spatial and temporal characteristics of tornado path directionchubasco.niu.edu/pubs/suckling and...

Spatial and Temporal Characteristics of Tornado PathDirection*

Philip W. SucklingTexas State University–San Marcos

Walker S. AshleyNorthern Illinois University

Common perception is that tornadoes travel in paths from the southwest quadrant of directions toward thenortheast. This study examines path directions for 6,194 tornadoes that occurred in the eastern two-thirds of theUnited States during the twenty-three-year period 1980–2002. At the national scale, nearly 70 percent oftornadoes included in the study propagated from the west, west-southwest, and southwest, with west-southwestbeing the highest frequency origin direction. Nevertheless, distinct seasonal and regional variations were found.In central and northern areas of the country, a more westerly or northwesterly path origin prevails during latespring and summer. The midtropospheric flow, convective typology, and synoptic patterns of tornado outbreaksare thought to contribute to the distributions observed in the climatology. Key Words: natural hazards,tornadoes, tornado path.

Introduction

Hazard mitigation studies indicate that overthe past half century improved warning

technologies and community preparedness havesubstantially reduced fatalities and injuries as-sociated with tornadoes (Boruff et al. 2003).Tornado hazard studies and associated risk-as-sessment models frequently focus on spatialcharacteristics such as track length and trackwidth, but often overlook path direction (e.g.,Schaefer, Kelly, and Abbey 1986; Grazulis,Schaefer, and Abbey 1993; Brooks 2004). Tor-nado path direction, however, is not a trivialmatter for tornado hazard mitigation. Whenconsidering the safest location within a buildingin the event of a tornado, at one time the south-west corner of a basement was considered thesafest, on the theory that wind-blown debrisfrom the damaged building would accumulatein the northeast corner of the basement as thetornado moved from the southwest toward thenortheast (Finley 1887; Flora 1954). Althoughtornado path direction, as we will show, is in-deed generally from the southwest, research byEagleman (1967) and others has shown that the

safest location is opposite to the southwest cor-ner since debris actually accumulates nearest tothe initial impact (Eagleman, Muirhead, andWilliams 1975; Grazulis 2001).

Current tornado safety recommendations forbuildings assert that one should seek shelter inan interior, reinforced room or corridor in thelowest level of a structure (AMS 2000; FEMA2004; NWS 2004). In addition to this primaryrule, safety guidelines, research, and accountscontinue to indicate that the ‘‘downstream’’portion of a building will be somewhat saferthan the ‘‘upstream,’’ relative to the approach-ing tornado (Eagleman 1967; Eagleman, Muir-head, and Williams 1975; Tornado ProjectOnline 1999; AMS 2000). Clearly, such guide-lines need to be revaluated in light of recentwind engineering research, advancements, andtechniques. For example, should ‘‘safe rooms’’and shelters be constructed away from the west-,southwest-, and south-facing walls, which,based on climatological data, have a higherprobability of receiving the highest wind loadsand therefore have a higher chance of collapse(Eagleman 1967)? Clearly, microscale vortexand structure interactions make it difficult to

* The assistance of Jamie Dyer (Climatology Research Laboratory, University of Georgia) is gratefully acknowledged. The authors thank theanonymous reviewers, whose comments helped us clarify and improve the manuscript.

The Professional Geographer, 58(1) 2006, pages 20–38 r Copyright 2006 by Association of American Geographers.Initial submission, March 2004; revised submission, November 2004; final acceptance, December 2004.

Published by Blackwell Publishing, 350 Main Street, Malden, MA 02148, and 9600 Garsington Road, Oxford OX4 2DQ, U.K.

determine where debris will fall. However, cli-matological information such as that derivedfrom this study can be utilized in future engi-neering studies to investigate the safest locationin structures for different tornado scenarios.

Additionally, data on tornado path directionare also extremely important to public officialsand information managers at local emergencyoperational facilities who must determine thedistribution of resources and storm spotters insevere weather situations. Results from thisstudy show that spotters should be positioned tohave an unobstructed view to the west andsouthwest horizons, where a majority of torna-does originate. Understanding climatologicalfeatures such as predominant tornado path di-rection can aid in general tornado awarenessand in obtaining critical information quickly sothat officials can make timely local warning dis-semination decisions.

Common perception is that U.S. tornadoestravel in paths from the southwest toward thenortheast (Eagleman, Muirhead, and Williams1975; Eagleman 1990; Grazulis 2001). Howev-er, there has been very limited research on tor-nado path direction for the United States, andmost studies that have examined tornado pathdirection have only considered specific states.For example, estimates from the early twentiethcentury indicated that in Kansas 61 percent oftornadoes came from the southwest, 16 percentfrom the west, and 11 percent from the north-west (U.S. Department of Commerce 1953). Itwas shown that in Illinois more than 80 percentof tornadoes traveled from the southwestthrough west directions (Wilson and Chang-non 1971). In a study of the state of Iowa basedon the thirteen-year period 1959–1971, Notisand Stanford (1973) reported that almost70 percent of Iowa tornadoes traveled fromthe southwest quadrant, but about 30 percentcame from the west or northwest directions.The descriptive climatology by Schaefer et al.(1980) examined national tornado path direc-tion, albeit briefly, and concluded that 87percent of conterminous U.S. tornadoes travelfrom the southwest quadrant. Schaefer et al.(1980) also noted that when tornado motion isconsidered on a seasonal basis, only summer-time storms differ markedly from this generalpattern, with one of four summer tornadoestraveling from the northwest quadrant of direc-tions. Fujita (1987) briefly examined tornado

path direction in a climatological study of tor-nado data from the period 1916–1985 and con-cluded that 59 percent of tornadoes moved fromthe southwest.

