Downtown Miami F1 Tornado Report



On May 12, 1997 a F1 tornado moved through the downtown Miami around 2:00 P.M. EDT. Even though the tornado produced significant damage, it will be remembered for the photographs and videos taken as it moved through the skyscrapers of downtown Miami. The images made newscasts and headlines around the world. Using the latest computer and radar technology the tornado event was well forecast at least 24 hours in advance by the National Weather Service Forecast Office in Miami.


A significant F1 tornado, with wind speeds estimated at 100 to 110 mph, produced an estimated $526,000 in damage as it moved through the downtown area of Miami around 1800 UTC (2:00 P.M. EDT) on May 12, 1997. The tornado developed just southwest of the city and first touched down in the Silver Bluff Estates area at approximately 1753 UTC (1:53 P.M.). The tornado moved east-northeast at 20 to 25 mph and cut a 30 to 150-yard wide path, 8 miles in length, and was on the ground for about 15 minutes (Figure 1) (Lushine 1997). After the initial touchdown in Silver Bluff Estates area, the tornado crossed interstate I-95 and then moved through downtown Miami before entering Biscayne Bay near Bicentennial Park and the MacArthur Causeway. The tornado then continued northeast and crossed both the MacArthur and Venetian Causeways and moved over Biscayne Island. The visible funnel lifted from the water as it crossed Biscayne Bay, but touched down again briefly on Miami Beach near Collins Avenue and Arthur Godfrey Road. Since the tornado moved through the downtown area of a major metropolitan city, thousands of people witnessed the storm. Several photographs and videotapes of the tornado were taken by amateur photographers (Figures 2 and 3). The tornado was even captured by the "tower cam" of a local television station (WPLG-ABC).

The tornado produced roof damage to an apartment complex and some houses in the Little Havana area. In the downtown area, windows were blown out of several buildings including nearly every south facing window on the first three floors of The Citadel Building on NW Fourth Street. Several cars were damaged by flying debris in the WTVJ-NBC parking lot. Other cars were overturned or blown several yards when the tornado moved through the parking lot of a Bell South office building. A Metro Mover car on Miami's public transportation elevated rail system was derailed as the tornado neared the Government Center in downtown Miami. About a dozen minor injuries were attributed to the tornado with the majority of them being cuts caused by flying glass and debris.

The downtown Miami tornado was an unusual event for South Florida because it was synoptically driven and the conditions for supercell thunderstorm development were well forecast by the medium range computer models, at least 24 hours in advance. Most South Florida tornadoes are F0 tornadoes which produce little damage and generally last only a couple of minutes. Usually, they are not as severe or destructive as those in the Midwest (Gerrish 1967). The purpose of this paper is to examine model data and WSR-88D products which were useful in allowing the Storm Prediction Center (SPC) and the National Weather Service Forecast Office (NWSFO) in Miami to issue accurate forecasts and timely Tornado Watches and Warnings during the period leading up to this significant weather event.


The 1200 UTC upper air observation from Miami indicated a moist and unstable atmosphere in place with an un-modified CAPE value of 1418 J/Kg, a lifted index of -4 degrees C, and a precipitable water of 1.75 inches (Figure 4). The most important factor evident on the sounding was the directional shear in the lowest ten thousand feet of the atmosphere.

The potential for severe weather was indicated by the models beginning on the previous day's (May 11) 1200 UTC cycle. The Lead Forecaster working the day shift on May 11, 1997 concluded in the State Forecast Discussion that wording for severe weather would be included in the Zone Forecast Product for north and central Florida but that South Florida would "not be spared." An old frontal boundary which was difficult to locate in surface observations was located across the South Florida Peninsula. The old frontal boundary was forecast to move northward into Central Florida on May 12, 1997. A weak low pressure system was forecast to develop in the central Gulf of Mexico and move northeast across the Florida Big Bend on May 12, 1997.

