The Update Climatology of Indiana Tornadoes, 1953 to 1989

The Update Climatology of Indiana Tornadoes, 1953 to 1989

Douglas A. Speheger
Department of Earth and Atmospheric Sciences
Purdue University
West Lafayette, IN 47907
In partial fulfillment of the
Undergraduate Honors Program
May 1990

1. Introduction

Indiana is no stranger to the occurrences of tornadoes. According to National Severe Storms Forecast Center (NSSFC) records, for the year 1953 through 1988, it ranks third in frequency of occurrence when the size of the state is taken into account (5.5 tornadoes per year per ten thousand square miles), behind Oklahoma (7.6) and Florida (7.5). The last effort to produce a tornado climatology for the state was by Keyser et al. (1977). Therefore, the present effort has been made to update the tornado climatology of Indiana. Records of tornadoes were taken from a variety of sources to compile as "complete" a record as possible of tornadoes in the state. While historical data extend back to 1880, data for 757 events since 1953 are the most complete and include date, specific counties affected and usually F-scale. Data from this period are analyzed more in depth, both geographically and over various time scales.
Fifty-five violent events, rated F4 and F5, have been identified in Indiana for the expanded time period of 1880 to 1989, 24 of them occurring since 1953. These violent tornadoes are deemed to be a very serious threat and therefore warrant special attention.

2. Data Sources/Periods of Records

According to Grazulis (1989), there are four general periods of time associated with tornado climatology. Events before 1880 are generally undocumented except for a few records like those in Ludlum (1970). This time period before 1880 is termed the Historic Period. The period between 1880 and 1915 has been termed the Early Period. The time from 1916 to 1952 is called the Middle Period. Reports from these three time periods are usually newspaper accounts. Unfortunately, these journalistic accounts are usually incomplete in a meteorological sense. Also, it is likely that events in this period were recorded only if they did enough damage or injury to be considered "newsworthy."
Since 1953, the National Severe Storms Forecast Center has attempted to record all of the tornadoes in the United States. Data from this time period are generally reliable and more complete. Since 1959, tornadoes and other severe weather events have been published monthly by the National Oceanic and Atmospheric Administration (NOAA) in the magazine Storm Data. In 1971, Dr. T. Fujita of the University of Chicago developed an intensity scale that rates tornadoes according to wind speed on a scale from zero to five. Most records during this Modern Period from 1953 to date include this F-rating.
Another important event in the history of tornado records is the recognition of the downburst beginning around 1977. Before this time, damage resulting from non-tornadic thunderstorm winds may have been characterized as tornadic and so included in the tornado records. However, since 1977, investigators are more likely to recognize the damage associated with thunderstorm winds and not mistakenly report these events as tornadoes.
An attempt has been made to gather as complete a data set as possible of reports of tornado occurrences in Indiana by using a variety of sources. The primary data source was the U.S. Department of Commerce publication Storm Data. All of the data in Storm Data, including the written descriptions of the events, were entered into a data base. The official tornado log of the National Severe Storms Forecast Center was also used. Missing events and other revisions were taken from Grazulis (1984,1990 personal communication). Violent events, both historical and modern, were taken from Grazulis (1984). Other historical records were taken from Ludlum (1970) and Flora (1953), but because of the incompleteness of the data, were not used in this study. This paper will use data from 1953 to date, except for the study of violent events.
Although we have attempted to create a data base of all reported Indiana tornadoes, this should not be considered a complete data set of all tornadoes in the state. In historical times especially, but even today, tornadoes that strike Indiana will not always be observed and recorded. Weak tornadoes that touch down in sparsely-populated rural areas and do no significant damage will occasionally miss being reported. Another problem with the tornado records is the misidentification of non-tornadic events as discussed earlier. This problem can not be easily solved for a tornado climatalogical study, so for this paper it is assumed that all events listed in Storm Data, the NSSFC log, etc... were in fact tornadoes.
For violent tornadoes, those of F4 and F5 intensity, a longer time period is used. Grazulis (1984) made a complete list of tornadoes that he considered to be of violent intensity dating back to 1880. It is probable that most violent tornadoes were indeed reported in newspapers because of the violence and extreme damage that often is a result of these intense storms. At the end of this paper, when violent storms are studied more closely, Grazulis' record of F4 and F5 tornadoes from 1880 to date is used.

