Center Logo

Natural Hazards Research and Applications Information Center
University of Colorado
482 UCB
Boulder, CO 80309-0482

Quick Response Report #161

An Analysis of the September 20, 2002, Indianapolis Tornado:
Public Response to a Tornado Warning
and Damage Assessment Difficulties

Jamie D. Mitchem

Hazards Research Lab
Department of Geography
University of South Carolina
Columbia, SC 29208



Return to the Hazards Center Home Page

Return to the Quick Response Report Index

This QR report is available in its entirety only in PDF format (11.7 Mbs). Below you
will find the body of the report without its extensive appendices.

This material is based upon work supported by the National Science Foundation under Grant No. CMS-0080977. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation or the Natural Hazards Research and Applications Information Center.

Citation: Jamie D. Mitchem. 2003. An Analysis of the September 20, 2002, Indianapolis Tornado: Public Response to a Tornado Warning and Damage Assessment Difficulties. Quick Response Research Report #161. Boulder, Colorado: Natural Hazards Research and Applications Information Center, University of Colorado. URL:


This research was funded by a Quick Response Program grant from the Natural Hazards Research and Applications Information Center at the University of Colorado at Boulder. Dr. Susan Cutter and Dr. Kirstin Dow from the University of South Carolina, as well as Dr. Arleen Hill from Bucknell University, assisted with the development of the survey forms. Their help is greatly appreciated. I would also like to thank David Lee for allowing me to conduct surveys at Washington Square Mall in Indianapolis, and I appreciate the assistance from Mike Shartran of the Indianapolis NWSFO for graciously providing me with the text of the warning messages. Finally, the hospitality and cooperation provided by the people of Marion County, Indiana during this difficult time is appreciated. Without their help, many questions about tornado safety would remain unanswered.


An anomalous date in the tornado record is September 20, which has, "the largest count for any day between the early summer and late spring" (NSSFC Staff, 1980: 54). The notorious Hurricane Beulah in Texas in 1967, which was a prolific producer of tornadoes, is responsible for many of the tornado reports on this date. Coincidentally, three tornadoes occurred in Indiana on September 20, 2002, with one hitting southern and eastern portions of Indianapolis in Marion County, Indiana (Appendix A: Fig. 1). The Indianapolis tornado had the longest track (a remarkable 112 miles) during this event, and it will be the subject of this study. Tornado paths longer than 100 miles are rare, with an average of almost one each year (Twisdale, 1982; Wilson & Morgan, 1971). This tornado proved to have the second longest track in Indiana's history. Pryor and Kurzhal (1997) claim that the more intense a tornado is, the more likely it is to have a long path length. The Indianapolis tornado also happened to be the most damaging tornado (F3; wind speeds of ~158-206 mph) during this outbreak.

This event provided the opportunity to study the effects of a tornado on an urbanized area. It is possible that the Fujita Scale does not describe urban types of damage effectively, which could lead to difficulties in the Fujita Scale damage assessment process. Also, urban dwellers might be faced with unique situations that affect their responses such as being in their cars, being at work, eating at restaurants, walking on the streets, or living in apartments. Since the tornado's path was so long, the focus of this study is only on the portion of the tornado's path in Marion County, Indiana (Appendix A: Fig. 2), which includes the city of Indianapolis.

The long-track tornado that is being studied caused enormous damages, affecting at least 313 buildings in Marion County, with a preliminary damage total of $36,605,800 for Marion County alone (Tan & Ellis, 2002). However, no fatalities or serious injuries were reported from this tornado, and only about 50 minor injuries were reported (Schneider, 2002). Early warnings, rapid dissemination, and an overall prepared citizenry all seem to have contributed to the "miraculously" low number of casualties.


Human Environment

Social factors can contribute as much or more to tornado vulnerability as the natural hazard occurrence (Blaikie et al., 1994). Therefore, risk is defined not only by the threat to people, but also by the product of cultural and social experiences (Kasperson, 1992). This is why it is important to understand the human environment as well as the physical environment when dealing with hazards. Boruff et al. (2003) illustrate how social vulnerability can amplify or decrease tornado vulnerability in the U.S. both spatially and temporally.