As illustrated in this review, there is limitedresearch on tornado path direction for the Unit-ed States, especially during the past twentyyears. Recent enhancement of tornado report-ing procedures by the National Weather Service(NWS), the development and enhancement ofan integrated warning system (Doswell, Moller,and Brooks 1999), increased population, in-creased public awareness, and the proliferationof video cameras have led to improvements inthe accuracy and consistency of the tornado re-port database. Despite inherent limitations ofthe raw tornado report database (e.g., see Do-swell and Burgess 1988; Grazulis 1993; Brooks,Doswell, and Kay 2003), these recent improve-ments and enhancements in the data allow fordetailed examination of the more recent portionof the database to uncover characteristics oftornado path directions. Given the potentialimportance of tornado path direction to hazardmitigation recommendations, this study seeksto provide a comprehensive geographic study ofpath direction for tornadoes occurring in thecentral and eastern United States. We also spec-ulate that the interaction of processes across amultitude of scales, including storm-scale dy-namics and synoptic-scale steering winds, areimportant controls on the distributions de-scribed.

Data and Methods

Data regarding U.S. tornadoes are now muchmore reliable and complete than ever before(Burgess, Donaldson, and Desrochers 1993;Schaefer et al. 1993; Grazulis 2001). Severalfactors account for the improvement in the tor-nado report database. These include emphasison and improvement of warning verificationpractices by the NWS since the early 1980s( Johns and Evans 2000), increased use and im-provement of spotter networks (Doswell, Moll-er, and Brooks 1999), modernization of theNWS during the early 1990s with improvedtechnologies including deployment of Dopplerradar (Friday 1994), and improved communi-cations (e.g., cell phones) and documenting me-dia (e.g., video cameras). Despite improvementin the quality and consistency of the data, there

Spatial and Temporal Characteristics of Tornado Path Direction 21

remain a number of limitations and biases in theraw tornado report dataset. These limitationsand biases in the tornado and other severeweather datasets have been described elsewhere(e.g., Doswell and Burgess 1988; Grazulis1993; Brooks, Doswell, and Kay 2003; Schaefer,Weiss, and Levit 2003) and involve such factorsas basic errors in reporting or recording of timeand location information, spatial and temporalvariability in the efforts to collect severe weatherreports for warning verification programs, ad-justments in the nature of detailed damage sur-veys, and population changes.

Detailed characteristics of tornadoes in theUnited States are collected by local NWS of-fices and, thereafter, published in the NationalClimatic Data Center (NCDC) publicationStorm Data and made available on the NCDCstorm events Web page1 or in the SeverePlotsoftware package2 (Hart 1993). For this study,we restricted our analyses to 1980–2002 for two

reasons: (1) prior to this period the data becomeincreasingly suspect and incomplete (e.g., seeKelly et al. 1978) and (2) a twenty-three-yearperiod should provide a long enough record toaccount for any climatological shifts in favor-able synoptic patterns for severe weather devel-opment (e.g., see the effects of favorable, long-term severe weather pattern shifts on derechoclimatologies presented by Bentley and Sparks2003 and Coniglio and Stensrud 2004).

Because tornadoes are rare west of the Con-tinental Divide, we restricted our analysis to athirty-seven-state area across the central andeastern United States. In an attempt to examinepossible region-specific shifts in the climatolo-gy, we divided that area into six regions (Figure1). The regions were divided primarily alongstate lines due to the state-by-state reportingstrategy utilized by Storm Data and becausemitigation efforts are often implemented on astate-by-state basis. Subdividing the thirty-sev-

Figure 1 Map of the six regions selected for the regional-scale study. Tornadoes that occurred within

these six regions were utilized for the national-scale analysis.

22 Volume 58, Number 1, February 2006

en-state area into regions was a somewhat sub-jective process and we realize that some differ-ences of opinion may arise as to which statesbelong in which regions.

To analyze tornado path direction, detailedlatitude and longitude coordinate informationfor the start and end points of each individualtornado is needed. Many tornadoes in the da-tabase are either small, short-lived, or lack post-storm survey team analysis and, therefore, in-clude only ‘‘touchdown’’ locations. In fact, morethan 70 percent of tornadoes in the study periodlacked full track information (Table 1), and in anumber of cases where a track length was indi-cated no tornado end-point information wasavailable. Since 97 percent of the tornadoes witha length of less than 1.61 km (1 mile) had in-complete track data, we restricted our analysesto tornado events greater than or equal to thislength. More than 6,000 tornadoes meeting thestart and end-point and minimal length criteriawere available for analysis (Table 1). Tracks forthese tornadoes are plotted in Figure 1, reveal-ing the distribution of events across the nation.

A tornado can indeed change path directionduring its life span. Nevertheless, for the pur-poses of this study, a straight path from startlocation to end location is assumed. We believethis is a safe assumption because tornadoescharacteristically have much smaller ‘‘side-to-side’’ deviation along the track in comparison tothe overall length of the along-track vector. Pathdirections according to path origin (i.e., the di-rection from which the tornado is traveling) arecategorized into sixteen directions (N, NNE,NE, ENE, etc.). Again, these directions refer tothe direction from which the tornado is ap-proaching (i.e., WSW means that the tornado istraveling from the west-southwest and movingtoward the east-northeast).

Path direction origin is illustrated using afigure similar to a wind-rose diagram, known asa ‘‘radar diagram,’’ with the frequency of torna-do events plotted according to the sixteen car-dinal directions (e.g., Figure 2). The radardiagrams in this study utilize a floating scale ofactual frequency numbers, instead of percent-ages, for the period in question. The actualmagnitudes are employed in the diagramsrather than percentages because they illustratethe importance of seasonality and sample sizeon developing a tornado climatology. None-theless, noteworthy percentages are indicatedthroughout the text.

