An Historical Perspective

John T. Snow[*], Amy Lee Wyatt, Ann K. McCarthy and Eric K. Bishop
The University of Oklahoma
Norman, Oklahoma


Many times each year tornadoes cause damage to countless homes and businesses in the United States. Great efforts have been made to understand how tornadic winds damage structures, and there have been a few investigations of low-altitude lofting and deposition of material by tornadic winds. However, the lofting of debris to great heights, its long-distance transport by tornadic thunderstorms, and its fallout far downstream have received little attention.

These phenomena pose interesting scientific questions and challenges, particularly when considering the risk posed by the lofting, transport and fallout of hazardous material, such as toxic chemical or radioactive waste. If long-distance transport of debris by tornadic thunderstorms is a common event, then tornadoes pose serious hazards well beyond the direct impact of their high winds.

1.1 The Barneveld, Wisconsin Tornado

Motivation for this study was provided by Anderson's (1985a,b) investigation of debris fallout from the 7/8 June 1984 outbreak of tornadoes in Wisconsin. One of these tornadoes, rated F5 (Fujita 1971, 1973), destroyed approximately 90% of the village of Barneveld. Debris from Barneveld was lofted, transported, and deposited in a swath running from southwest to northeast across Wisconsin. Following the outbreak, Anderson and his students conducted a ground survey and a mail and telephone campaign to obtain data on the fallout of debris. Anderson classified fallout from this storm as follows:

Anderson's survey (Anderson 1985a) revealed a "more or less continuous area of debris stretching from Barneveld to west of Neenah, some 102 miles northeast", with a width ranging from "a few hundred yards just north of Barneveld to a maximum of 20 miles some 65 miles downstream." Within the contiguous area, heavy material lay along a narrow path about 85 miles in length. Lighter materials "were not deposited symmetrically around the heavy debris path but [lay] more to the east, making the heavy debris path lie nearer the western edge of the contiguous deposit area." Paper debris was found in widely scattered locations throughout and north of the contiguous area, some as far as 135 miles away (to the northeast). (See Fig. 1 in Anderson 1985a, reproduced as Fig. 2b in Snow et al. 1995, for a mapping of the fallout.)

1.2 Objectives

While showing conclusively that lofting, transport, and fallout occur, Anderson's study is the only one of its type in modern times (see Peterson 1993 for discussion of three early European studies). It does not provide a sufficient basis from which to judge either how often these phenomena occur or what happens in tornadoes of lesser intensity. We would like to be able to answer questions such as

Here we address these questions by reviewing accounts in the historical record, and draw some preliminary conclusions.


A review of accounts of tornadoes -- by no means exhaustive -- finds a number of reports of debris being transported long distances. While many of these accounts -- especially those prior to 1950 or so -- are highly subjective and likely exaggerated, when taken together they provide useful insights on lofting, transport, and fallout. In Grazulis (1993), for the period 1871 through 1990, 86 out of 12,651 tornado accounts provided 121 reports of debris found more than five miles downstream. A review of other publications on significant tornadic events which occurred in this same 120-year period provided an additional 42 reports. Items reported to have been transported by tornadic thunderstorms range from paper products, such as canceled checks and photographs, to large, heavy objects, such as an airplane wing and a carton of deer hides.

A year-by-year distribution of the 93 tornado events which yielded the 163 reports from the historical record shows that although tornadoes with reports of long-distance transport occur with regular frequency, more than half (50 events, 90 reports) occurred in the fifty years prior to 1920. In the period 1871 through 1990, the number of tornadoes reported annually has increased from less than 100 in the period 1871-1916 (Fujita 1987, p. 36) to greater than 1100 in 1990. Therefore, the number of tornado events with reports of long-distance debris transport, per year, as a fraction of the total number of recorded tornadoes for that year, has decreased in recent years. We conjecture that this is due in part to a shift (beginning in the early 1950's and with increased emphasis since the mid-1970's) from detailed analyses of each year's relatively few significant events, to a risk-assessment approach that is focused on characterizing the intensities for all tornadoes that occur in a given year. In using the current F scale for this characterization, only damage in the path of a tornado need be assessed to ascertain intensity. Entries in Storm Data tend to be short and limited to items that substantiate estimates of intensity. Further, this approach to documenting all tornadoes has increased most notably the number of weak (F0 and F1) tornadoes reported each year, events for which debris lofting and long-distance transport seem less likely.