The total population of Marion County, Indiana in 2000 was 860,454 (USCB, 2002a). It has the highest population of any county in Indiana. The population density was 2172 persons per square mile in 2000. The city of Indianapolis is the 12th largest city in the U.S. based on population, but its population density ranked 183rd (USCB, 2002b). Population density is especially important when discussing tornado impacts because tornadoes typically affect a narrow swath rather than the entire county, and a high population density means more people are likely to be affected.

The median age of Marion County residents is 33.6 years old (USCB, 2000a). Only 11% of the population in Marion County is 65 years of age or older, but 26% of the county's residents are under the age of 18. Seven percent of the population of Marion County consists of small children under the age of five. Schmidlin (1993) has found that the very young (<5) and the elderly (>65) are at highest risk for death in tornadoes. Also, there are 157,908 persons over the age of 5 with disabilities in Marion County. These residents are particularly vulnerable to tornadoes because it is frequently more difficult for them to receive warnings and to respond to them.

There are slightly fewer males (48%) than females (52%) in Marion County, Indiana. The age and gender composition of Marion County is consistent with the state average for Indiana. However, the racial composition of the population in Marion County differs from the rest of Indiana. The black population is much higher in Marion County (24%) than it is for the entire state of Indiana (8%). The white population percentage for Marion County (71%) is thus lower than the average for the state of Indiana (88%). The Hispanic population is the third most common racial group in Marion County according to the 2000 Census, but they account for only 4% of the population in both Marion County and the state of Indiana. It has been suggested that both gender and ethnicity contribute to vulnerability to tornadoes.

There were 352,164 households in Marion County, Indiana, in the year 2000, with an average of 2.39 persons per household. Fifty-three percent of Marion County residents who are 15 years old or older have either never been married, are separated, are divorced, or are widowed (USCB, 2000b). Being unmarried or divorced has been shown to increase vulnerability to tornadoes, possibly due to where these persons frequently live (e.g. apartments, mobile homes) (Schmidlin et al., 1998).

Other factors that can contribute to vulnerability are income and education levels, which generally affect the type of home one can afford and possibly have an influence on preparedness levels. Illiteracy can make people particularly vulnerable (Pearce, 1994). Generally, a low education level is related to a low income level and high vulnerability. The median household money income in 1999 for Marion County residents was $40,421, and the per capita money income was $21,789. Unfortunately, 11% of the population in Marion County lives below the poverty level. This is just slightly higher than the state average. 82% of Marion County residents over the age of 25 are high school graduates, and of the population over 25 years old has a bachelor's degree or higher education level. This level of college-educated residents is higher than the state average of 19%.

The population in Marion County is fairly mobile with only 47% of persons over the age of 5 living in the same house in 2000 as they were living in back in 1995. This is lower than the state average of 55%. 19% of Marion County residents lived in another county, state, or country in 1995 (USCB, 2000b). This affects preparedness because residents that are new to an area may not have the level of awareness that long-term residents have. 5% of Marion County residents were born outside of the United States, and 7% of residents over the age of 5 speak a language other than English at home. This affects preparedness levels and warning dissemination. Warnings are usually only broadcasted in English. These numbers are slightly higher than the state average.

Finally, the mean travel time to work for residents over the age of 16 is 23 minutes. Only 10% of households do not own a vehicle in Marion County (USCB, 2000b). Being in a vehicle reduces the likelihood of receiving warnings, and vehicles are not the best shelter choice during tornadoes. Unfortunately, automobile fatalities from tornadoes can occur in high numbers, particularly during rush hour times when traffic jams are more common as was true in Wichita Falls, Texas on April 10, 1979 (Glass et al., 1980).

Physical Environment

Marion County is located at the heart of the Indianapolis Metropolitan Statistical Area, and it contains the city of Indianapolis. Marion County, Indiana has an area of 396 square miles. The city of Indianapolis has the 9th largest land area of any city with a population over 100,000 in the U.S. (USCB, 2002b). Berz et al. (2001) claim that the potential for catastrophes increases greatly as economic activity is concentrated in large cities and industrial areas. Urban sprawl increases the probability that a city will be hit by a tornado because it increases the area of the city. Unfortunately, Indianapolis does lie in an area of frequent tornado occurrence. Pryor and Kurzhal (1997) discuss the tornado climatology of Indiana in detail.