The old frontal boundary was depicted best on the ETA 300 K isentropic surface pressure, wind, and relative humidity progs. The isentropic analysis (Figures 5 and 6) indicated strong isentropic lift across the entire Florida Peninsula with an indication of the old frontal boundary in the relative humidity field with higher relative humidity values north of the front. Using early morning visible satellite images, it was noted in the State Forecast Discussion issued at 1300 UTC (9:00 A.M. EDT) that clearing was occurring across extreme South Florida and that heating would take place south of the old frontal boundary. North of the boundary it appeared that the threat of severe weather was somewhat diminished with the activity being in the form of rain with embedded showers and thunderstorms.

The ETA model from 0000 UTC on May 12, 1997 indicated a 90 kt jet streak would move across the South Florida Peninsula between 1800 UTC May 12, 1997 and 0000 UTC May 13, 1997. It was noted by the 1200 UTC Miami upper air observation that the model may have been slightly underestimating the strength of the jet streak, as 250mb winds were already observed at 81 knots.

The model indicated during the afternoon of May 12, 1997 the southern portion of the Florida Peninsula would be in the favorable left front quadrant of the jet streak with good 250mb-850mb Qn vector convergence (Figures 7 and 8). The Qn vector convergence was a significant indicator in the amount of vertical lift which would occur. Since the Qn vectors across South Florida had a large component across the isotherms from warm to cold air, the thermal wind balance in that layer was altered. The atmosphere response was to counter the effect by increasing the thermal gradient. The QG theory can explain this as Barnes and Colman (1993) found that an induced ageostrophic circulation will attempt to tilt the isentropes toward the vertical by lifting the cold air and subsiding the warm air causing a thermally indirect circulation. Therefore, the QG forcing would contribute to upward motion across the area.

One very important feature was the model forecast indicating that the speed shear would dramatically increase across the Florida Peninsula during the early afternoon of May 12, 1997 (Figures 9 and 10). The 1200 UTC Miami sounding indicated that the wind near the surface was from the southeast and turned quickly to the south and southwest between five and ten thousand feet above the surface. At that time the speed shear was not as impressive as the directional shear as speeds ranged between 15 and 20 knots from about two thousand to ten thousand feet. However, as the day progressed the low level winds strengthened and the speed shear increased over the Florida Peninsula. Using the VAD Wind Profile (Figure 11) from the WSR-88D at the time of the tornado, the wind speeds had increased.

The increased shear forecast by the ETA model was evident in the model's forecast of a dramatic increase in the surface to 3 kilometer helicity values over the Florida Peninsula between 1200 UTC and 1800 UTC. The observed 0-3 km helicity value on the 1200 UTC Miami sounding was 131 m2s-2. The ETA forecast for 1800 UTC indicated helicity values between 200 and 300 m2s-2 over much of the central and south central Florida Peninsula, with values in the Miami area between 150 and 200 m2s-2 forecast for 1800 UTC (Figures 12 and 13). At 1630 UTC (12:30 P.M. EDT), it was noted during the Miami NWSFO daily map briefing that surface wind observations farther inland had veered to the south-southwest, but along the immediate coast and offshore buoys the wind remained out of the south-southeast creating stronger directional shear. These observations enabled forecasters to correctly conclude at the briefing that any severe weather would likely occur near the coast. Modifying the 1200 UTC Miami sounding using the Miami Beach wind observation and the VAD Wind Profile from the Miami WSR-88D the 0 to 3 km storm relative helicity increased to 205 m2s-2. Davies-Jones et al. (1990) suggested that a rough range of helicity values needed to produce a weak tornado is 150 to 299 m2s-2. Even though the ETA model indicated that the strongest helicity values would occur over central Florida, the fact that the model accurately predicted the significant increase in helicity was enough to realize the potential for supercell development.


At 1529 UTC (11:29 A.M. EDT) the Storm Prediction Center (SPC) issued a Tornado Watch for all of mainland South Florida effective until 2100 UTC (5:00 P.M. EDT). In the bulletin alerting South Florida of the Tornado Watch the SPC predicted that, "storms will continue to increase across South Florida in a very moist and unstable atmosphere. The potential for supercell tornadoes will increase during the early afternoon as the shear profile strengthens."

