Jeffrey A. Zogg
National Weather Service Office
Davenport (Quad Cities), Iowa
As part of the National Weather Service (NWS) Modernization and Associated Restructuring program (MAR), the NWS Davenport (Quad Cities) office, like many other NWS field offices, is in a period of transition. The old Weather Service Office (WSO) in Moline, Illinois, has evolved into the new Next Generation Radar (NEXRAD) WSO (NWSO) in Davenport, Iowa (DVN). Among the MAR's final steps for NWSO DVN is the evolution into a Warning and Forecast Office (WFO). Since severe thunderstorms pose one of the most serious meteorological threats to this WFO's County Warning Area (CWA) of eastern Iowa and northwestern Illinois, a climatological study was warranted. The results of this work will enable NWS personnel to improve their severe thunderstorm preparedness and serve as a training resource for use in public awareness of severe thunderstorms. This improved preparedness may ultimately lead to a reduction of life and property losses due to severe thunderstorm activity.
In a recent climatology of significant tornadoes (tornadoes producing F2 or greater damage and/or at least one fatality (Grazulis 1993; hereafter Grazulis)) for the Des Moines, Iowa (DMX), modernized CWA, Prentice (1991) suggested "future research efforts could expand to include all future WSFOs [Weather Service Forecast Offices]." Following that suggestion, DiPlacito and Kwiatkowski (1993) developed a climatology of significant tornadoes for the Goodland, Kansas (GLD), modernized CWA. This study expands on these previous significant tornado climatologies to provide a climatology of severe thunderstorms, including convective wind damage (speed greater than or equal to 58 mph) events, large hail (hail greater than or equal to 3/4 inch diameter) events, significant tornadoes, and tornado-related fatalities in the DVN CWA. The significant tornado information is presented in a manner similar to the DMX and GLD climatologies to provide for direct comparison with these past studies. The wind and hail climatology provides a proposed standardization for future research of these types of events.
The DVN CWA (Figure 1) encompasses 34 counties, 21 in eastern Iowa and 13 in northwestern Illinois. Figure 2 illustrates that the DVN CWA population increased from the 1870s through the 1970s, decreased during the 1980s, and increased to a 1995 population of 1.41 million (U.S. Census Bureau 1991, 1982, 1921; Rand McNally 1995; Funk & Wagnalls 1996). The largest cities in the DVN CWA are, in descending order, Cedar Rapids, Davenport, Dubuque, Iowa City, and Burlington. The largest metropolitan area is the Quad Cities which includes Scott County, Iowa, and Rock Island and Henry counties, Illinois. The Quad Cities' 1995 population was 355,175. The DVN CWA has a land area of 19,204 mi² which yielded a 1995 population density of 73.4 people per mi².
Figure 1. The WFO Quad Cities, Iowa-Illinois, modernized County Warning Area.
Figure 2. Decadal population distribution, 1870-1995, WFO Quad Cities modernized CWA.
Several considerations resulted in choosing the Grazulis database for significant tornadoes and the NWS National Severe Storms Forecast Center (NSSFC) database for convective wind damage and large hail events. They are the following:
Although the NSSFC database is more current than the Grazulis database, the Grazulis database was chosen for significant tornadoes. The reasons for this decision were:
The DVN CWA is similar to the DMX CWA because the relatively homogeneous spatial distribution of small farms provides an excellent opportunity for damage-based tornado climatologies, especially when Prentice (1991) says "F-scale ratings are closely tied to the destruction of residential buildings . . ." According to Grazulis, "At the turn of the century, there were thousands of 160-acre homesteads, each with a barn and house approximately a quarter-mile away from the next barn and house." The DVN CWA housing density is higher than the turn-of-the-century figure. As of 1990, there were 567,182 housing units over 19,204 mi², an average of 30 housing units per mi² (U.S. Census Bureau, 1991; Funk & Wagnalls, 1996).