Figure 1 shows the number distribution of the 163 debris reports as a function of debris type and distance transported. Using Anderson's definitions, the transported items were categorized as "paper", "light", or "heavy." If the account of an event was unclear as to the size or weight of the debris, the item was classified as having "unknown" weight. (While it would be desirable to also characterize transported objects in terms of their aerodynamic shapes, this proved impractical due to the lack of information in most historical accounts. An important exception is "paper", especially personal bank checks, since such items are very often of standard size.)

Figure 1. Plot of number of debris reports as a function of type and distance for the 120-year period 1871-1990. Debris is categorized as paper, light (weighing less than 1 lb), heavy (weighing more than 1 lb), or unknown weight (not paper). Total number of reports plotted: 163. As a result of an arbitrarily selected cutoff used in screening the tornado database, there are no reports in the interval 0-5 miles.

The overall shape of the distribution in Figure 1 is roughly what one would anticipate. Most items fall out close to their points of origin. Heavy items tend to come down closest to their source locations. Paper items, the lightest materials, are carried the farthest downstream. The secondary peak in the distribution which occurs at a distance of 85-90 miles (135-145 km) in Figure 1 (and also in Figure 2) is due to multiple reports of debris found at the same location, all associated with the F4 Great Bend, Kansas, tornado of 10 November 1915.

Figure 1 suggests that most heavy debris falls out within 50 miles (80 km) of its source, while most light debris falls out within 90 miles (145 km) from its source. Paper debris, on the other hand, can be deposited much farther. The farthest reported distance for any fallout is 210 miles (335 km); this was a canceled check, found in Palmyra, Nebraska, transported by the thunderstorm that produced the 1915 Great Bend, Kansas tornado. This thunderstorm also transported light and heavy objects farther than any other storm on record. Clothing, shingles, and fragments of books were found in Glasco, 87 miles (140 km) to the northeast of Great Bend. A sack of flour, categorized as heavy, was found about 110 miles (175 km) to the northeast. Grazulis (1993, personal communication) noted this to be "perhaps the longest distance ever recorded for an object weighing more than one pound." This particular report may be anomalous, since the second-farthest report of transport of a "heavy" item is only 50 miles (80 km).

Table 1 shows the 163 debris reports as a function of the F scale rating for the lofting tornado. These and subsequent statistics using the F scale must be interpreted with caution, especially if inferences are to be made about tornado dynamics and wind speeds, as the F scale is really a tornado damage rating scale -- see Doswell and Burgess (1988). (Recall that the F-scale value assigned to an event characterizes the most severe damage observed; typically this will have occurred over only a small portion of the tornado's track.) While there is some correlation between damage rating and tornado strength, the correlation is not a perfect one. Further uncertainties enter in the assignment of F-scale values from anecdotal information. The net effect is that the F-scale value assigned to many events probably contains an uncertainty approaching +/-1; there also appears to be a tendency to err on the high side in many events.

F scaleHeavyLightPaperUnknownAll types
F23 (3)2 (1)1 (1)4 (4)10 (9)
F33 (3)9 (9)7 (3)15 (15)34 (27)
F420 (13)28 (12)32 (20)22 (20)102 (49)
F54 (4)2 (2)8 (4)3 (3)17 (8)
Total30 (23)41 (24)48 (28)44 (42)163 (93)

Table 1. Distribution of 163 debris reports as a function of tornado F scale for the 120-year period 1871-1990. Number of events associated with debris reports is listed in parentheses.

In Table 1, there are no data on fallout linked to F0 or F1 tornadoes since Grazulis focused on strong (F2 and F3) and violent (F4 and F5) tornadoes. The largest number of reports of long-distance debris transport (about 75% of the total) are associated with violent tornadoes, particularly those rated F4 in intensity. While the number of reports associated with F5 tornadoes appears small, this is a result of the fact that damage of this intensity is very rare (on the order of 0.1% of all tornadoes, based on tornado accounts 1953 through 1990).