A Tornado Warning was issued by the National Weather Service Forecast Office (NWSFO) in Miami at 1755 UTC (1:55 P.M. EDT). A family member of a NWS employee reported seeing a "large" funnel cloud shortly before 1755 UTC. The warning was based on both the funnel cloud report and rotation detected by the Miami WSR-88D weather radar. Meteorologists at the Hurricane Research Division (HRD), located on Key Biscayne just southeast of downtown Miami, also reported a "large" funnel cloud over downtown Miami shortly after the warning was issued. Around 1758 UTC (1:58 P.M. EDT) a local television station phoned the WSFO to report that the funnel cloud was on the ground and they could see the tornado from their "tower cam". A Severe Weather Statement was issued using the WSR-88D and the live pictures available on television. This aided in the forecasters' ability to understand the magnitude of the storm and to warn residents that the tornado would move across Biscayne Bay and through Miami Beach.


The thunderstorm which produced the tornado developed in Southern Dade County just southwest of the radar site (in the Country Walk area) around 1645 UTC. The thunderstorm moved northeast near 20 mph and crossed Coral Gables and South Miami around 1730 UTC and approached the southwest corner of the City of Miami at 1743 UTC. The base reflectivity products at the 0.5 and 1.5 degree elevation angles indicated the developing tornado and eventually detected a rather distinct hook echo over Biscayne Bay. A four-panel display of the 1748 UTC base reflectivity products indicated the presence of a developing inflow notch in the lowest elevation angle along with higher reflectivity cores elevated aloft. On the 1753 UTC volume scan, about the time of the initial tornado touch down, a well-defined hook echo was evident on the 1.5 degree elevation angle. The WSR-88D continued to indicate a rather distinct hook echo on both the 0.5 and 1.5 degree elevation angles (Figures 14 and 15) as the tornado moved across downtown Miami and into Biscayne Bay.

The WSR-88D did not label the mesocyclone until the 1753 UTC volume scan which was the time of initial tornado touchdown. The radar continued to label a mesocyclone at 1758 UTC but then did not label a mesocyclone again until 1808 UTC about the time the tornado touched down briefly on Miami Beach. The 1.5 degree and especially the 2.4 degree elevation angle SRM products did detect significate rotation and shear in the storm. The height of the beam at both 1.5 degree and 2.4 degree elevation is approximately 2800 feet and 4400 feet respectively. Beginning with the 1748 UTC volume scan the 1.5 degree elevation angle SRM indicated rotational velocities (Vr) between 26 and 31 knots at a range of 14 to 19 nm during the life cycle of the tornado. Shear values remained between .010/s and .014/s during the time just prior to and while the tornado was on the ground. In this case the rotational velocities produced a minimal mesocyclone on the Mesocyclone Recognition Criteria developed by the Operation Support Facility (OSF) (Andra 1994).

The 2.4 degree elevation angle SRM indicated a moderate mesocyclone and higher shear values during the life cycle of the tornado. At 1748 UTC an operator defined minimal mesocyclone with a shear value of .010/s was evident. Between 1753 UTC and 1808 UTC a moderate mesocyclone was indicated by the WSR-88D. The rotational velocities continued to increase with a maximum of greater than 45 knots at 1758 UTC, however the maximum inbound and outbound velocities were not gate to gate and a broad mesocyclone with a radius of 1.8 nm was indicated. The mesocyclone weakened slightly at 1803 UTC but at 1808 UTC an inbound-outbound, gate to gate velocity of 30 to 40 knots was detected, which yielded a rotational velocity of 35 knots with a very impressive shear value of .065/s. This correlated extremely well in both time and location to the second tornado touchdown on Miami Beach.