Results: National Scale

At the national scale, tornadoes travel in pathsdominantly from the W, WSW, and SW direc-tions (69.3 percent of tornadoes included in thestudy) with WSW having the modal frequency(Figure 2). The frequency of tornadoes trave-ling from the NW (3.2 percent) and WNW (7.0percent), and from the SSW (8.1 percent) and S(4.5 percent) is notable, but these directions arefar less common than the three main directionsof W (21.6 percent), WSW (26.2 percent), andSW (21.5 percent). Tornadoes do travel in pathsfrom all nine of the remaining cardinal direc-tions, but far less often. In the thirty-seven-stateregion, 65.1 percent of all tornadoes move intothe northeast quadrant. This percentage is farless than the 87 percent indicated for nationwidetornadoes from 1950–1978 by Schaefer et al.(1980). From 1980–2002, 28.2 percent of torna-does propagated into the southeast quadrant.Tornadoes moving toward the west were ex-tremely rare during this period, with only 4.0percent of events moving into the northwestquadrant and 2.7 percent of tornadoes movinginto the southwest quadrant.

The dominant easterly and northeasterly mo-tion of tornadoes described in our twenty-three-year climatology is likely due to severalfactors, including the predominance of the mid-latitude westerlies across the study region, tor-nado-producing storm typology, and thetendency for tornado outbreaks in the easterntwo-thirds of the United States to occur in areasof southwesterly flow at 500 hPa with a longwave trough to the west of the outbreak area(Beebe 1956; Miller 1972; Galway and Pearson1981).

Table 1 Number of tornadoes during the twenty-three-year period 1980–2002

F-scale Totalnumber

Trackinformationincomplete

� 1.61 kmpath

length

� 1.61 kmlength withcomplete

track information

F0 11754 10140 2615 349

F1 7017 4474 4195 2452

F2 2438 826 2060 1598

F3 708 77 691 629

F4 158 5 156 153

F5 13 0 13 13

Total 22088 15522 6730 6194


The motion of tornadoes is undoubtedly af-fected by the parent convective storms thatspawn them. In turn, the parent convectivestorms are steered by the prevailing winds inwhich they are embedded (Fujita 1987). Notisand Stanford (1973) discovered that Iowa tor-nado directions of propagation were closely tiedto 500-hPa flow. The mean 500-hPa flow for the1,927 days during which tornadoes occurred inour dataset was predominantly westerly tosouthwesterly across the United States exceptfor a small portion of the north-central region(Figure 3A). Most storm systems producingtornadoes would have propagated to the east or

northeast assuming that the 500-hPa level is agood indicator for steering-level flow.

In addition, it is important to understand thefactors that determine cellular and, in particular,supercellular convective propagations sincemost tornadoes are produced by these convec-tive typologies (Moller et al. 1994; Tessendorfand Trapp 2000). Weisman and Klemp (1986)determined that supercell motion is primarilydue to the advection of the storm by the meanwind and the internal storm dynamics that formwhen convective updrafts develop in a verticallysheared environment (e.g., see Bunkers et al.2000). In many cases, supercells, including

All Tornadoes

0

300

600

900

1200

1500

1800N

NNE

NE

ENE

E

ESE

SE

SSE

S

SSW

SW

WSW

W

WNW

NW

NNW

Weak (F0-F1) Tornadoes Significant (F2+) Tornadoes Violent (F4+) Tornadoes

A

B C D

Figure 2 Radar diagrams illustrating the frequency of tornado path direction origin for (A) all, (B) weak, (C)

significant, and (D) violent tornadoes.


those producing tornadoes, have been observedmoving slower and at 201, 301, or greater to theright of the mean wind—so called, right-turn-ing supercells (Maddox 1976). In fact, the de-viant motion (in terms of mean flow) ofsupercells is a distinct characteristic of thisstorm type and specific techniques for calculat-ing supercell motion have been developed (e.g.,see Bunkers et al. 2000). We hypothesize thatthe unique motion of right-turning supercellslikely accounts for some shift in the path clima-tology to a more westerly direction. Addition-ally, there are likely a number of noncellulartornadoes in our dataset. Tessendorf and Trapp(2000) illustrated that quasi-linear convectivesystems and hurricane rain bands may account

for nearly 24 percent of all tornadoes in thecontiguous United States. It is likely that tor-nadoes spawned by hurricanes may account forsome of the more unusual directions identifiedin our study.

Finally, eighty-one tornado outbreaks3 oc-curred during the study period—an outbreakbeing defined as a day with at least five F2þ tor-nadoes and a Destruction Potential Index(Thompson and Vescio 1998) of 50 or greater.All but ten of these tornado outbreaks (i.e., 86percent) occurred within 500-hPa flow thatcontained a southwesterly component acrossthe outbreak region. Nearly 81 percent of alltornadoes in the dataset that occurred duringtornado outbreak days propagated from the

Figure 3 Mean 500-hPa

heights, winds, and anomalies

for days in which a tornado

occurred (A) during the entire

twenty-three-year period of

study, and (B) during the summer

months of June, July, and August

over the twenty-three-year

period.


southwesterly quadrant, indicating that tornadodirections of motion do appear to be closely tiedto midtropospheric flow, as suggested by Notisand Stanford (1973).

Comparing ‘‘weak’’ (F0–F1) tornadoes withsignificant (F2þ ) and extreme (F4þ ) eventsillustrates that weak tornadoes have a greaterconcentration of W path direction than moresevere tornadoes (Figures 2B–2D). Significantand extreme tornadoes are tightly clustered, withmost of these events traveling from the WSWorSW. Nevertheless, there are several examples ofstrong tornadoes that have propagated in direc-tions different from these typical path directions.For example, the F5 Jarrell, Texas, tornado thatoccurred on 27 May 1997 propagated from thenortheast toward the southwest.