3. General Statistics

A total of 757 tornadoes were recorded in Indiana between 1953 and 1989. (This is about 13 more than the NSSFC official log which has 718 recorded through 1988 and an additional 26 more occurring in 1989. This is due to the fact that we included other sources in our records as listed above). The 757 tornadoes averages about 20 per year. In one calendar year, as many as 49 (1973) and as few as four (1982, 1984) have occurred. Figure 1 shows the number of tornadoes for each year. The activity from year- to-year fluctuates greatly. There was only moderate tornadic activity around 1970, followed by a drastic increase in 1973. The general trend appears to decrease through the late 1970's and into the early 1980's, when Indiana experienced some uneventful years with only 4, 5, 4, and 8 tornadoes per year in 1982 through 1985, respectively. In the most recent four years, tornadic activity has recovered some, but still cannot be considered very active. Not since 1980 has there been a year with more than 30 tornadoes.
Part of the recent decline in the number of reported tornadoes may not be meteorologically related. It could be due to downbursts and other non- tornadic causes of damage being more accurately diagnosed in recent years and to the fact that damage by some of these causes may have mistakenly been attributed to tornadoes earlier.
Another important feature in Indiana's tornado climatology is that three of the nation's most significant tornado events affected the state. The infamous Tri-State tornado finished it's 219-mile devastating trek in Southwestern Indiana on March 18, 1925. Seventy-six Hoosiers died and 400 more were injured by that event. During it's three-and-one-half hours on the ground in Missouri, Illinois, and Indiana, almost 700 people were killed and over 2,000 injured (Wilson and Chagnon, 1971).
On April 11, 1965, 31 tornadoes struck the Great Lakes region, including ten in Indiana. 137 were killed in Indiana that day, known as the Palm Sunday Outbreak, and 1,795 others were injured in Indiana's worst tornado event in terms of loss of life and injuries.
On April 3, 1974, the Super Outbreak brought 148 tornadoes to thirteen states, twenty-one of them affecting Indiana. The town of Monticello, Indiana was especially hard hit by a tornado that travelled 121 miles from one side of the state to the other. In total, 49 were killed and an additional 896 were injured by the tornadoes of that day.
Although outbreaks like those of 1965 and 1974 are rare, they are the most serious threat to the lives and property of residents of Indiana. These outbreaks also play a considerable role in the tornado statistics. A closer look at outbreaks is discussed later in the paper.

4. Tornado Distribution (F-Scale)

Fujita (1971) developed an intensity scale which rates tornadoes on a scale from zero to five. Table 1 shows the categorization of tornadoes on this scale. Figure 2 shows the distribution of tornadoes in Indiana by their F-scale ratings. Most tornadoes that occur are weak (F0 and F1). In Indiana, fifty-four percent of recorded tornadoes have been classified as weak. Thirty-seven percent have been characterized as strong (F2 or F3) tornadoes. Fortunately, only about three percent of tornadoes in Indiana have been in the category of violent tornadoes (F4 or F5). Six percent of the tornadoes recorded have no F-scale classification.

5. Monthly Distribution

The monthly distribution of tornadoes shows that the most favored time of year for tornadoes is the spring and early summer. Figure 3 shows peaks in April and June of 165 and 175, respectively. Statistics for the month of April are heavily influenced by the two large outbreaks that hit the state in 1965 and 1974. But even without these two outbreaks included in the April total, a secondary peak (of 134) would still remain. The "tornado season" begins in March, when 85 tornadoes have been reported since 1953. At the end of the season, the events drop off rapidly from 175 in June, to 74 in July, to 35 in August. The monthly totals then slowly decrease to a minimum in January of 10 in the last 37 years. Another interesting feature of the distribution is the relatively "calm" month of May. Although 101 tornadoes have been recorded in that month, it is significantly less than the peaks in April and June on either side.
Fifty-eight percent of the tornadoes in Indiana occur in the months of April, May, and June. 79 percent occur from March through July. This springtime peak is expected because the conditions that lead to tornado development are more often in place in the spring transition from the cold winter conditions to the warm summer months. Strong cold fronts bringing down cold air from Canada are common events this time of year. When a surge of cold air meets warm moist air spreading northward from the Gulf of Mexico, the contrast of air masses sets up conditions favorable to the development of severe thunderstorms and tornadoes.
The monthly variation of tornado days (days on which one or more tornadoes occurs), shows a significant contrast to the actual number of tornadoes. The double peak of April and June does not exist with tornado days. Although April had 165 tornadoes, they only occurred on fifty-seven different days. April averaged about three tornadoes per tornado day. The two large outbreaks occurring in April certainly have an influence in that statistic, but it still appears that in general, the early spring is more likely to have more tornadoes per tornado day than in the summer and fall periods. A similar ratio of about 3 tornadoes per tornado day is evident for February and March. In June, there were fifty percent more tornado days than April, to produce a similar number of tornadoes. Each tornado day averages only two tornadoes in June.
Not only do springtime tornado days produce more tornadoes, but springtime tornadoes are stronger on the average. Figure 4 shows the distribution of tornadoes for each month according to their strength. Nineteen violent F4/F5 tornadoes occurred in April, or twelve percent of that month's tornadoes. In contrast, since 1953 not one violent tornado has occurred in June. In fact, 23 of the 24 violent tornadoes since 1953, and all five F5 events have occurred in March, April, and May. A similar result is shown for strong F2/F3 tornadoes. Fifty-four percent of March's tornadoes, and 40 percent of April's are classified as strong, while only about 25 percent of the tornadoes in May, June, and July are strong. A majority of summer tornadoes are weak F0/F1 events. For May, June, and July, the percentage of weak tornadoes is 64, 70, and 63 percent, respectively. Only about 40 percent of March and April tornadoes are weak. Again, this probably is a result of temperature contrasts being stronger in springtime. Once summer heat settles into Indiana, temperature contrasts are not as pronounced as in the transition time between winter and summer. With a decrease in this driving force, thunderstorms and associated tornadoes, in general are weaker. Tornadoes still occur in the summer because there are a large number of thunderstorms occurring, but the storms are more of the air-mass variety that occur with strong surface heating, and these are not as well organized and are not as strong as the cold front-induced thunderstorms.