In 2000, there were 387,183 housing units in Marion County. However, 32% of these housing units were in multi-unit structures, a much higher percentage than the state average of 19%. It is difficult to modify a multi-unit structure because more than one household is affected, and the occupants are often renters. The homeownership rate for Marion County is 59%, which is lower than the state average of 71%. This is probably due to the large number of apartments and renters in Marion County. People who rent their residences are also more vulnerable than homeowners because renters carry less insurance, and their residences are often not well constructed or maintained. Apartment managers often do not provide safe shelters for their tenants either. In addition, renters are frequently over-represented by the poor, the elderly, and the young (Pearce, 1994). 56% of Marion County's structures were built before 1970 (USCB, 2000b). Fortunately, mobile homes make up only 2% of the housing stock in Marion County (USCB, 2000b). Mobile homes are not safe during tornadic storms. However, the median value of owner-occupied housing units was $99,000, which was $4700 higher than the state average. Therefore, it is not difficult for a tornado to cause significant property damage values in Marion County.

Research Questions


After driving to Indiana on September 23, 2002, the principal investigator began documenting the tornado's path with photographs and notes about the resultant damage. It is important to photograph damage before it can all be cleaned-up to get a realistic impression of what occurred. The pictures taken for this study were compared to index pictures of tornado damage patterns (Appendix A: Figure 3) taken by the tornado pioneer, Ted Fujita (Fujita, 1971) in order to estimate the Fujita Scale rating of the tornado at each photograph location. This information can then be mapped to determine the tornado's path and strength variations throughout Marion County (Appendix A: Figs. 4, 5, and 6).

In addition to damage assessment, this study determines if and how people heard about the tornado warnings and how they responded to the warnings. Therefore, public surveys were conducted at two different sites, one a major retail center on the south side of Indianapolis very close to the tornado's path (Appendix A: Fig. 4) and the other a mall on the east side of Indianapolis near the tornado's track (Appendix A: Fig. 5). On September 25, the survey was conducted at the south side site, and there were 29 participants. On September 26 and 27, surveys were conducted at the east side site with 48 and 25 participants, respectively. A copy of the survey used in this study is attached (Appendix D). The survey results were entered into a spreadsheet for analysis, and the X2 statistic was used to test for significance when appropriate. Again, quick response is necessary to ensure that respondents remember the event clearly, and the public is generally more eager to answer questions soon after a disaster

After returning to Columbia on September 29, the 6 o'clock news reports from the NBC, CBS, ABC, and FOX affiliates in Indianapolis for September 20, 21, and 22 were ordered to study their coverage and obtain more information about the tornadoes. The National Weather Service Office in Indianapolis was also contacted, and they provided the original wording and times of the tornado warnings issued on September 20 (Appendix A: Figs. 7 and 8).

Sample Size and Characteristics

The public surveys in Indianapolis included 102 people. The majority of the respondents were female (64%), which is probably a result of the demographics of the customers at the survey sites at the time that the surveys were conducted. It is possible that men were more likely to be working or cleaning up debris at the time of the surveys. It was easily apparent that more women than men were entering the major retail center and the mall in Indianapolis during the daylight hours while the surveys were being conducted.

Also, most of the respondents were fairly young. For example, 67% of the respondents were between the ages of 18 and 47. Only 1/3 of respondents were over the age of 47 (Appendix A: Fig. 9). There were slightly more 18-27 year old respondents who were male and slightly more 28-57 year old respondents who were female (Appendix B). Again, this is probably a result of the demographic construction of the mall and retail center's customers.

The race of most respondents was white (75%), but almost (24%) of respondents were black. The racial gap was slightly larger for females than males (Appendix B), but the difference is not statistically significant, X2 (1,N=101)=0.34, p>0.10. Black respondents were generally younger than white respondents in this case (Appendix B). Again the demographic profile of the areas surveyed is probably responsible for these results, and the proportions agree with the statistics found for Marion County by the U.S. Census Bureau.