The radar volume products such as the Vertically Integrated Liquid (VIL) and the mid-Layer (24 to 33 thousand feet) Reflectivity Maximum (LRM) products indicated a rather typical south Florida thunderstorm until after tornado had been on the ground for about five minutes. VIL values between 35kg/m2 and 39kg/m2 were indicated prior to the tornado touchdown on both the 1743 and 1748 UTC volume scans. At the time of the initial tornado touchdown at 1753 UTC the VIL increased to 45 kg/m2 and as the tornado was producing damage in downtown Miami the VIL increased to a maximum of 49kg/m2 at 1758 UTC. Similarly the mid-layer LRM products indicated values between 46 and 49 dbz prior to the tornado touchdown on both the 1743 and 1748 volume scans. One volume scan later at 1753 UTC, approximately the time of initial touchdown, the LRM increased substantially to 63 dbz and then peaked at 1758 UTC at 66 dbz.


The F1 tornado in downtown Miami on May 12, 1997 was an interesting weather event in South Florida. Typically, tornadoes in South Florida are generally weak and short-lived. This tornado was a strong F1 on the Fujita Scale which lasted about 15 minutes and affected the downtown area of a major metropolitan city. Most severe weather and tornadoes in South Florida occur in pulse type severe thunderstorms along the interaction of sea breeze and other boundaries which can be difficult to forecast more than a few minutes in advance. Since this event was synoptically driven, the conditions which were indeed favorable for producing supercell thunderstorms were well forecast by medium range computer models, the Storm Prediction Center, and the National Weather Service Forecast Office in Miami.

In this case, the ETA model indicated sufficient forcing over the area produced by an upper level jet streak and Qn vector convergence. The model also indicated that speed shear and helicity values would increase over South Florida during the early afternoon. Timely Tornado Watches and Warnings were issued based on correct analysis of model data, spotter reports and rotation detected by the WSR-88D. Shortly after 2:00 P.M. the magnitude of the event was realized when local television stations began broadcasting damage reports and footage of the event. Since the F1 tornado on May 12, 1997 was well documented by pictures as it weaved through the skyscrapers of downtown Miami, it will be remembered by the millions of residents in South Florida for years to come. In addition, the spectacle of a tornado moving through a major metropolitan area in the middle of the day was shown in newscasts around the world.


Several people are recognized for their guidance and assistance with this study. Appreciation is extended to Jack Gross, Science and Operations Officer, and Paul Hebert, Meteorologist in Charge at the NWSFO in Miami, for their expertise, review, and input on this manuscript. Joel Rothfuss, lead forecaster at the NWSFO in Miami, is recognized for detecting the potential for severe weather in advance. His knowledge assured that adequate personnel would be available. Scott Carroll, a meteorologist intern, was extremely helpful in the dissemination of the Tornado Warning. James Lushine, Warning Coordination Meteorologist, completed a storm damage survey and preliminary storm report which were helpful for this study.


Andra, D., 1994: Operational Recognition of Mesocyclones: Criteria and Application. THE WSR- 88-D Operator's Guide to Mesocyclone Recognition and Diagnosis, Appendix C., 1-5.

Barnes, S.L. and B.R. Colman, 1993: Quasigeostropic Diagnosis of Cyclogenesis Associated with a Cutoff Extratropical Cyclone-The Christmas 1987 Storm. Mon. Wea. Rev., 121, 1613-1634.

Davies-Jones, R. and D. Burgess, 1990: Test of Helicity as a Tornado Forecast Parameter. Preprints- 16th Conf. On Severe Local Storms, Kananskis Park, Alta., Canada, Amer. Meteor. Soc., 588-592.

Gerrish, H.P., 1967: Tornadoes and waterspouts in South Florida Area. Proc. 6th Army Conf. Tropical Meteorology, Coral Gables, Fl., 8-9 June 1967, 62-76.

Lushine, J., 1997: Preliminary Report of Tornado in Downtown Miami, May 12, 1997, Public Information Statement, Miami, FL: National Weather Service Forecast Office. is the U.S. government's official web portal to all federal, state and local government web resources and services.