Adapted from Preprints, 23rd AMS Conference on Hurricanes and Tropical Meteorology, Dallas, 11-15 Jan 1999


Roger Edwards

Storm Prediction Center

Norman, OK


The spatial and temporal distribution of tropical cyclone (TC) tornadoes near landfall is climatologically well-documented. Tornadoes have predominantly occurred within 24 hours following landfall, north through east-southeast of the center (Hill et al. 1966, Novlan and Gray 1974, Gentry 1983), or with a more disperse distribution weighted toward the right-front portion when using a cyclone motion-relative coordinate system (McCaul 1991). A few case studies have examined TCs yielding singular or outbreak tornado events well inland, or more than a day after landfall (e.g., Hoadley 1981, Vescio et. al. 1996), one of which -- Danny of 1985 (McCaul 1987) -- returned to sea after episodically spawning tornadoes throughout its passage over land.

Tornadoes associated with tropical cyclones that are moving back out to sea are uncommon but not unique. The climatology compiled by Weiss (1987) included some such TC tornadoes, but did not separately categorize them. No explicit tornado climatology has heretofore existed for such storms, defined here as "exiting" events. These may present an uncommon forecast challenge, as with Hurricane Danny of 1997. The remnants of Danny (1997) spawned eight tornadoes1 in the Carolinas shortly before moving offshore, including a strong (F2 damage), killer tornado near Gaston, SC (NCDC 1997). This was a difficult forecast problem, for Danny had weakened after making landfall on the central Gulf coast several days earlier and produced only two brief, weak tornadoes. It later intensified during a baroclinic interaction while maintaining tropical classification, as it crossed the southern Appalachians and before its re-entry over water. Hurricane Earl (1998) also made landfall off the Gulf, then, after a relative minimum in post-landfall tornadoes, produced a killer tornado in South Carolina before moving offshore. This article documents tornado occurrences with exiting TCs and their distribution during the 1955-1998 period.


1One of these tornadoes was not listed as a tornado in Storm Data, but instead as a waterspout (which moved onshore, damaging two homes and injuring one person). No F-scale damage rating was given.


Exiting TC candidates were identified using the National Hurricane Center's (NHC) online maps of tropical cyclone tracks, available through their Internet website (, and published annually in Weatherwise season summaries (e.g., Rappaport 1998).

To qualify as an exiting TC, a cyclone must have been classified by NHC as a tropical depression, tropical storm or hurricane as its center made landfall on the U.S. Gulf or Atlantic coasts. It must have remained inland for a minimum of 12 hours and/or 200 nautical miles (371 km), and moved back offshore as a classified cyclone of either tropical or extratropical character from the U.S. Atlantic coast north of Florida. [Because of the short overland residence of TCs crossing peninsular Florida, and the resulting similarity of their landfall and exit environments, those systems were excluded. Similar reasoning precluded those TCs making only brief passages inland from the Atlantic coast.]

Exiting TCs before 1995 were systematically examined for all tornado occurrences using SVRPLOT software (Hart 1993). That SVRPLOT version used tornado county-segments; however, no tornadoes in this study were found to have crossed county lines, rendering the segment data representative of whole tornado reports. Since SVRPLOT-format raw data files are not yet available for subsequent years, tornado tabulation and mapping for 1996 and 1997 were compiled from Storm Data (NCDC 1996, 1997). Final storm data -- including damage ratings -- for Hurricane Earl of 1998 was not available as of this writing; so its tornado information was culled from preliminary operational severe weather logs compiled at SPC. For consistency, all 1995-1998 tornado data for exiting TCs was mapped using SVRPLOT format for consistency.

For systems making Gulf landfall, exiting-TC tornadoes are defined as those occurring closer to the Atlantic coast (north of Florida) than to the Gulf coast. For TCs making Atlantic landfall, associated tornadoes must be closer to the exit point than the entrance. One example of an exiting TC tornado event, Danny of 1997, is shown in Fig. 1 (below). F-scale damage ratings (Fujita 1981) were also accumulated from SVRPLOT and Storm Data for available events.

Figure 1. Path of TC Danny (July 1997) before exit, and location of associated tornadoes (small filled triangles). TC position dates/times are in UTC.