Almost half (49%) of the respondents indicated that they were very close (within 2 miles, but not hit) to the tornado (Appendix A: Figure 10). This makes sense because the survey sites were very close to the tornado's path, and most people probably lived near the survey sites. Only 10% of respondents indicated that the tornado hit their location. Also, only 9% of respondents were more than 10 miles away. Due to the proximity of the tornado, many respondents were very enthusiastic about taking the survey, and this is an opportunity to get vital information from the public that should not be missed.


Fujita Scale Assessment

Narrative Account of Tornado Track

The tornado damage pictures and accompanying notes indicate a continuous damage path across southern and eastern Marion County, with damage ranging from F1 to F3 intensity on the Fujita Scale (Appendix A: Fig. 2). This agrees overall with the National Weather Service Assessment (Appendix A: Fig. 1), but the damage assessment in this study indicates smaller-scale variations in F-scale intensity and the shape of the tornado's swath of damage. The tornado's track in Marion County is divided into 3 segments from southwest to northeast as it moved through the county, and the damage path is described in Appendix C.

Damage Assessment Difficulties

Using the descriptions of characteristic damage types and the index pictures, it is fairly straightforward to rate tornado damage to houses on the Fujita Scale (Appendix A: Fig. 28). Damage to apartments is not very difficult to classify either (Appendix A: Fig. 29). However, it is possible that apartment buildings are more susceptible or less likely to be damaged by tornadoes than houses due to their size and construction as well as wind flow patterns around them. Wind engineering studies could address this issue.

The same can be said for businesses and factories (Appendix A: Fig. 30). The large-span roofs on factories probably make them more susceptible to wind damage in a similar manner as churches and gymnasiums. The vacant Pier 1 Imports store that was destroyed by this tornado was probably more susceptible to damage due to its type of construction and the glass walls that are common in these types of stores.

Damage to vehicles was more difficult to classify (Appendix A: Fig. 31). Vehicles are not typically damaged by tornadoes less than F1 in strength unless a limb falls on them. At F1 intensity, some cars are blown around on roads. When F2 winds are encountered some vehicles are lifted off the ground. Debris also poses a major hazard to vehicles by crushing them or breaking the windows (Appendix A: Fig. 31). Cars can be lifted and thrown by F3 tornadoes. This results in tremendous damages and terrifying situations. Debris can flatten cars and blow out all windows (Appendix A: Fig. 31). During this tornado a minivan with a mother and teenage daughter inside was rolled over several times through a parking lot at a shopping center while they prayed for their lives. Fortunately, they had on their seat belts and survived, and they were thankful to be alive. Another amazing story broadcast on local television in Indianapolis tells of a mother and her toddler son who were driving home ahead of the tornado. The mother realized the tornado was about to strike them, and she pulled into her church parking lot. She got in the back of the minivan and lay on top of her son. The tornado picked up the minivan and tossed it through the air for three blocks. Witnesses described seeing the minivan flying over several houses. Miraculously, the minivan landed on its wheels, and neither mother nor son was physically harmed.

Damage to trees was somewhat challenging to classify. In this case, there were usually buildings nearby to assess the damage, but in rural areas that are heavily wooded damage assessment could prove difficult. Damage below F3 is especially difficult to rate (Appendix A: Fig. 32) because it depends on so many variables (e.g. root depth, plant health, height, leaf mass, wind speed, and thickness of forest). Ecological models and biologists could improve damage classification with more research.

Many have noted problems with the Fujita Scale, and a symposium addressing its problems was held at the 2003 American Meteorological Society's Annual Meeting in Long Beach, CA. The Fujita Scale does not describe how to classify damage to many structures that are commonly found such as signs (Appendix A: Fig. 33), power lines, fences, small walls and borders, or streetlights (Appendix A: Fig. 34). It is also challenging to try to classify tornadoes based on odd debris that is found some distance from where it was originally located (Appendix A: Fig. 33). Of course if no damage indicators are present (e.g. corn fields), it is extremely difficult to determine the intensity of the tornado.