To illustrate variations through the year, tor-nado path direction origin is plotted in radardiagrams for monthly pairs (Figure 4). The di-agrams are designed to illustrate dominant di-rection rather than total frequency, thereforethe frequency scale varies: the March–April andMay–June monthly pairs have greater total fre-quency of tornadoes, the January–February andSeptember–October pairs have the least fre-quency. For January–February, tornado pathsare dominantly from the SW (31.8 percent),followed by WSW (29.5 percent) and, to a lesserextent, W (17.9 percent) directions. By March–April, there is a slight shift to a greater westerlycomponent, with WSW and W accounting for31.0 percent and 21.0 percent of all tornadopaths, respectively. As spring progresses towardsummer, a shift in dominant origin direction toW is evident in the May–June diagram with Wand SW path origins accounting for 25 percenteach. For the summer months of July–August,tornado paths are primarily from the north-western quadrant rather than from the south-western quadrant. During July–August, themodal frequency is W (26.0 percent), howeverWNW (15.7 percent) and NW (7.8 percent)frequencies remain relatively high. By fall, tor-nado path origins shift back to the southwesternquadrant. For September–October, WSW(21.2 percent) and SW (20.5 percent) are mo-dal, with a wide swath of directions being evi-dent including WNW (5.3 percent) and W(17.4 percent), through SSW (11.5 percent) andS (10.6 percent). For November–December,tornado path direction origins return to a nar-rower range, similar to the January–February

case, with SW (35.2 percent) and WSW (29.2percent) dominant, followed by SSW (13.2percent) and W (11.8 percent).

The bimonthly and monthly (not shown)analyses illustrate subtle, yet distinct, seasonalshifts in the tornado path climatology. Tornadopaths propagate from a primarily southwesterlydirection during January, February, and March,then from a predominantly westerly directionfor the next six months (April–September), be-fore returning to again propagating from thesouthwesterly direction toward the end of theannual cycle. Previous research has suggestedthat the seasonal shifts in tornado occurrenceare strongly linked to upper-air synoptic-scalemeteorological patterns (Notis and Stanford1973; Bluestein and Golden 1993; Davis, Stan-meyer, and Jones 1997; Monfredo 1999; Brown2002). As previously illustrated, Notis and Stan-ford (1973) noted seasonal shifts in Iowa torna-does and suggested that direction of tornadomovement is intimately related to the 500-hPaflow. Indeed, mean 500-hPa charts for the tor-nado days in the bimonthly cases (not shown)indicate a distinct association between the500-hPa flow across the study domain and thehighest frequency path directions. For example,during June, July, and August, tornadoes fromthe N, NNW, NW, and WNWaccount for 25.3percent of all events during this period, but dur-ing the remainder of the year only 8.8 percent oftornadoes propagate from these directions.These warm-weather months are the only peri-od where the mean tornado day 500-hPa analysisis from the west-to-northwest across the majorityof the study area (Figure 3B). Johns (1982) dis-covered that during the summer months a largenumber of severe weather outbreaks, includingthose producing tornadoes, occur with west-northwest and northwest flow in the midtropo-sphere. Hence, the climatological maximum innorthwest-flow severe weather outbreaks duringthe summertime (c.f., Figure 3 in Johns 1982) islikely responsible for the distinct shift towardmore easterly and southeasterly moving torna-does during this warm-weather period.

Results: Regional Scale

The national-scale results illustrate that thereare potentially important seasonal dynamics in-volved with the analysis of tornado path direc-tion. Tornado season varies considerably by


location (region) within the country, governedby changes in synoptic-scale meteorologicalcontrols (Bluestein and Golden 1993; Davis,

Stanmeyer, and Jones 1997; Monfredo 1999;Brown 2002). Therefore, we repeated the anal-ysis of tornado path direction for six similar-

January-February Tornadoes

0

30

60

90

120

150

N

NNE

NE

ENE

E

ESE

SE

SSE

S

SSW

SW

WSW

W

WNW

NW

NNW

March-April Tornadoes

0

100

200

300

400

500

N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW

May-June Tornadoes

0

200

400

600

800

N

NNE

NE

ENE

E

ESE

SE

SSE

S

SSW

SW

WSW

W

WNW

NW

NNW

July-August Tornadoes

0

50

100

150

200

N

NNE

NE

ENE

E

ESE

SE

SSE

S

SSW

SW

WSW

W

WNW

NW

NNW

September-October Tornadoes

0

20

40

60

80

100

120

N

NNE

NE

ENE

E

ESE

SE

SSE

SSSW

SW

WSW

W

WNW

NW

NNW

November-December Tornadoes

0

50

100

150

200

250

N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW

Figure 4 Radar diagrams illustrating the frequency of tornado path direction origin for monthly pairs.


sized regions (Figure 1). The total number oftornadoes and the monthly frequency for eachof these regions are presented in Table 2.

Tornado season peaks in March, April, andMay for the southernmost regions (South Cen-tral, Southeast), nevertheless many tornadoesare still evident during January and February.Tornadoes also are experienced throughout thefall in both of these regions, with November infact being the overall peak month for tornadooccurrence in the Southeast. Very few tornadoesoccur in either region during the summermonths of July and August.

In contrast, the northernmost regions (NorthCentral, Ohio Valley–Mid Atlantic, Northeast)experience few tornadoes during the winter(December, January, February). Tornado seasonpeaks during spring and summer (May, June,July for the North Central region; April throughJuly for the Ohio Valley–Mid Atlantic region;May through July for the Northeast). An April,May, June peak is evident for the Central region,but summer and fall tornadoes are also experi-enced. Similar to the northernmost regions, theCentral region has few winter tornadoes.

In the following sections, diagrams for eachof the six regions illustrate the seasonal distri-bution of tornadoes, and radar diagrams plottornado path direction for monthly pairs.

South Central Region

Spring is the peak tornado season in the SouthCentral region (Figure 5, top left). For March–April, tornado paths are dominantly from theW, WSW, and SW directions (Figure 5), whichis quite similar to the national scale for thesemonths illustrated previously in Figure 4. Othermonthly pairs illustrated in Figure 5 do not varymuch from the March–April case. However, thepath origin directions for September–Octoberindicate a wide distribution among directions inthe southwest quadrant. This relatively broad

distribution could be an artifact of the weaktropospheric flow that occurs across this regionbefore the strong westerlies return duringthe winter season. Supercell motion, when themean wind is relatively weak, has been observedto deviate by greater than 301 to the right ofthe mean wind (Davies 1998; Bunkers et al.2000), and that deviant motion might explainthis unique distribution. Moreover, tornadoesgenerated by tropical storms or hurricanescould account for the increase in tornadoeswith a southerly component during thisperiod. Path direction origins are from a nar-rower range for the other monthly pairs, espe-cially November–December. Because summertornadoes are rare, no radar diagram is present-ed for July–August. The rarity of tornadicevents during this period is likely caused by anincreasingly capped environment in this region(Farrell and Carlson 1989), a decrease in thethermodynamic instability of the atmosphere( Johns 1982), and a general weakening of themidtropospheric westerlies leading to an overalldecrease in the deep-layer shear needed tofacilitate the growth of tornado-producing con-vection ( Johns 1982).