One of the non-qualifying TCs, Tropical Storm Beryl of 1994, was an unusually prolific tornado producer over several Atlantic coast states -- well after a relatively tranquil Gulf coast landfall (Vescio et al. 1996). It was not mapped by NHC as a classified tropical or extratropical cyclone by the time its remnants moved offshore from New England; however, its path, evolution and tornadic character otherwise closely resembled a few TC cases used here. It may, therefore, be exempted from criteria and included in the database upon further review.


Tornadic exiting-TC events are listed in Table 1. [Click here for a map of all TC exit tornadoes.] With a small sample size (8 exiting-TC cases in 42 years), there are naturally some concerns about climatological representativeness. There was no clear preference for whether TCs produced exit-tornadoes at higher proportions based on the basin from which the cyclones moved inland. Of tornadic exiting TCs, 25% made Atlantic landfall, corresponding closely to the portion of Atlantic landfalls among all exiting TCs (27%). Of 22 total exiting TCs for the study period, eight (36%) spawned tornadoes during their exit phase. This is 11% greater tornado production than for tropical storms and hurricanes as a whole as tabulated by Novlan and Gray (1974). [Their database did not include tropical depressions; and more modern classification practices are not be the same.]

There was a strong preference (75%) for hurricane strength at landfall among the exit-tornado producing systems, regardless of exit classification; however, one TC (Alma of 1970) had weakened to tropical depression status in the Caribbean -- over 1200 km from its Florida landfall. [Click here for a cyclone track map of all TCs which produced exit tornadoes.]

In most exiting-TC tornado cases, there was a well-defined spatial separation or clustering of tornadoes meeting "exit" criteria, relative to tornado occurrences (if any) near landfall. For example, tornadoes were reported over central Florida before the time of landfall of Hurricane Earl (Fig. 2). Though it had a characteristically tropical warm core aloft, operational surface, satellite and upper air analyses at SPC indicated distinct baroclinic characteristics including a warm front analyzed east of its center -- beginning before landfall. Most tornadoes from Earl occurred in or within 100 km north of the warm front, with a pronounced lull during the period the front moved northward, remaining east to northeast of the cyclone center across northern Florida and most of southeastern Georgia.

Figure 2. Preliminary mapping of tornadoes (filled triangles) associated with Hurricane Earl, September 1998. Thick line represents Earl's path, with approximate landfall and exit times pending final NHC "best track" data.

Apparent baroclinic influences appear in every exiting-TC tornado event. Half the cyclones which produced exit tornadoes were classified as extratropical upon moving offshore; and three of the four cyclones which produced at least eight tornadoes were extratropical. Examination of daily weather maps, rawinsonde plots and/or operational map analyses for the remainder indicated the presence of low level fronts and/or mid-tropospheric baroclinic waves within 1000 km of the cyclone centers at exit.

F-scale ratings were available for 38 tornadoes; their spectral distribution is shown in Fig. 3. Slightly less than one-fifth (19%) were significant events (Hales 1988) with F2 damage as compared to 32% of all TC tornadoes between 1964-1983 (Weiss 1987). Although F3 and F4 tornadoes have occurred with landfalling TCs (Grazulis 1993, Vescio et al. 1996), none were found in exiting systems.

Figure 3. Distribution of available F-scale ratings for tornadoes with exiting TCs, 1955-1998.


More intensive examination of surface and upper air data is warranted for the eight exiting TC cases in order to determine the extent and magnitude of various parameters associated with TC supercells and tornadogenesis. Such parameters include dry air entrainment into the cyclones (Hill et al. 1966), vertical shear in the lowest 1 km (Novlan and Gray 1974), and various derived kinematic and bulk parameters (McCaul 1991). This should include more thorough analyses for each event of observed proximal baroclinic features surface and aloft as possible contributors to the environmental shear and instability profiles favorable for TC tornadoes.

Hurricane-spawned tornadoes are often associated with mini-supercells, which have been commonly observed using WSR-88D radar data (e.g., Spratt et al. 1997), and which have been successfully numerically simulated (McCaul and Weisman 1996). These articles detailed some observational aspects of TCs Allison (1995) and Danny (1985) respectively, primarily during that portion of their lifespans prior to exit phase. Each storm produced exit tornadoes also; and further analysis is needed to compare kinematic and thermodynamic profiles in the exit stage with conditions closer to landfall that were well-documented in their research.