Damage assessment is often quite challenging. In urban areas, damage assessment is probably easier than usual because there are so many structures to compare to the index pictures, which are only structures. Some possible biases in urban areas are that the amount of debris created in urban areas during tornadoes could exacerbate the damage levels, and the types of buildings in urban areas (e.g. apartments, businesses, factories, restaurants, etc.) might not react to winds in the same manner as houses. Future research could improve the damage assessment process by addressing some of the problems discussed in this section.

Warning Dissemination, Shelter Decisions, and Public Awareness

The public survey results from Indianapolis indicate that most of the 102 respondents (47%) had no visual or audible cues to indicate that a tornado was nearby (Appendix A: Fig. 11). Of those that did see or hear tornado indicators, the most common cues were odd-colored or greenish clouds (31%), a loud, roaring sound (25%), or a funnel-shaped cloud (11%). Those respondents who were at a location that was struck by the tornado frequently experienced all of the visual or audible cues associated with tornadoes, except for hail or odd-colored clouds (Appendix B). They were the only group to likely have seen the funnel-shaped cloud associated with the tornado. However, to of persons in all distance groups claimed to have experienced no sensory cues of the tornado. The respondents who were not hit by the tornado frequently saw odd-colored clouds. Besides those who were hit, respondents that were very close or not close often heard the roar of the tornado, too.

Fortunately, 81% of the 102 respondents were aware of the tornado warning in this case. Those least likely to be aware of the warning were either directly hit by the tornado, or they were more than 10 miles away from the tornado's path (Appendix B). Distance from the tornado did not significantly alter the tornado warning awareness level, X2 (1,N=102)=0.37, p>0.10. A minimal racial difference in tornado warning awareness existed with white persons being slightly less likely to be aware of the warning than black persons (Appendix B), but the difference was not statistically significant, X2 (1,N=101)=0.82, p>0.10. Also, female respondents were more likely to have been aware of the tornado warning than males (Appendix B) in this case, but again the difference was not statistically significant, X2 (1,N=102)=1.24, p>0.10. Finally, middle-aged respondents (age 28-57) were more likely to be aware of the warning than retirement-age (over 57) or young adult (18-27) respondents (Appendix B). However, the relationship between age and warning awareness was not statistically significant, X2 (2,N=102)=0.41, p>0.10, either.

Unfortunately, of the 83 respondents who were aware of the warning, only 31% felt that they were really in danger. Those who did not feel that they were in danger were much more likely to continue with business as usual (Appendix B). As expected, those that felt they were in danger when they heard the warning were less than 10 miles from the tornado's track (Appendix B), but the relationship between distance and awareness was not significant, X2 (1,N=83)=1.28, p>0.10. Most of these 83 people either heard about the warning from the tornado sirens sounding (70%) or from local television scrolls and news coverage (46%; this adds up to more than 100% because more than one answer could be chosen) (Appendix A: Fig. 12). However, the respondents that were most likely to perceive the threat of the tornado (>50% response) received the warning from NWS Weather Radio or a friend or relative (Appendix B).

Though the warning lead-time was between 20 and 40 minutes for Marion County, depending on what part of the county is considered, 22% of these 83 respondents reported less than 5 minutes to respond (Appendix B). However, many of the respondents (41%) who were aware of the warning did seek shelter immediately when they became aware of the tornado warning. Unfortunately, the second most common response to the tornado warning was to continue with business as usual (27%; again more than one answer could be chosen) (Appendix A: Fig. 14). Those most likely to seek shelter were within 20 miles of the tornado, but those who were very close to the path also were most likely to continue with business as usual (Appendix B). The relationship between distance from the tornado and shelter-seeking behavior was not statistically significant, X2 (1,N=102)=1.89, p>0.10. One frightening result of the survey is that 49% of the 102 respondents did not seek shelter at any time during this tornado, and only about half of those who did seek shelter had rehearsed their plan or gone to that shelter before. Being aware of the tornado warning did raise the chances of the respondent seeking shelter (Appendix A: Fig. 15), and the difference in proportions is significant, X2 (1,N=101)=4.52, p<0.05. The almost half of respondents who did seek shelter most commonly went to a basement (19%), an interior hallway (11%), or into a large building like a mall or Wal-Mart store (8%) (Appendix A: Fig.16). Females were more likely to seek shelter than males (Appendix A: Fig. 17), and the difference in proportions is significant, X2 (1,N=102)=3.21, p<0.10. Younger respondents were more likely to seek shelter than elderly respondents with 18-57 year olds being statistically more likely to seek shelter than those age 58 or older, X2X2 (1,N=101)=0.96, p>0.10. White persons who took shelter most often went to a basement, but black persons who took shelter most often went to a hallway (Appendix B). The respondents who where at a location that was struck by the tornado usually either never took shelter or they went to a basement (Appendix B). Those who went to a basement, large building, or bedroom were most likely to have rehearsed their plan before, but those who went to someone else's house or a vehicle were not likely to have practiced their plan before (Appendix B).