6. Time of Day Distribution

Tornadoes can occur at all times of the day. The afternoon and evening hours, however, are most preferred. Figure 5 shows the istribution of tornadoes during hour intervals of the day for the hour beginning with the x- axis value. 71 percent of the tornado occurrences within the last 37 years have been between the hours of 2pm and 10pm Eastern Standard Time, and 83 percent between noon and midnight. Again this should be expected. The peak time for thunderstorms in general is in the afternoon and evening hours. This reflects storm initiation during the warmest part of the diurnal temperature cycle (roughly 1 to 4 pm). This aids convection as the surface layer is heated and stimulates upward motion of the warm surface air. However, tornadoes cannot be ruled out during the morning hours. 129 tornadoes (an average of 3.5 per year) occur in the hours between midnight and noon. These tend to be weaker, however. Only five morning tornadoes have been rated as F3, and no morning tornadoes have reached the violent category in Indiana. The majority (106) are in the F1 and F2 intensity classification.
Table 2 shows the distribution by time of day for each month. A slight trend in the peak time can be detected, shifting slightly later in the spring and early summer, then again shifting back in the late summer. This can most likely be attributed to the fact that the sun sets later in the day approaching the beginning of summer. In March, the peak time for tornadoes occurs between 3pm and 5pm. Between April and June, a longer period of peak activity is seen, but the maximum time is seen to be just after 5pm in April and May, and slightly after 6pm in June.

7. Distribution by Day of the Week

One statistic that should be uniform is the occurrences of tornadoes by day of the week. There is no reason to believe that a bias exists that favors one day of the week over another for tornado occurrences. But the distribution by day of the week shows that tornadoes are not evenly distributed. Figure 6 shows the distribution of Indiana tornadoes by the day of week. A maximum of 134 is seen on Monday, declining to 87 on Thursday. A second peak occurs on Friday and Saturday, followed by a slight decline on Sunday. If the two major outbreaks are taken out, it does not significantly change the results. The 1974 outbreak occurred on a Wednesday with 21 tornadoes, and the 1965 outbreak with 10 tornadoes occurred on a Sunday. When these two day's events are not considered, this only amplifies the mid-week and Sunday minima.
A chi-squared test was performed on the distribution by day of week, and the hypothesis of a uniform distribution could be rejected at the 99% confidence level. Wilson and Kelly (1977) examined the day of week distribution of the tornadoes of the United States through 1976, and also found a distribution that was not uniform. However, their results showed that Thursday and Friday have the largest number of tornadoes reported, while ours show that Thursday is actually the minimum for Indiana. Figure 7 shows a comparison of Indiana tornadoes and U.S. tornadoes from Wilson and Kelly by day of the week. They also found their minimum for the United States on Saturday, and attributed this fact primarily to the reduction in number of Sunday newspapers compared to the other six days. We do not see similar results for Indiana.
Stratifying Indiana tornadoes according to their intensity was done to see if removing weak events would show that stronger tornadoes were distributed more uniformly. The results showed that the F4/5 category were dominated by the outbreak days on Wednesday and Sunday. The F2/3 category roughly mirrored the total distribution with a minimum in the middle of the week (Wednesday and Thursday), and dual peaks early in the week (Monday and Tuesday), and again on Friday.
However, if we look at tornado days, this statistic does appear to be close to uniform. The tornado days has a maximum on Saturday with 59, followed by a minimum of 46 on Sunday. A chi-squared test was performed on the tornado days, and we cannot reject the uniform distribution hypothesis at the 90 % confidence level. Again, this is different than the distribution of the U.S. tornado days discussed by Wilson and Kelly. They show a tornado day minimum on Saturday and maximum on Tuesday, but they also found close to a uniform distribution.

