending on whether the WATADS Radar algorithm default motion of the mesocyclone or its observed motion is used.

As mentioned above Doppler Radar data as viewed from WATADS program alone is not sufficiently detailed as to discriminate mesocyclones which produced tornadoes from those which did not. A valuable possible aid is a fine resolution numerical model equipped with 4-D variational initialization and assimilation (4DVAR). NCAR's MM5 is obviously available for experimentation.

In future studies we will stratify mesocyclones by range from Radar to see if significant differences between tornadic and non-tornadic mesocyclones can be gleaned.

Accomplishments of the University Participant (U.P.)

Accomplishments of NWS participants

Summary of benefits

The U.P. and his students became aware of the apparently different mechanisms governing the TC tornado formation and the variance in size, reflectivity,
spectrum width, CAPE and SRH of the mesocyclones that produced tornadoes. The damage a tornado caused is not apparently related to the size and strength of the parent mesocyclone.

The U.P. became aware of the rich mesoscale observations in the vicinity of Melbourne which are available on a fine spatial and temporal scale. These should be fully exploited for further TC mesocyclone studies.

The U.P. became aware of the potentialities of the mesoscale observations such as soil moisture in the State of Oklahoma and their use in understanding tornadogenesis.

Benefits to the NWS partners

The WFO forecasters at Melbourne (Hagemeyer and Spratt) traditionally base their short-term warnings of TC tornadoes on the Radar signatures of mesocyclones. Because of safety considerations there is a tendency to call for tornado occurrences when a mesocyclone is detected. Because tornadoes can be produced in a variety of ways and not all mesocyclones produce tornadoes this practice results in "over warning" culminating in a host of false alarms. This generation of false alarms is a matter of grave concern to the WFO forecasters. Although the mean characteristics studied here do not show distinct characteristics between tornado producing and not producing the tendency fields of shear seem to show a promise.

The SPC forecasters (Schaefer and Edwards) found the MM5 simulations of the TC environment by SLU to be helpful in ascertaining the meso-beta scale kinematic and thermodynamic changes which tend to occur either in connection with or just prior to the onset of tornadogenesis in peripheral TC supercells. Watches and mesoscale discussions may key on these features which distinguish tornadic versus non-tornadic environments. Potentially, this understanding can make the difference between issuing an unverified watch for a landfalling TC which produces no tornadoes and saving such a watch for a more threatening TC tornado situation. Also because of the mesoscale detail of the MM5 indicated trends watches and mesoscale discussions can be confined in spatial extent to those areas which contain the highest probabilities for supercells to form which can produce tornadoes.

Presentations at National Meetings

G. V. Rao, J. Scheck and R. Edwards presented a talk entitled, Radar derived structures of tornadic mesocyclones from tropical cyclone Frances in Texas and Louisiana. Preprints. 25th AMS Conf. on Hurricanes and tropical meteorology, 617-618, 29 April-3 May 2002.

J. Scheck (SLU), R. Edwards and G. V. Rao presented a talk entitled, Radar examination of a bow echo in South Carolina within the remains of TC Danny (1997). Preprints, 21st AMS Conf. on Severe Storms, San Antonio, TX. 517-519, 12-16 August 2002.

3. G. V. Rao, J. Scheck, D. Gallagher and K. Santhanam presented at the NWA Annual meeting in Fort Worth, TX , a poster entitled, Observed WSR-88D Radar structures and MM5 simulated kinematic structures.


We had studied the Radar and other characteristics of mesocyclones throughout their
life time. For the purposes of stressing their characteristics at the time of tornado occurrence we narrowed down the sample to 3 observations: one, 5 minutes before the tornado occurrence, two, at the time of occurrence and three, 5 minutes after the occurrence. Our study deals with 12 incidences of tornadoes making the total population 36.

Based on this population and taking an average of 36 data pieces we draw the following conclusions.

  1. The low-level circulation at 0.5 degree elevation is situated at 570 meters above ground level (AGL).

  2. The circulation at 1.5 degree elevation is situated at 1700 meters AGL.

  3. The diameter of the mesocyclone at 570 m is 6.05 km.

  4. The diameter of the mesocyclone at 1700 m is 6.92 km.

  5. The horizontal shear at 570 m is 0.006 s-1.

  6. The horizontal shear at 1700 m is 0.004 s-1.

  7. The duration of the tornadic mesocyclone (sample 12) is 1 h 15 minutes.

  8. The duration of the non-tornadic mesocyclone (sample 2) is 40 minutes.

  9. Mesocyclones at the time of tornado occurrence are located in the right front quadrant of the TC at angle of 56.5 degrees at a distance of 354.5 km from the center.

  10. The average VIL (Vertically Integrated Liquid Water) content of a mesocyclone as estimated by WATADS is 20.36 g per meter sqd.

  11. The average maximum reflectivity is 42.6 dBZ.

  12. The average height of maximum reflectivity is 3.48 km.

Since the number of mesocyclones that did not produce tornadoes in North Carolina we studied was only two we did not summarize here. They are in the power point presentation which is enclosed.

Future plan: We wish to submit a condensed version of our findings to Monthly Weather Review towards Notes and Correspondence in June 2003. This Note will include the following NWS COMET participants: J. Schaefer, R. Edwards, B. Hagemeyer and S. Spratt.