Of the 19 respondents who were not aware of the tornado warning, the most common reasons for not being aware of the warning were being in a car or being at work, respectively (Appendix A: Fig. 19). Most of these people were either never aware that there was a tornado threat, or they had less than 1 minute to respond when they did see or hear the tornado (Appendix A: Fig. 20). Therefore, 68% of them never responded to the tornado threat. This illustrates the importance of disseminating the warnings widely and effectively, and it indicates that we must improve warning notification processes at businesses and industries and in automobiles.

The respondents seemed to be most satisfied with the performance of the National Weather Service (63%), rescue personnel (54%), and the media (50%) during this tornado (Appendix A: Fig. 21). These groups should be commended for saving many lives. Most people (61%) thought everyone performed adequately during this tornado. Females (and white persons) were slightly more satisfied than males (and black persons) with all groups' responses, except for the public's, during this tornado (Appendix B). All age groups, especially the middle-aged were happy with the performance of the National Weather Service during this tornado. Young persons reported being satisfied with the rescue personnel slightly more often than older persons, but older respondents reported being satisfied with the media slightly more often than young respondents (Appendix B).

The respondents who did note some problems most often criticized the performance of the public (16%), the government (11%), and the media (7%) during this tornado (Appendix A: Fig. 22). Gender did not play a noticeable role in dissatisfaction of various groups during this tornado. The public was criticized by some respondents for making bad sheltering decisions, for causing massive traffic problems, for looting, and for apathy. White respondents were dissatisfied with the public's performance slightly more than black respondents (Appendix B). The government was most often criticized for not providing adequate shelter and assistance following the tornado, for keeping people out of their damaged homes, and for some decisions local schools made during the tornado. One respondent stated that children were crouched on the floor of a second-story hallway in a two-story school building that took a direct hit from the tornado. There was roof and water damage, but fortunately the children were not harmed. School officials should be instructed that children should always seek shelter on the lowest floor of the school during a tornado, and the sheltering plan should be practiced during tornado drills. Tornado drills also provide an opportunity to educate people about tornado facts and safety information. Black respondents were dissatisfied with the government's performance more often than white respondents (Appendix B). Also, young respondents were dissatisfied with the government's performance more often than older respondents (Appendix B). Finally, the media received some criticism, from white respondents (Appendix B), for hyping the event, for causing panic, and for insensitivity to victims. However, again the majority of people were satisfied with everyone's performance during this tornado.