Since the Grazulis database includes significant tornado data only, the official NSSFC severe weather database for 1955-1993 was used for convective wind damage and large hail. Due to differences in data collection and analysis between Grazulis and the NWS, the NSSFC large hail and convective wind damage database may be less complete and consistent than is the Grazulis significant tornado database. Sammler (1994) said ". . . a number of external factors such as population density, warning verification policy, subjectivity of event reporting and report inclusion criteria, significantly impact the [NSSFC] database." Still, the NSSFC database was the convective wind damage and large hail database most readily available for this study.
SIGNIFICANT TORNADO CLIMATOLOGY
From 1870-1991 the DVN CWA experienced 205 significant tornadoes and 138 significant tornado days (days with one or more significant tornadoes). Thirty-nine killer tornadoes struck the DVN CWA and killed 106 people. Figures 3 and 4 show the Fujita (1971) tornado intensity scale (F-scale) distribution of significant tornadoes responsible for all significant tornadoes, killer tornadoes, and fatalities. The reader should note the F1 tornadoes in Figures 3 and 4 are only those F1 tornadoes responsible for tornado-related fatalities, and as such the actual number of F1 tornadoes is higher than indicated. The F-scale of tornado intensity is shown in Table 1.
Figure 3. Significant tornado intensity distribution by number, 1870-1991, WFO Quad Cities modernized CWA.
Figure 4. Significant tornado intensity distribution by percent of all significant tornadoes, 1870-1991, WFO Quad Cities modernized CWA.
|F Rating||Tornado Type||Wind Speed (mph)|
(Source: Fujita 1971)
Figure 5 depicts the distribution of significant tornadoes and tornado-related fatalities by decade. An average of 17 significant tornadoes, 11 significant tornado days, and 3 killer tornadoes occur each decade. The 1930s through the 1970s were active with a peak in the 1960s. The 1880s, 1920s, and 1980s were relatively inactive. While killer tornado frequency is considerably lower than significant tornado frequency, the killer tornado frequency trends are similar to those of significant tornadoes.
Figure 5. Significant tornadoes, significant tornado days, killer tornadoes, and tornado-related fatalities by decade, 1870-1991, WFO Quad Cities modernized CWA.
An average of nine tornado-related fatalities occur each decade. Next to the 1870s, the 1890s was the deadliest decade with 33 fatalities. Most of these fatalities were due to the May 18, 1898 F4 tornado which killed 28 people on its 85-mile track through Cedar, Clinton, and Jackson Counties, Iowa, and Carroll County, Illinois. This twister was the DVN CWA's deadliest and longest-lived tornado. A secondary peak of fatalities occurred in the 1870s, mostly due to the April 5, 1873 F2 tornado which killed nine people in Des Moines County, Iowa, and the May 22, 1873 F4 tornado which killed eight people on its 45-mile path through Keokuk, Washington, and Louisa counties, Iowa. Another peak occurred in the 1940s, mainly due to the June 22, 1944 F4 tornado which killed nine people on its 80-mile track through Lafayette County, Wisconsin, and Stephenson County, Illinois.
An increase in killer tornadoes and fatalities occurred in the 1980s despite a decrease in significant tornadoes. Perhaps the upward trend of killer tornadoes and fatalities in the 1980s was due to tornadoes striking populated areas by chance and was not necessarily due to NWS warning quality. Regardless, since 1870 the trend in DVN CWA fatalities has been downward. A similar pattern appears in the GLD CWA, noted by DiPlacito and Kwiatkowski (1993) when they said "while [significant] tornadoes [in the GLD CWA] have decreased in frequency over the last few decades, the data reflects that deaths have decreased far more rapidly [there]." These facts at least partially support the idea that improved NWS warning technology leads to fewer deaths.