Southeast Region

For the Southeast region, March–April is againthe bimonthly period with the greatest numberof tornadoes (Figure 6, top left). To an evengreater extent than in the South Central region,March–April tornado paths in the Southeast aredominantly from the W, WSW, and SW direc-tions (Figure 6). This pattern is also prevalentfor the late fall and winter months of Novem-ber–December and January–February.

Fewer tornadoes occur during the summer,and tornado paths are somewhat more variablefor May–June (with a W dominance) and Sep-tember–October (with greater variation fromWNW through to SSE and SE). Some of the

Table 2 Number of tornadoes, by region, during the twenty-three-year period 1980–2002 with pathlength � 1.61 km and complete track information

Region Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total

South Central 126 52 177 298 465 154 14 20 44 119 119 100 1688

Southeast 80 110 230 223 154 95 27 37 65 56 235 64 1376

Central 7 11 87 325 460 375 122 44 38 76 32 22 1599

Ohio V/Mid Atlantic 11 4 52 78 126 162 91 40 38 37 46 1 686

Northeast 1 1 3 11 41 65 69 22 16 5 12 0 246

North Central 0 0 14 34 100 210 131 52 32 21 5 0 599

Total 225 178 563 969 1364 1061 454 215 233 314 499 187 6194


latter events were associated with tropical sys-tem-generated tornadoes rather than being themore common supercell-generated tornadoes.In addition, the relatively wide variation in pathdirections during the late warm season in theSoutheast may be attributable to waterspouts

coming onshore or from tornadoes forming inlow-shear environments on convergence linesassociated with sea breezes (Brooks, Doswell,and Kay 2003). Again, no radar diagram is pre-sented for July–August given the rarity of sum-mer tornadoes in this region.

South Central

0

50

100

150

200

250

300

350

400

450

500

J F M A M J J A S O N D

Nu

mb

er o

f T

orn

ado

es


0

10

20

30

40

50

60

70

NNNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

50

100

150

NNNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW

May-June Tornadoes

0

50

100

150

200

NNNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

10

20

30

40

NNNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

20

40

60

80

100

NNNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW

Figure 5 Seasonal distribution of tornadoes (top left) and radar diagrams illustrating the frequency of

tornado path direction origin for monthly pairs, for the South Central region.


Central Region

The peak months for tornado events in the Cen-tral Region are April, May, and June (Figure 7, topleft). Tornado paths for March–April are againdominantly from the southwestern quadrant in-

cluding W, WSW, and SW directions, but alsowith a significant number from the SSWand somefrom the S (Figure 7). For the smaller number oftornadoes occurring in the preceding wintermonths (not shown), tornado paths are from avery narrow range of directions dominated by SW.

Southeast

0

50

100

150

200

250


Nu

mb

er o

f T

orn

ado

es


010

2030

40

5060

NNNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

50

100

150

200N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW

May-June Tornadoes

0

20

40

60

80N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

5

10

15

20

25N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

20

40

60

80

100

120N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


tornado path direction origin for monthly pairs, for the Southeast region.


The May–June monthly pair experiences themost tornadoes for this region. The modal pathdirection is from the W, with a notable numberof events occurring in a wider swath from theNWand WNW, and from the WSW, SW, SSWthrough to S directions. For this region, July–

August summer tornadoes are more numerousthan in the South Central and Southeast re-gions, although still much smaller in numberthan the March–April and May–June time pe-riods. Path direction origins are notably differ-ent during the summer months of July and

Central

050100150200250300350400450500


Nu

mb

er o

f T

orn

ado

es


020406080

100120140

NNNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW

May-June Tornadoes

0

50

100

150

200

250N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

10

20

30

40

50N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

51015202530

NNNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

5

10

15

20N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


tornado path direction origin for monthly pairs, for the Central region.


August with W, NW, and even a N spike in thenumber of events. The unique shift to morewesterly and northwesterly path origins duringthese summer months is likely due to the cli-matological maximum in northwest flow out-

breaks across the region during this period (see,e.g., Figures 4, 7, and 8 in Johns 1982). By fall,there is a dramatic shift back to tornado pathsfrom the WSW through S directions. Similarly,the mean midtropospheric flow during this time

Ohio Valley - Mid Atlantic

0

20

40

60

80

100

120

140

160

180


Nu

mb

er o

f T

orn

ado

es


0

10

20

30

40

50N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW

May-June Tornadoes

0

20

40

60

80

100N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

10

20

30

40

50N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

5

10

15

20

25N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

5

10

15N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


tornado path direction origin for monthly pairs, for the Ohio Valley–Mid Atlantic region.


period for tornado days in the dataset shifts backpredominantly from the southwest, indicatingthat this shift in the path direction climatologyis governed primarily by the upper-level steer-ing flow. Finally, although November–Decem-ber tornadoes are rare, they primarily have pathorigins from the WSW and SW in this region.

Ohio Valley–Mid Atlantic Region

Tornado events for the Ohio Valley–Mid At-lantic region are most numerous during thespring and early summer months of April, May,June, and July (Figure 8, top left). Notablenumbers of tornadoes also are evident for themonths of August through November, but win-ter tornadoes are rare.