Of the 102 respondents, approximately 30% could not correctly identify what a tornado watch means, and 21% could not correctly identify what a tornado warning means. The majority of respondents knew what both a tornado watch and warning meant, and knowing one raised the chance of knowing the other at a significant level, X2 (1,N=102)=22.80, p<0.005. However, the most frequent mistakes were not knowing what a tornado watch meant and switching the definitions of tornado watches and warnings (Appendix B). There is not much of a gender difference in the comprehension of the meaning of a tornado watch and warning (Appendix B), and the differences are not significant, X2 (1,N=102)=0.02 and ?2 (1,N=102)=0.10, p>0.10, respectively. Both groups understood tornado warnings better than tornado watches. However, black respondents were less likely to understand the meaning of a tornado watch or warning (especially a tornado warning) than white respondents (Appendix A: Fig. 23), and both differences were significant (watch: X2 (1,N=102)=3.63, p<0.10; warning: X2 (1,N=101)=8.33, p<0.005). Alarmingly, 18-27 year olds were most likely to not understand what a tornado watch means (Appendix B). However, the difference in proportions was not quite significant, X2 (1,N=102)=2.36, p>0.10. This could mean that education efforts are decreasing or complacency is creating less awareness. Those aged 28-47 and 68-77 were most likely to know what a tornado watch means. Another frightening trend is that the youngest group (18-27) was least likely to know what a tornado warning means. 18-37 year olds were significantly less likely to know what a tornado warning meant than those age 38 or older, X2 (1,N=102)=3.68, p<0.10, (Appendix A: Fig. 24). It appears that education about tornadoes in public schools has either been recently reduced, or it is ineffective. This indicates that further public education and awareness campaigns, especially using the media, would be beneficial. These campaigns should particularly target young adults and the black population of Marion County.

22% of respondents had never been in a tornado drill, and almost the exact same number of respondents did not have a family emergency plan at home. However, having been in a tornado drill did not raise the likelihood of a person having a family emergency plan at home. Respondents over the age of 58 were least likely to have been in a tornado drill before (Appendix B), and this finding is significant, X2 (1,N=102)=13.403, p<0.005, (Appendix A: Fig. 25). However, young respondents (18-37) were least likely to have a family emergency plan at home (Appendix B), and this finding is significant also, X2 (1,N=101)=7.00, p<0.01, (Appendix A: Fig. 26). White respondents were less likely to report having been in a tornado drill at work than black respondents (Appendix B), and the difference is significant, X2 (1,N=101)=3.34, p<0.10, (Appendix A: Fig. 27). However, both groups usually had a family emergency plan at home (Appendix B), and the difference was not significant, X2 (1,N=101)=1.31, p>0.10. Again, education campaigns could target these age and racial groups to improve safety. There was not a gender difference in tornado drill participation or family emergency planning.


The Marion County, Indiana tornado of September 20, 2002 provided many lessons to improve tornado warning information and damage assessment procedures. The damage from this tornado illustrated that urban areas could have biased Fujita Scale estimates due to the types of structures present (e.g. factories, churches, restaurants, apartments) and the amount of airborne debris that is created. Certain types of damage such as damage to signs, fences, streetlights, and power lines were difficult to classify using the current Fujita Scale. Tree and vehicle damage proved difficult to rate also. This study illustrates that the Fujita Scale damage assessment is as much an art as a science, and it contains inherent subjectivity.

The public surveys conducted in Marion County indicated that many respondents (61%) who received the tornado warning did not take it seriously. The results also implied that disseminating warnings many different ways is crucial to reach the most people. Tornado sirens and local television stations proved to be extremely effective methods for disseminating the tornado warning in this case. This is an important finding because budgetary constraints are making some towns abandon future funding for tornado siren coverage. Also, the StormReady program, in which Marion County participates, appears to help reduce vulnerability to tornadoes via its requirements for entry into the program. Travelling, recreation, power outages, and working all seem to inhibit the warning dissemination process. Creative methods for reaching people in these situations should be devised. Finally, not being aware of a warning substantially decreased the time to respond to the tornado threat, which illustrates the importance of effective warning dissemination methods.

Other findings from the surveys were that 46% of Marion County residents did not seek shelter during this tornado. However, those who did seek shelter most often went to a basement or hallway as recommended during tornadoes. Warning awareness did increase shelter-seeking behavior, which further strengthens the point that warnings must be disseminated effectively, particularly since most respondents did not experience any sensory cues of the approaching tornado. Fortunately, most respondents did report having been in a tornado drill and having a family emergency plan at home.

Some recommendations to improve safety in Marion County during tornadoes are to emphasize the importance of seeking shelter, especially focusing on males and those ages 58 or older. Also, tornado drills need to include white persons and those age 58 and older more often in Marion County. Respondents who were age 18-37 should develop family emergency plans. Education campaigns in Marion County should target black residents and residents ages 18-37. These actions could ameliorate some of the problems that were displayed by the survey results after the September 20, 2002 tornado.