8. Geography

The tornado geographic distribution was strongly influenced by urban areas. Maximum areas occur in Marion county (Indianapolis), Tippecanoe county (Lafayette), and Elkhart county (Elkhart). Figure 8 shows the number of tornadoes affecting each Indiana county, and the number of tornadoes that started their path in each county. Looking past these urban influences, it appears as if there may be a slight north to south gradient of tornadoes with more activity in the northern part of the state. To test this, a three- dimensional-least squares fit using the x and y coordinates as points approximately at the centroid of the county, and the z coordinate as the number of touchdowns in each county divided by the area of the county, a plane was fit to the data. The total number of tornadoes per county was not used since tornadoes crossing county borders would be counted more than once. For example, the Monticello tornado of April 3, 1974 passed through eleven different counties as it moved across the state. The result of this least- squares-fit plane is shown on Figure 9. A slight north-northwest to south- southeast gradient is observed. There are several possible explanations of why the southern part of the state would have fewer tornadoes reported. The terrain difference, with Southern Indiana having more hills, could affect the formation of tornadoes. Also, in general, the southern part of the state is more sparsely populated than the north and, therefore, would contain fewer people to actually observe tornadoes.
However, this plane does not represent a perfect distribution of tornadoes in the state. If an even distribution of tornadoes per area is used to define the state's distribution, the sum of the squares of the residual is .0107. If the least-squares-fit plane mentioned above is used, the sum of the squares of the residual is .00923. This means that the fitted plane only reduces the variation of tornadoes by fourteen percent. Again, the population bias probably plays a larger role in the variation of tornadoes than does the geography.

8.1. Changes by month or season

To see if the monthly distribution was a function of the geography of the state, the state was divided into three sections of roughly equal area. Figure 10 shows the division of Indiana into a Northern, a Central, and a Southern section. With one small exception, the division was along the boundary of National Weather Service forecast zones. The tornadoes were then separated according to the counties affected. If a tornado crossed from one section into another, the tornado was recorded in both sections.
Figures 11a, 11b, and 11c show the monthly distribution for the Northern, Central, and Southern sections respectively. The north-to-south tornado occurrence gradient is also apparent in this separation. The northern section of the state recorded 304 tornadoes, the central section tallied 276, while the number fell off to 191 in the south. Focusing first on the Northern section, Figure 11a shows a complete lack of tornadoes in January and February and a sharp rise in March and April. The twin peaks of April and June are very obvious, with both being almost double the value of the months of March, May, and July. The number of tornadoes declines again in August, and a small relative maximum is again observed in October.
Figure 11b shows that the distribution for Central Indiana also contains twin peaks in April and June; however, the April peak is much smaller. It also shows March approaching the value of April. There still exists in Central Indiana a relative lull of tornadic activity in May. The distribution drops off very sharply in July and decreases until September followed by a late season relative maximum in October and November.
Figure 11c shows that Southern Indiana has a number of tornado occurrences every month of the year, but that the significant activity does not start until later than the Northern and Central parts of the state, waiting until April. The twin peaks shown in the other parts of the state do not occur in Southern Indiana. In fact, even though June is the maximum time of tornado activity in the rest of the state, the number of tornadoes actually decreases from April to May and again from May to June in the South. After June, the activity drops significantly again in July showing that there are only three months of significant tornado activity in the Southern third of the state. Two late season relative maxima are observed here in September and November.