However, F4 events are also a small fraction of all tornadoes (on the order of 1%). Hence, that 57 violent events appear here suggests that a relatively high percentage of violent tornadoes (roughly 4%) leads to long-distance debris fallout. In contrast, while there are some reports of long-distance debris transport by strong tornadoes, the number here is a relatively small percentage of all such events (roughly 0.4%). Since strong events account for roughly 28% of all tornadoes, this suggests debris transport by thunderstorms producing such events is rare.

Figure 2 shows the same number distribution as Figure 1, but with the bars now showing dependence on F scale. The items which traveled the farthest distances, 205 and 210 miles, were both transported by thunderstorms producing F4 tornadoes. The greatest distance traveled by debris associated with an F2 tornado was 130 miles by a piece of copper plate from the 20 March 1888 tornado in Calhoun, Georgia (Grazulis 1993).

Figure 2. Plot of number of debris reports as a function of F scale and distance for the 120-year period 1871-1990. Overall distribution is the same as in Figure 1, but each bar is now differently proportioned.

Sixty-eight debris reports from 31 of the 93 tornado events in the historical record -- ones in which the descriptions of the fallout were particularly detailed -- were combined to produce Figure 3, a composite plot of fallout locations with respect to estimated storm motion. Cases used in the plot were those in which a debris report was linked to a town name, or in which a direction anddistance from the debris source were given. Locations of fallout finds and source points were given in 48 reports. In the other 20 reports, a direction and distance from the source were provided. (Although there were actually 83 usable reports, only 68 are plotted in Figure 3; the other 15 overlap, i.e., same event, same type, same fallout location.)

Figure 3. Composite pattern of 68 debris reports associated with 31 tornadic storms, 1871-1990. The arrow indicates "storm direction" and has arbitrarily been oriented SW to NE. Debris items are plotted by azimuth and distance from their source.

Using latitude and longitude coordinates of the noted towns or other locations, an azimuth and distance from the source was computed for each fallout site. The direction of storm movement for each event was estimated very roughly (to within +/- 10°, say) as the direction of the line tangent to the beginning portion of the tornado track; here this will be referred to as "storm direction." The tracks were then rotated to align along a common direction (here drawn aesthetically as 225°, or from southwest to northeast) and relative debris locations plotted accordingly. Because of the small number of reports, no effort was made to stratify these data by tornado intensity.

Figure 3 shows that, of the 68 debris items plotted, 53 items (78%) fell to the left of the composite "storm direction". Of the 15 items which fell to the right, only two items (one "paper" and one "light") fell beyond -10°. Therefore, 66 items of debris (97% of the reports) fell either approximately along (i.e., within 10° to the right or left) or to the left of "storm direction."

Relative to the "storm direction", the composite debris pattern shown in Figure 3 suggests the most concentrated area of debris is between 10° and 35°, that is, to the left of the composite "storm direction"; 54% of the points fall in this area. Thirty-two percent fall along the "storm direction". Paper was scattered the farthest, for distances as great as 210 miles, between angles of -40° and 70°. Light debris was found up to 87 miles downstream, between angles of -38° and 75° heavy items, with the exception of the flour sack, were carried up to 50 miles, between angles of -10° and 75°. Items of unknown weight were carried up to 130 miles, between angles of -5° and 80° with respect to "storm direction."

This pattern of long-distance debris transport and fallout has similarities to that found by Anderson for the Barneveld tornado. In constructing a fallout map, Anderson found that the "...contiguous area of debris [lay] mainly to the west and north of the...tornado path", that is, most of the fallout was to the left of the storm's track. (Similarly, Peterson 1993 has described three European events in which the debris was to the left of the tornado track: the St. Claude, France tornado of 1890; the Woldegk, Germany, tornado on 29 June 1764; and the 24 July 1930 tornado in the Treviso-Udine district of Italy.)


Although it is premature to draw general conclusions about these phenomena, the historical record does suggest the following points:

Acknowledgements. The late Dr. Charles E. Anderson provided the inspiration for this work through his study of the Barneveld, Wisconsin tornado. The historical review could not have been done without the extensive analysis of tornado records that has been carried out by Mr. Thomas P. Grazulis over the last decade. Mr. Michael Magsig, Mr. Carl Levison and Dr. Chris Church read several rough drafts and provided excellent sounding boards. This work is supported by the National Science Foundation under Grant ATM9411767.