Path direction origins illustrated in the radardiagram sequence indicate a gradual shift from theMarch–April case dominated by the Wand WSWdirections to more westerly to northwesterly or-igin directions in July and August (Figure 8).Similar to the Central region, the increase in W

and WNW directions during July and August islikely the result of an increase in northwest flowsevere weather outbreaks that tend to occur acrossthe region during these two months (e.g., seeFigures 7 and 8 in Johns 1982). By fall, pathdirections shift back to the southwestern quadrantfor both the September–October and November–December radar diagrams (the latter diagramreally represents November since there is onlyone December event). Tornadoes are also quiterare during the winter months of January andFebruary.

Northeast Region

Illustrative of the climatological return of low-level moisture and low static stability to theNortheast during late spring and summer, tor-nado events for this region are most numerousduring the months of May, June, and July (Fig-ure 9, top left). Throughout May and June in theNortheast, tornado paths originate from pri-marily the W. During July and August, the tor-

Northeast

0

10

20

30

40

50

60

70

80


Nu

mb

er o

f T

orn

ado

es

May-June Tornadoes

0

10

20

30

40

50N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

5

10

15

20

25

30N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


tornado path direction origin for monthly pairs, for the Northeast region.


nado path direction shifts, with the modal originbecoming WSW. Nevertheless, a large numberof Northeast tornadoes arrive from the north-west quadrant of motion during the summer.Tornadoes during the remainder of the year inthis region are rare.

North Central Region

Tornadoes are very rare during the winter half ofthe year for the North Central region (Figure 10,top left). Only two monthly pairs (May–June,July–August) have a sizeable number of torna-

North Central

0

50

100

150

200

250


Nu

mb

er o

f T

orn

ado

es


0

5

10

15N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW

May-June Tornadoes

0

20

40

60

80N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

10

20

30

40

50

60N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


0

5

10

15

20N

NNE

NE

ENE

E

ESE

SE

SSES

SSW

SW

WSW

W

WNW

NW

NNW


tornado path direction origin for monthly pairs, for the North Central region.


does. Nevertheless, four radar diagrams are pre-sented in order to illustrate changes in tornadopath direction for this region (Figure 10).

The small numbers of events that occur dur-ing March–April have paths from the SSW, SW,WSW, and, to a lesser extent, W. For May–June,when there are far more tornadoes, origins shifta bit and are primarily from the W and WSW.This shift to a more westerly and northwesterlydirection continues through July and August,where W and WNW are the predominant pathorigin directions. The modal travel directionfor these summer events is from the WNW,with paths ranging from the NNW and NWthrough to W, WSW, SW, and SSW. Unique tothis region, the primary path origin for the six-month period May–October is from the W. Thepredominance of westerly midtroposphericflow (Figure 3) and the influence of northwest-erly flow severe weather outbreaks are the prob-able reasons for this unique distribution.

Conclusions

Although forecasters and atmospheric scientistsare aware of the generalization that tornadoestypically propagate from the southwest quad-rant, no study has comprehensively examinedthe characteristics of U.S. tornado path direc-tions. Such research has important implicationsfor hazard mitigation and engineering studies.For example, knowledge about tornado pathdirection can be useful when deciphering thesafest location within a structure.

Results from this study illustrate that thecommon perception that most U.S. tornadoestravel from the southwest quadrant of directionsis generally valid. For the thirty-seven-state re-gion, tornadoes moving from the west-south-west toward the east-northeast were the mostfrequent, with tornadoes traveling from thesouthwest and west directions also having highfrequencies. Even on this national scale, subtleseasonal shifts in tornado path direction are ev-ident. In late spring (May–June), west and west-southwest are the modal origin directions fortornado paths. By the summer months ( July–August), when tornadoes are far less commonnationally, path directions are dominantly fromthe west, west-northwest, and northwest direc-tions. The climatological maximum in north-west flow severe weather outbreaks ( Johns1982) and a shift in the mean tornado day trop-

ospheric flow to the west-to-northwest acrossthe study area during summer may be the pri-mary contributing factors leading to this ob-served shift in tornado path directions.

Regionally, important variations occur in tor-nado path direction origins. In the North Cen-tral and Northeast regions, most tornadoesoccur in late spring (May–June) and summer( July–August). Tornadoes during these month-ly pairs travel dominantly from the west andwest-southwest during May and June, and pri-marily from the northwest quadrant of direc-tions during July and August. Additionally,although tornado path direction is most com-monly from the southwestern quadrant for boththe Central and Ohio Valley–Mid Atlantic re-gions, variations do occur. The months of April,May, and June represent peak tornado seasonfor these regions. By May–June, the modal di-rection for tornado path origin is west or west-southwest, similar to the case for the NorthCentral and Northeast regions. Although sum-mer tornadoes are far less frequent in the Cen-tral and Ohio Valley–Mid Atlantic regions, pathdirections are dominantly from the west and thenorthwest quadrant of directions when they dooccur. For the remaining seasons in these re-gions, the commonly perceived southwestquadrant of directions dominates.

For the southernmost regions (South Centraland Southeast), spring and winter tornadoesdominate. Summer tornadoes are very rare.Tornado path direction, as expected, is predom-inantly from the southwest quadrant; thereforethe common perception that tornadoes travelfrom the southwest quadrant of directions ismost valid for these southern regions.

Seasonal shifts in tornado occurrence arestrongly linked to upper tropospheric synoptic-scale meteorological patterns (Bluestein andGolden 1993; Davis, Stanmeyer, and Jones1997; Monfredo 1999; Brown 2002). Notisand Stanford (1973) demonstrated that tornadopath direction is closely linked to the 500-hPalevel flow pattern. The results from our studysupport Notis and Stanford’s conclusions andillustrate further that the midtropospheric flowis the primary control on the overall tornadodirection climatology. In addition, more than 23percent of tornadoes in the dataset were asso-ciated with one of eighty-one tornado out-breaks. In a large majority of these outbreaks,the steering-level flow was from the southwest.