9. Injuries/Fatalities

In the past thirty-seven years, 208 people have died in Indiana from tornadoes, and 3,388 injuries have occurred. This corresponds to an average of about six deaths and 92 injuries per year. Indiana not only ranks third in number of tornadoes per 10,000 square miles, but also ranks third in the nation in tornado deaths per area with 57 deaths per 10,000 square miles behind Massachusetts (120) and Mississippi (72). Figures 12a and 12b shows the distribution of deaths and injuries by the intensity of the tornado which caused them. Although weak F0 and F1 tornadoes account for fifty-four percent of the tornado occurrences in Indiana, they cause minimal loss of life and injuries. Only five deaths (2.4%) and ninety injuries (2.7%) have been attributed to weak tornadoes. Another fifteen percent of deaths and twenty percent of injuries are caused by strong F2 and F3 tornadoes. In contrast, despite only about three percent of tornadoes being classified as violent, these storms account for 83 percent of the deaths and 77 percent of the injuries caused by tornadoes in the state. Since these relatively few tornadoes cause a disproportionate percentage of death and injury, a closer look at violent tornadoes in Indiana is discussed in the next section.

10. Violent Events

As was discussed earlier, violent F4/F5 tornadoes contributed a disproportionate number of the deaths and injuries occurring in the state. Since these are a large threat to the state, a brief look at the Indiana violent tornado climatology seems warranted.
Grazulis (1984) recorded tornadoes dating back to 1880 that he considered to be of the violent category. The violent tornadoes used here are primarily those listed by Grazulis. Because of the small number of events, an expanded time period of 1880 to 1989 is used instead of the 1953 to 1989 time frame used elsewhere in this paper. Because of the intense destruction associated with violent tornadoes, it is reasonable to assume that most of these storms between 1880 and 1952 were reported although many of their weaker counterparts may not have been.
Fifty-five violent tornadoes have been recorded in Indiana between 1880 and 1989, twenty-four of them since 1953. Figure 13 shows the distribution of violent events by year. The two prominent spikes represent the two major outbreaks of 1965 and 1974. The Palm Sunday Outbreak spawned eight violent tornadoes, two of F5 intensity, while the 1974 Super Outbreak spawned nine violent tornadoes, three of which were F5. Figure 14 displays the monthly distribution of violent tornadoes. In Indiana, all the violent events have been restricted to the months of March through July. These months, especially the spring months of March through May, are the months when the weather conditions would most likely favor strong tornado development as discussed earlier [footnote: This is not to say that violent F4/F5 tornadoes can not happen in other times of the year. Indiana's western neighbor Illinois, which has roughly the same type of climate, has had violent tornadoes touch down in September, and even December, January, and February. (Grazulis, 1984)]
Figure 15 shows the distribution of violent tornadoes by time of day. The activity is confined to the hours between 12 noon and 9 pm. Since even more favorable conditions are needed to make a violent tornado compared to a weaker tornado, we would expect the afternoon and early evening hours to be the best time for violent tornado occurrences.

11. Contribution of outbreaks

As shown in the last section, seventeen of the fifty-five violent events of the past 110 years happened on two days, the April 11, 1965 Palm Sunday Outbreak and the April 3, 1974 Super Outbreak. These incredible days heavily influenced many of the statistics shown in this paper, and other results that were not shown because the statistics were so heavily biased by these events that they were of little use. For example, ninety percent of the deaths that have occurred as a result of tornadoes since 1953, happened on these two days, so distribution of deaths by year, month, or even location would be skewed by these two days of tornadoes. Over half of the injuries since 1953 happened on the day of the Palm Sunday tornadoes. About 25% of the injuries in the past 37 years were caused by just one tornado, the Palm Sunday Russiaville tornado that caused 835 injuries.