Anderson, C.E., 1985a: The fall-out pattern for debris for the Barneveld, WI tornado: an F-5 storm. Preprints, 14th Conf. on Severe Local Storms, Indianapolis, IN, Amer. Meteor. Soc., 264-266

______, 1985b: The Barneveld tornado: a new type of tornadic storm in the form a spiral mesolow. Preprints, 14th Conf. on Severe Local Storms, Indianapolis, IN, Amer. Meteor. Soc., 289-292.

Corliss, W.R., 1983: Tornados [sic], Dark Days, Anomalous Precipitation, And Related Weather Phenomena. The Sourcebook Project, Glen Arm, MD, 196 pp.

Davies- Jones, R.P., 1981: Acquisition and analysis of severe storms and tornado field observations. Section 4 (pp. 21-66) of Summary of AEC-ERDA-NRC Supported Research At NSSL 1973-1979 (J.T. Lee, ed.), NOAA Technical Memorandum ERL NSSL-90, 93 pp.

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

Felknor, P.S., 1992: The Tri-State Tornado. Iowa State University Press, 131 pp.

Fujita, T.T., 1971: Proposed characterization of tornadoes and hurricanes by area and intensity. Satellite and Mesometeorology Research Project Research Paper 91, Department of the Geophysical Sciences, The University of Chicago, 44 pp.

______, 1973: Experimental classification of tornadoes in FPP scale. Satellite and Mesometeorology Research Project Research Paper 98, Department of the Geophysical Sciences, The University of Chicago, 15 pp.

______, 1987: U.S. Tornadoes. Part One: 70-Year Statistics. Satellite and Mesometeorology Research Project Research Paper 218, Department of the Geophysical Sciences, The University of Chicago, 122 pp.

Grazulis, T.P., 1980: Indiana Killer Tornadoes, poster.

______, 1993: Significant Tornadoes 1680 - 1991. The Tornado Project of Environmental Films, 1326 pp.

Hayes, M.W., 1927: The St. Louis Tornado of September 29, 1927. Mon. Wea. Rev., 55, 405-407.

Levison, C.H., M.A. Magsig, J.T. Snow, A.L. Wyatt, 1996: Fallout of debris from tornadic thunderstorms 2: Examples from 1994 and 1995. Preprints, 18th Conf. on Severe Local Storms, San Francisco, CA, Amer. Meteor. Soc.

O'Toole, J.M., 1993: Tornado! 84 Minutes, 94 Lives. Databooks, 284 pp.

Peterson, R.E, 1993: Far-field tornado debris patterns. Preprints, 17th Conf. on Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc., 319-322.

Smith, H.E., Jr., 1982: Killer Weather. Dodd, Mead & Company, 224 pp.

Snow, J.T., A.L. Wyatt, A.K. McCarthy, E.K. Bishop, 1995: Fallout of debris from tornadic thunderstorms: An historical perspective and two examples from VORTEX. Bull. Amer. Meteor. Soc., 76, 1777-90.

Stanford, J.L., 1987: Tornado: Accounts of Tornadoes in Iowa. 2nd Ed. Iowa State University Press, 143 pp.

Tanner, R.W. (Ed.), 1990: Southwestern Nebraska Tornado of June 15th. Storm Data, 32(6), 25-27.

Walz, F.J., 1917: Tornado of March 23, 1917, at New Albany, Ind. Mon. Wea. Rev., 45, 169-171.

[*] Corresponding author address: Dr. John T. Snow, College of Geosciences, 100 E. Boyd St., Suite 710, Norman, OK 73019-0628; e-mail jsnow@ou.edu.


This article contains much of the same information contained in Fallout of Debris from Tornadic Thunderstorms: An Historical Perspective and Two Examples from VORTEX, less the discussion of the two cases from 1994. Also, Table 1 has been updated to contain additional information.

For a list of our most recent findings, see the Addendum to the BAMS article.

Return to publications page

Return to Tornado Debris Project home page