In fact, 90 percent of all tornadoes that occurredduring these outbreaks propagated from theSSW, SW, WSW, or W, indicating that themidlevel tropospheric flow is a controlling fac-tor in tornado path direction in these scenarios.Finally, right-turning supercells and their asso-ciated tornadoes in the climatology are specu-lated to have contributed to a greater easterlypropagation of direction of motion. However,until a climatology of supercells and their var-ious typologies is developed, such hypotheseswill remain unsupported.

Additional study, incorporating synoptic cli-matological methods, could be useful in iden-tifying the progression of atmospheric featuresthat lead to the character and seasonality iden-tified in the climatological distributions shownin this research. Such research would be usefulin confirming the potential consistency of tor-nado path direction in summer versus othertimes of the year.

This study concludes that the common per-ception that tornadoes travel from the south-west quadrant of directions (especially SW,WSW, and W) is generally valid for the south-ern United States, and also valid for other partsof the country especially during spring, fall, andwinter. However, by late spring, a more westerlycomponent often dominates tornado path di-rection origin. Where summer tornadoes occurin significant numbers, a westerly to north-westerly directional origin is most prevalent.Since most tornadoes in the North Central re-gion occur during late spring and summer, tor-nadoes in this area often travel from thenorthwest quadrant of directions (includingwest). It is essential that these seasonal and re-gional variations in tornado path directions beincorporated into future engineering and haz-ard studies to further improve tornado mitiga-tion efforts.’

Notes

1 Available online at http://www4.ncdc.noaa.gov/cgi-win/wwcgi.dll?wwEvent�Storms.

2 SeverePlot v2 is freely available online at http://www.spc.noaa.gov/software/svrplot2/.

3 These tornado outbreak days have been identified byJohn Hart of the NWS’s Storm Prediction Center.The list is available online at http://www.crh.noaa.gov/mkx/climate/torout.htm.

Literature Cited

American Meteorological Society (AMS). 2000. Pol-icy statement: Tornado preparedness and safety.Bulletin of the American Meteorological Society81:1061–65.

Beebe, R. G. 1956. Tornado composite charts. Month-ly Weather Review 84:127–42.

Bentley, M. L., and J. A. Sparks. 2003. A 15 yrclimatology of derecho-producing mesoscale con-vective systems over the central and eastern UnitedStates. Climate Research 24:129–39.

Bluestein, H. B., and J. H. Golden. 1993. A review oftornado observations. In The tornado: Its structure,dynamics, prediction, and hazards, ed. C. Church, D.Burgess, C. Doswell, and R. Davies-Jones, 319–52.Geophysical Monograph 79. Washington, DC:American Geophysical Union.

Boruff, B. J., J. A. Easoz, S. D. Jones, H. R. Landry,J. D. Mitchem, and S. L. Cutter. 2003. Tornadohazards in the United States. Climate Research24:103–17.

Brooks, H. E. 2004. On the relationship of tornadopath length and width to intensity. Weather andForecasting 19:310–19.

Brooks, H. E., C. A. Doswell, and M. P. Kay. 2003.Climatological estimates of local daily tornadoprobability for the United States. Weather andForecasting 18:626–40.

Brown, M. E. 2002. The spatial, temporal, and ther-modynamic characteristics of southern-AtlanticUnited States tornado events. Physical Geography23:401–17.

Bunkers, M. J., B. A. Klimowski, J. W. Zeitler, R. L.Thompson, and M. L. Weismanm. 2000. Predict-ing supercell motion using a new hodograph tech-nique. Weather and Forecasting 15:61–79.

Burgess, D. W., R. J. Donaldson Jr., and P. R. Des-rochers. 1993. Tornado detection and warning byradar. In The tornado: Its structure, dynamics, predic-tion, and hazards, ed. C. Church, D. Burgess, C.Doswell, and R. Davies-Jones, 203–21. Geophysi-cal Monograph 79. Washington, DC: AmericanGeophysical Union.

Coniglio, M. C., and D. J. Stensrud. 2004. Interpret-ing the climatology of derechos. Weather and Fore-casting 19:595–605.

Davies, J. M. 1998. On supercell motion in weakerwind environments. Preprints, 19th Conference onSevere Local Storms, 685–88. Minneapolis, MN:American Meteorological Society.

Davis, R. E., T. M. Stanmeyer, and G. V. Jones.1997. A synoptic climatology of tornadoes in Vir-ginia. Physical Geography 18:383–407.

Doswell, C. A., III, and D. W. Burgess. 1988. On someissues of United States tornado climatology. Month-ly Weather Review 116:495–501.


Doswell, C. A. III, A. R. Moller, and H. E. Brooks.1999. Storm spotting and public awareness sincethe first tornado forecasts of 1948. Weather andForecasting 14:544–57.

Eagleman, J. R. 1967. Tornado damage pattern inTopeka, Kansas, June 8, 1966. Monthly Weather Re-view 95:370–74.

———. 1990. Severe and unusual weather, 2nd ed. Le-nexa, KS: Trimedia Publishing.

Eagleman, J. R., V. U. Muirhead, and N. Williams.1975. Thunderstorms, tornadoes, and building damage.Lexington, MA: D. C. Heath.

Farrell, R. J., and T. N. Carlson. 1989. Evidence forthe role of the lid and underunning in an outbreak oftornadic thunderstorms. Monthly Weather Review117:857–71.

FEMA (Federal Emegency Management Associa-tion). 2004. Surviving the storm, tornadoes. Avail-able online at: http://www.fema.gov/pdf/hazards/survtorn.pdf (last accessed 13 October 2005).

Finley, J. P. 1887. Tornadoes. New York: The Insur-ance Monitor. Available online at: http://hsci.cas.ou.edu/exhibits/exhibit.php?exbid=30&exbpg (lastaccessed 13 October 2005).

Flora, S. D. 1954. Tornadoes of the United States.Norman: University of Oklahoma Press.

Friday, E. W. 1994. The modernization and associat-ed restructuring of the National Weather Service:An overview. Bulletin of the American MeteorologicalSociety 75:43–52.