Violent (F4, F5) and F3 tornadoes have not been observed with exiting TCs. Given the inherent interpretative subjectivity of the F-scale (Doswell and Burgess 1988), however, the possibility of such tornadoes producing F3 or greater damage cannot be disregarded. Storm-relative azimuthal mapping of both tornado locations and their F-scale is planned, analogous to the motion-relative F-scale sums plotted by McCaul (1991). In addition to most-intense spot damage -- the basis for each F-scale rating -- numbers and areal coverage of tornadoes are factors in the public tornado hazard and in the SPC forecast decision process. For this reason, comparative analysis of total tornado event threat associated with exiting TCs may be performed using the Destruction Potential Index (Thompson and Vescio, 1998). Such an analysis could be extended to a much broader spectrum of TC tornado events in a separate study.


The author thanks John Hart, Dave Imy, Bob Johns, Steve Weiss and Mike Vescio of SPC, as well as Jack Beven of NHC, Bill McCaul and Lon Curtis, for their ideas, aid in analyzing data and/or intelligent discussions while compiling this article.


Doswell, C. A., III, and D. W. Burgess, 1988: On some issues of United States tornado climatology. Mon. Wea. Rev., 116, 495-501.

Fujita, T.T., 1981: Tornadoes and downbursts in the context of generalized planetary scales. J. Atmos. Sci., 38, 1511-1533

Gentry, R.C., 1983: Genesis of tornadoes associated with hurricanes. Mon Wea. Rev., 111, 1793-1805.

Grazulis, T.P., 1993: Significant Tornadoes: 1680-1991. Environmental Films, St Johnsbury VT, 1326 pp.

Hales, J.E., 1988: Improving the watch/warning system through use of significant event data. Preprints, 15th Conf. Severe Local Storms, Amer. Meteor. Soc., Baltimore, 165-168.

Hart, J.A., 1993: SVRPLOT: A new method of accessing and manipulating the NSSFC severe weather database. Preprints, 17th Conf. Severe Local Storms, St. Louis, Amer. Meteor.Soc., 40-41.

Hill, E.L., W. Malkin and W.A. Schulz, Jr., 1966: Tornadoes associated with cyclones of tropical origin -- practical features. J. Appl. Meteor., 5, 745-763.

Hoadley, D.K., 1981: A Tropical Storm David tornado in Fairfax County - September 1979. Bull. Amer. Meteor. Soc., 62, 498-507.

McCaul, E.W., Jr., 1987: Observations of the Hurricane "Danny" tornado outbreak of 16 August 1985. Mon. Wea. Rev., 115, 1206-1223.

_____, 1991: Buoyancy and shear characteristics of hurricane-tornado environments. Mon Wea. Rev., 119, 1954-1978.

_____ and M.L. Weisman, 1996: Simulations of shallow supercell storms in landfalling hurricane environments. Mon. Wea. Rev., 124, 408-429.

NCDC, 1996: Storm Data, 38, 10, 155 pp.

_____, 1997: Storm Data, 39, 7, 347 pp.

Novlan, D.J., and W.M. Gray, 1974: Hurricane-spawned tornadoes. Mon Wea. Rev., 102, 476-488.

Rappaport, E.N., 1998: Atlantic Hurricanes. Weatherwise, 51, 2, 43-46.

Spratt, S.M., D.W. Sharp, P. Welsh, A.C. Sandrik, F. Alsheimer and C. Paxton, 1997: A WSR-88D assessment of tropical cyclone outer rainband tornadoes. Wea. Forecasting, 12, 479-501.

Thompson, R.L., and M.D. Vescio, 1988. The Destruction Potential Index - a method for comparing tornado days. Preprints, 19th Conf. Severe Local Storms, Amer. Meteor. Soc., Minneapolis, 280-282.

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

Weiss S.J., 1987: Some climatological aspects of forecasting tornadoes associated with tropical cyclones. Preprints, 17th Conf. Hurricanes and Tropical Meteor., Amer. Meteor. Soc., Miami, 160-163.