Berz, G., W. Kron, T. Loster, E. Rauch, J. Schimetchek, J. Schmieder, A. Siebert, A. Smolka, and A. Wirtz, 2001: "World Map of Natural Hazards - A Global View of the Distribution and Intensity of Significant Exposures." Natural Hazards. 23, 443-465.

Blaikie, P., T. Cannon, I. Davis, and B. Wisner, 1994: At Risk: Natural Hazards, People's Vulnerability and Disasters. London, Routledge.

Boruff, B.J., J. A. Easoz, S. D. Jones, H. R. Landry, J. D. Mitchem, and S. L. Cutter, 2003: We're Not Just in Kansas Anymore: A Half Century of Tornado Hazards in the U.S. Climate Research (forthcoming).

Fujita, T. T., 1971: Proposed characterization of tornadoes and hurricanes by area and intensity. SMRP Research Paper 91, University of Chicago, Chicago, IL, 42 pp. (Available from Wind Engineering Research Center, Box 41023, Lubbock, TX 79409).

Glass, R.I., R.B. Craven, D.J. Bregman, B.J. Stoll, N. Horowitz, P. Kerndt, and J. Winkle, 1980: "Injuries from the Wichita Falls Tornado: Implications for Prevention." Science. 207, 734-738.

Kasperson, R.E., 1992: "The Social Amplification of Risk: Progress in Developing an Integrative Framework." In Krimsky, S. and D. Golding (Eds.) Social Theories of Risk. Westport, CT, Praeger, 153-178.

McNally, Rand, 2000: Indianapolis, Indiana City Map.

National Severe Storms Forecast Center's Staff Members, 1980: "Tornadoes: When, Where, How Often." Weatherwise. 33 (2), 52-59. (Contributors: J.T. Schaefer, D.L. Kelly, C.A. Doswell, III, J.G. Galway, R.J. Williams, R.P. McNulty, L.R. Lemon, and B.D. Lambert).

National Weather Service Forecast Office, Indianapolis, Indiana, 2002: September 20 2002 Tornados.

Pearce, L., 1994: The Vulnerability of Society to Atmospheric Hazards.

Pryor, S.C., and T. Kurzhal, 1997: "A Tornado Climatology for Indiana." Physical Geography. 18 (6), 525-543.

Schmidlin, T.W., 1993: "Tornado Fatalities in Ohio, 1950-1989." In The Tornado: Its Structure, Dynamics, Prediction and Hazards. Geophysical Monograph No. 79, C. Church, D.W. Burgess, C.A. Doswell, and R. Davies-Jones (Eds.), American Geophysical Union, 459-466.

---, P.S. King, B.O. Hammer, and Y. Ono, 1998: Risk Factors for Death in the 22-23 February 1998 Florida Tornadoes. Quick Response Report No. 106.

Schneider, M.B., 2002: Miracle: No Dead in Storms. The Indianapolis Star. 09/21/2002.

Tan, S., and M. Ellis, 2002: Federal Teams Start Tallying the Wreckage. The Indianapolis Star. 09/24/2002.,news.

Twisdale, L.A., 1982: Regional Tornado Data Base and Error Analysis. Preprints, 20th Conference on Severe Local Storms. Orlando, FL, American Meteorological Society, 45-50.

United States Census Bureau, 2000a: American FactFinder.

United States Census Bureau, 2000b: Profile of selected Social Characteristics.

United States Census Bureau, 2002a: State and County QuickFacts.

United States Census Bureau, 2002b: County and City Data Book.

United States Department of Housing and Urban Development, 2002: The Daily Focus, October 7, 2002.

Wilson, J.W., and G.M. Morgan, Jr., 1971: Long-Track Tornadoes and Their Significance. Preprints, 7th Conference on Severe Local Storms. Kansas City, MO, American Meteorological Society, 183-186.

Read Complete Report

Return to Top

Return to the Hazards Center Home Page

Return to the Quick Response Report Index

March 5, 2002