Fujita, T. T. 1987. U.S. tornadoes Part 1 70-year sta-tistics. Satellite and Mesometeorology ResearchProject (SMRP) Research Paper No. 218, Univer-sity of Chicago, 122 pp.

Galway, J. G., and A. Pearson. 1981. Winter tornadooutbreaks. Monthly Weather Review 109:1072–80.

Grazulis, T. P. 1993. Significant tornadoes 1680–1991.St. Johnsbury, VT: Environmental Films.

———. 2001. The tornado: Nature’s ultimate windstorm.Norman: University of Oklahoma Press.

Grazulis, T. P., J. T. Schaefer, and R. F. Abbey Jr. 1993.Advances in tornado climatology, hazards, and riskassessment since tornado symposium II. In The tor-nado: Its structure, dynamics, prediction, and hazards,ed. C. Church, D. Burgess, C. Doswell, and R. Da-vies-Jones, 408–26. Geophysical Monograph 79.Washington, DC: American Geophysical Union.

Hart, J. 1993. SVRPLOT: A new method of access-ing and manipulating the NSSFC severe weatherdatabase. Preprints, 17th Conference on Severe LocalStorms, 40–41. St. Louis: American Meteoro-logical Society.

Johns, R. H. 1982. A synoptic climatology of north-west flow severe weather outbreaks. Part I: Natureand significance. Monthly Weather Review 112:449–64.

Johns, R. H., and J. S. Evans. 2000. Comments on ‘‘Aclimatology of derecho-producing mesoscale con-

vective systems in the central and eastern UnitedStates, 1986–95. Part I: Temporal and spatial dis-tribution.’’ Bulletin of the American MeteorologicalSociety 81:1049–54.

Kelly, D. L., J. T. Schaefer, R. P. McNulty, C. A.Doswell III, and R. F. Abbey Jr. 1978. An augment-ed tornado climatology. Monthly Weather Review106:1172–83.

Maddox, R. A. 1976. An evaluation of tornado prox-imity wind and stability data. Monthly Weather Re-view 104:133–42.

Miller, R. C. 1972. Notes on the analysis and severe-stormforecasting procedures of the Air Force Global WeatherCentral. Air Weather Service Tech. Rept. 200 (Rev.),Air Weather Service, Scott Air Force Base, IL, 190pp.

Moller, A. R., C. A. Doswell III, M. P. Foster, and G.R. Woodall. 1994. The operational recognition ofsupercell thunderstorm environments and stormstructures. Weather and Forecasting 9:327–47.

Monfredo, W. 1999. Relationships between phases ofthe El Nino–southern oscillation and character ofthe tornado season in the south-central UnitedStates. Physical Geography 20:413–21.

Notis, C., and J. L. Stanford. 1973. The contrastingsynoptic and physical character of northeast andsoutheast advancing tornadoes in Iowa. Journal ofApplied Meteorology 12:1163–73.

NWS (National Weather Service). 2004. Thunder-storm, tornadoes, lightning: A preparedness guide.Available online at: http://www.nws.noaa.gov/om/brochures/ttI.pdf (last accessed 13 October 2005).

Schaefer, J. T., D. L. Kelly, and R. F. Abbey. 1986. Aminimum assumption tornado-hazard probabilitymodel. Journal of Climate and Applied Meteorology25:1934–45.

Schaefer, J. T., D. L. Kelly, C. A. Doswell III,J. G. Galway, R. J. Williams, R. P. McNulty, L. R.Lemon, and B. D. Lambert. 1980. Tornadoes.When, where, how often: A study of 29 years oftornado data. Weatherwise 33:52–9.

Schaefer, J. T., R. L. Livingston, F. P. Ostby, and P. W.Leftwich. 1993. The stability of climatological tor-nado data. In The tornado: Its structure, dynamics,prediction, and hazards, ed. C. Church, D. Burgess,C. Doswell, and R. Davies-Jones, 459–66. Geo-physical Monograph 79. Washington, DC:American Geophysical Union.

Schaefer, J. T., S. J. Weiss, and J. J. Levit. 2003.The frequency of severe thunderstorm winds overthe contiguous United States. 11th InternationalConference On Wind Engineering, Lubbock, TX,8 pp.

Tessendorf, S. A., and R. J. Trapp. 2000. On the cli-matological distribution of tornadoes within quasi-linear convective systems. Preprints, 20th Conferenceon Severe Local Storms, Orlando, FL, American Me-teorological Society, 134–73.


Thompson, R. L., and M. D. Vescio. 1998. The De-struction Potential Index—A method for compar-ing tornado days. Preprints, 19th Conference on SevereLocal Storms, Minneapolis, MN. American Meteor-ological Society, 280–82.

Tornado Project Online. 1999. Safety. Available on-line at: http://www.tornadoproject.com/safety/safety.htm (last accessed 13 October 2005).

U.S. Department of Commerce. 1953. Tornado occur-rences in the United States. Technical paper no. 20.Washington, DC: U.S. Government Printing Of-fice.

Weisman, M. L., and J. B. Klemp. 1986. Character-istics of convective storms. In Mesoscale meteorologyand forecasting, ed. P. S. Ray, 331–58. Boston: Amer-ican Meteorological Society.

Wilson, J. W., and S. A. Changnon Jr. 1971. Illinoistornadoes. Circular 103. Champaign: Illinois StateWater Survey.

PHILIP W. SUCKLING is Professor and Chair ofthe Department of Geography at Texas State Univer-sity–San Marcos, San Marcos, TX 78666. E-mail:[email protected]. His research interests involve cli-matology and natural hazards.

WALKER S. ASHLEY is Assistant Professor in theMeteorology Program, Department of Geography atNorthern Illinois University, DeKalb, IL 60115.E-mail: [email protected]. His research interests involvesynoptic and mesoscale climatology/meteorology,especially with respect to severe weather.


spatial and temporal characteristics of tornado path directionchubasco.niu.edu/pubs/suckling and...

Documents