Figures 6 and 7 show the monthly and seasonal distribution of DVN CWA significant tornadoes and tornado-related fatalities. Figure 7 indicates the significant tornado season extends from mid-March through mid-November. Significant tornado activity peaks from mid-April to late May with 49 percent of all significant tornadoes and 46 percent of all killer tornadoes occurring during this period. If this period is extended through early July the figures increase to 65 percent and 67 percent respectively. A secondary maximum occurs in the fall from late October to mid-November. Prentice (1991) noted a secondary maximum in the DMX CWA centered from mid to late September, and Galway (1977) noted "the existence of a secondary maximum of Great Plains [tornado] outbreaks in the fall." Another curious peak is centered on late January. Further inspection revealed this peak is due to the January 24, 1967 tornado outbreak when ten significant tornadoes tracked from southwest to northeast across DVN's CWA. If this outbreak is removed from the data, significant tornadoes are rare from mid-November to mid-March (and are non-existent in February).
Early April through mid-July is the deadliest time of the year for the DVN CWA when 84 percent of all fatalities occur. A distinct peak in fatalities is centered on mid-May, due primarily to the May 18, 1898 F4 tornado mentioned earlier. Three other secondary peaks are centered on early April, late June, and late September. The peak in early April is due to the April 5, 1873 F2 tornado mentioned earlier. In summary, the data in Figures 6 and 7 indicate that the number of fatalities increases as the number of significant tornadoes increases.
Figure 6. Significant tornadoes, significant tornado days, killer tornadoes, and tornado-related fatalities by ten-day interval, 1870-1991, WFO Quad Cities modernized CWA.
Figure 7. Significant tornadoes, significant tornado days, killer tornadoes, and tornado-related fatalities by month, 1870-1991, WFO Quad Cities modernized CWA
In his study of Southern Plains tornado outbreaks, Moller (1979) identified tornado outbreaks as "corridor" and "cluster" events. Moller (1979) defines a "corridor" outbreak as "an outbreak of three or more tornadoes that advances generally from west-to-east within a narrow corridor of land. Tornado activity must translate through the corridor with good time continuity." A "cluster" outbreak is defined as "when four or more tornadoes occur within a roughly circular area of about 5,600 square miles. All tornadoes within a cluster must occur within a single severe thunderstorm episode."
Moller (1979) found that tornado "corridor" outbreaks:
Moller (1979) also found that tornado "cluster" outbreaks:
"Cluster" events in the DVN CWA are rareonly two obvious events occurred between 1870 and 1991. The first outbreak occurred in Henry, Whiteside, Warren, and Knox Counties, Illinois, between 1700-1900 CST 5 August 1875 and left two people dead. The second outbreak occurred in Henderson, Warren, and Henry Counties, Illinois, between 1715-1830 CST 3 May 1914 and left one person dead. A third outbreak not included in the Grazulis database occurred on May 9, 1995 (Wolf et al. 1996). No fatalities occurred during this event. These three outbreaks illustrate the rarity of "cluster" outbreaks in the DVN CWA.
Figure 8 indicates the hourly (CST) distribution of significant tornadoes and tornado-related fatalities in the DVN CWA. However, this climatology will discuss the hourly distribution of significant tornadoes using the concept of Normalized Solar Time (NST). NST was chosen to illustrate the effects of diurnal heating on significant tornado occurrence. Doswell (1985) notes that NST:
In other words, both daytime and nighttime are each given 12 NST hours, even though they usually span different amounts of (absolute) time. Thus, since summer days span longer amounts of absolute time than do summer nights, a NST hour of the summer day would span less absolute time than would a NST hour of the summer night. The reverse would be true during the winter.
Figure 8. Significant tornadoes, killer tornadoes, and tornado-related fatalities by hour (CST), 1870-1991, WFO Quad Cities modernized CWA.