12. Conclusion

Over the past 37 years, between 1953 and 1989, 757 tornadoes have struck the state of Indiana. Indiana thus ranks third in the nation with about 5.5 tornadoes per 10,000 square miles per year, behind Oklahoma and Florida but ahead of other "tornado alley" states like Kansas, Iowa, and Nebraska. During this time, there have been 208 deaths and 3,388 injuries in the state caused by tornadoes. The favored times of tornado development are in the springtime and in the afternoon and early evening hours, although tornadoes have occurred in all months and at all times of the day. Over two-thirds of tornadoes happen in the months between March and July, and over sixty percent strike between the hours of 2pm and 9pm Eastern Standard Time. A relative calm has prevailed over the state over the past nine years, especially in the early 1980s. Not since 1980 has there been a year with more than 30 tornadoes, and in 1982 through 1984 there were a total of only 13 tornadoes. The state was divided into three sections showing three somewhat distinct monthly distributions. Both Northern and Central Indiana showed maximum activity in June with a secondary peak in April. Southern Indiana, however, shows a maximum in April decreasing through August.
Geographically, a bias towards populated areas seems to exist because there are more potential observers of tornadoes in urban areas. Marion county (Indianapolis), and Tippecanoe county (Lafayette) are the counties of maximum reported tornadoes. Looking past the urban biases, a north-south gradient seems to appear. A three-dimensional plane was fit to the tornado data. The x and y coordinates were taken to be coordinates of the centroid of the county, with the z coordinate being the number of touchdowns in that county divided by the county's area. The plane showed a gradient of tornadoes with the most occurring in the northwest corner of the state, decreasing to the south-southeast.
Although weak tornadoes account for over half of the tornadoes in Indiana, it is the violent tornadoes, rated F4 and F5 on the Fujita intensity scale, that present the biggest danger to the population. Only three percent of the tornadoes in Indiana since 1953 have been characterized as violent, but these few storms have accounted for 83% of the tornado deaths, and 77% of the injuries.
A special look at the climatology of violent tornadoes in Indiana for the expanded time period of 1880 to 1989 shows that the temporal distribution is even more pronounced than their weaker counterparts. All violent events in the past 110 years have been restricted to the months between March and July and the time of day between 12 noon and 9 pm.
Seventeen of these violent events, and ninety percent of the deaths since 1953 have been attributed to two days, the 1965 Palm Sunday Outbreak, and the 1974 Super Outbreak. The events of these two days also heavily bias many statistics of the climatology of Indiana tornadoes.
Although all tornadoes are potential killers if citizens are not prepared, the real danger lies in those few devastating events that are hard to protect against. These violent tornadoes with wind speeds over 200 miles per hour, although rare, are the most serious threat to the safety of residents. Outbreaks of violent events are especially dangerous. The Palm Sunday Outbreak of April 11, 1965 brought ten tornadoes to Indiana, eight of them violent. In just this one day, 137 were killed and 1,795 injured just in the state of Indiana, 835 injured by just one tornado. The Super Outbreak of April 3, 1974 struck 13 states with over 130 tornadoes. 21 occurred in Indiana, nine of them classified as violent. 49 died and another 896 injured in Indiana on that day. One violent tornado travelled over two hours and for 121 miles across the state on that day. It is these violent and outbreak events that we must learn more about in order to protect the lives of the people.
Easterling, D.R., and P.J. Robinson, 1985: The Diurnal Variation of Thunderstorm Activity in the United States. J. Clim. Appl'd Meteor., 24, 1048-1058.
Flora, S.D., 1953: Tornadoes of the United States. University of Oklahoma Press, 193 pp.
Fujita, T.T., 1971: Proposed characterization of tornadoes and hurricanes by area and intensity. SMRP Res. Paper No. 91, Dept. of Geophys. Sci., Univ. of Chicago, Chicago, IL, 42 pp.
Grazulis, T.P., 1984: Violent Tornado Climatography, 1880 - 1982. NUREG/CR- 3670, U.S. Nuclear Reg. Comm., Washington D.C., 165 pp.
Grazulis, T.P., 1989: Significant Tornadoes...A 107 year perspective. Sixth U.S. National Conference on Wind Engineering (Houston, TX), 21 pp.
Kelly, D.L., J.T. Schaefer, R.P. McNulty, C.A. Doswell, and R.F. Abbey, 1977. Presentation and Interpretation of An Expanded Tornado Climatology. Preprints, 10th Conf. on Severe Local Storms (Omaha, NE), AMS, Boston, MA, 186-187.
Keyser, D.A., E.M. Agee, and C.R. Church, 1977: The Modern Climatology of Indiana Tornadoes. Proc. Ind. Acad. Sci., 86, 380-390.
Ludlum, D.M., 1970: The History of American Weather: Early American Tornadoes, 1586-1870. AMS, Boston, MA. 219 pp.
Neter, J., and W. Wasserman, 1974: Applied Linear Statistical Models. Richard J. Irwin, Inc, 179-268.
Schmidler, T.W., 1988: Ohio Tornado Climatology, 1950 - 1985. Preprints, 15th Conf. on Severe Local Storms (Baltimore, MD), AMS, Boston MA. 523- 524.
Wilson, J.W., and S.A. Changnon, 1971: Illinois Tornadoes. Circ. 103, Illinois State Water Survey, Urbana, IL. 58 pp.
Wilson, L.F. and D. Kelly, 1977: Tornado Climatology by Day of the Week. Preprints, 10th Conf. on Severe Local Storms (Omaha, NE), AMS, Boston MA. 194-198.