Figure 9 shows the hourly (NST) distribution of significant tornadoes and tornado-related fatalities in the DVN CWA. Ninety-six percent of all significant tornadoes occur between 1200-0000 NST. Eighty-five percent of all significant tornadoes occur between 1300-2000 NST and 71 percent occur between 1400-1900 NST. A distinct peak occurs from 1700-1800 NST, just before sunset. A similar peak was observed in the Denver, Colorado, CWA (Wolf, personal communication 1995). The DMX CWA experienced a peak from 1500-1600 NST, earlier than the DVN CWA peak. This lag in peak activity between the DMX and DVN CWAs may be due to the eastward progression of tornadic thunderstorms across Iowa. Another possible explanation for this activity peak before sunset may be boundary layer decoupling which could allow for enhanced thunderstorm inflow. A much smaller secondary maximum occurs between 0400-0500 NST which may be due to nocturnal mesoscale convective systems (MCSs). These findings suggest significant tornado occurrence in eastern Iowa and northwestern Illinois is tied to the diurnal heating cycle.
Figure 9. Significant tornadoes, killer tornadoes, and tornado-related fatalities by hour (NST), 1870-1991, WFO Quad Cities modernized CWA.
All tornado-related fatalities occur between 1200-0000 NST (1100-0000 CST), coincident with the period of greatest significant tornado activity. A distinct peak occurs at 1500 NST (1600 CST), partly due to the May 18, 1898 F4 tornado. Another peak occurs at 1700 NST (1900 CST), due in part to the June 22, 1944 F4 tornado. When these two events are removed from the data, the fatality trends are similar to those of significant and killer tornadoes.
Figure 10 depicts the geographical distribution of DVN CWA significant tornadoes in units of significant tornadoes per 1000 mi² per 100 years. The figure indicates a relative maximum in DVN CWA significant tornado activity bounded by Delaware and Johnson Counties, Iowa, and Rock Island County, Illinois. Delaware, Scott, and Cedar Counties, Iowa, the most active counties in the DVN CWA, have densities of 21.2, 21.0, and 18.3 significant tornadoes per 1000 mi² per 100 years respectively. Relative activity minima exist across the southwest and northeast thirds of the DVN CWA. Jo Daviess County, Illinois, the least active county in the DVN CWA, has a density of 5.4 significant tornadoes per 1000 mi² per 100 years.
One issue which may affect the data in Figure 10 is that of the possible relationship between population density and significant tornado density. Indeed, the significant tornado density maxima across Scott County, Iowa, and Rock Island County, Illinois (the Quad Cities), occurs in an area of relatively high population density (341.2 people per mi² in 1995 compared to the DVN CWA population density of 73.4 people per mi², U.S. Census Bureau 1991; Rand McNally 1995; Funk & Wagnalls 1996). Also, relatively low population density occurs in the relatively inactive areas of the southwest and northeast thirds of the DVN CWA. For example, Van Buren County, Iowa, and Jo Daviess County, Illinois, had 1995 population densities of 15.5 and 36.3 people per mi² respectively. However, Delaware County, Iowa, the most active county in the DVN CWA, had a 1995 population density of 36.7 people per mi², less than the DVN CWA average population density. Thus, without additional research these data are inconclusive in supporting a possible relationship between significant tornado density and population density in the DVN CWA.
In his book Significant Tornadoes 1680-1991, Grazulis noted areas of the United States with maxima or minima of significant tornado activity. One of those areas is a minima in the southern part of DVN's CWA, as evidenced by Figure 10. According to Grazulis:
Among the most persistent of all tornado minima in the Midwest is one in northeast Missouri, southeastern Iowa, and extreme western Illinois. This may be a "quiet zone," too far west for eastern tornado activity in the spring, and too far east for western summer activity. Many of the significant tornadoes in this area took place in unusual outbreaks such as January 24, 1967.
While Grazulis' "quiet zone" idea offers a possible explanation for this tornado minima, (low) population density may provide another reason. In his book, Grazulis noted ". . . [the] maxima and minima [of significant tornado activity] . . . virtually all seemed to have some population component." Indeed, the "quiet zone" region of northeast Missouri, southeastern Iowa, and extreme western Illinois, has a relatively low population density compared to that of the surrounding area (U.S. Census Bureau 1991, 1982, 1921). Finally, although a tornado minima exists in the "quiet zone," significant tornadoes influence this region too. While not included in the 1870-1991 Grazulis database, the Henderson County (Raritan, Illinois) May 13, 1995 tornado reached F4 intensity in the heart of the "quiet zone" (Wolf, personal communication 1995).
Figure 10. WFO Quad Cities, Iowa-Illinois, modernized County Warning Area geographical distribution of significant tornadoes, 1870-1991, in units of significant tornadoes per 1000 mi2 per 100 years.
CONVECTIVE WIND DAMAGE CLIMATOLOGY
From 1955-1993 the DVN CWA experienced 1,476 convective wind damage events and 504 convective wind damage days, an average of 38 convective wind damage events and 13 convective wind damage days each year. Figure 11 indicates the yearly distribution of convective wind damage events. The data indicate a steady upward trend in events from 1955 to 1993 with maximas in 1974 and 1993. No reports were noted in 1972. Convective wind damage days demonstrated more year-to-year continuity from 1955-1993 than did the number of events.
Figure 11. Convective wind damage reports and days by year, 1955-1993, WFO Quad Cities modernized CWA.
Figure 12 shows the monthly distribution of convective wind damage events. The summer months of June through August have the highest number of events (62 percent). The data indicate a steady increase from April through June and a decrease from June through October. Compared to significant tornadoes, the peak for convective wind damage events is delayed by one month.
Figure 12. Convective wind damage reports by month, 1955-1993, WFO Quad Cities modernized CWA.
Figure 13 depicts the hourly (CST) distribution of convective wind damage events. Seventy-eight percent of all convective wind damage events occur from 1200-0000 CST with a peak from mid-afternoon to early nighttime (1500-2200 CST). A secondary peak occurs during the early morning from 0200-0400 CST which, like the secondary significant tornado temporal peak, may be due to nocturnal MCSs.
Figure 13. Convective wind damage reports by hour (CST), 1955-1993, WFO Quad Cities modernized CWA.
LARGE HAIL CLIMATOLOGY
From 1955-1993 the DVN CWA experienced 561 large hail events and 263 large hail days, an average of 14 large hail events and 6 large hail days each year. Figure 14 indicates the annual distribution of large hail events and large hail days. While the data indicate an upward trend in large hail reports from 1955 to 1993, the upward trend is less gradual than is the one for convective wind damage events for the same time. No reports were noted in 1972, and a secondary minima occurred in 1983. A large maxima occurred in 1974 and a secondary maxima occurred in 1992. Like convective wind damage days, large hail days demonstrate more year-to-year continuity than do the number of large hail events.
Figure 14. Large hail reports and days-by-year, 1955-1993, WFO Quad Cities modernized CWA.
Figure 15 shows the monthly distribution of large hail events. The mid-spring to early summer months of April through June have the highest number of events (65 percent). The data indicate a jump from March through April, a more gradual increase from April through June, a sharp drop from June through July, and a slight rise from July through August. Note also the maxima of large hail distribution is April through June, while the maxima for convective wind damage events is June through August. A similar relationship occurs in the Jacksonville, Florida, CWA (Anthony 1994). In a severe thunderstorm climatology for that area, Anthony stated:
"The dynamics supporting these storm systems as they move into the region tend to lower freezing level heights such that the thunderstorms associated with the systems are more likely to produce hail events at the surface. As the seasons progress and thunderstorms become more frequent, the mean wet bulb temperature increases, and thunderstorms are less likely to produce hail."
Figure 15. Large hail reports by month, 1955-1993, WFO Quad Cities modernized CWA.
Figure 16 illustrates the hourly (CST) distribution of large hail events. Most large hail events (82 percent) occur between 1200-0000 CST, 63 percent occur from 1400-2100 CST, and activity peaks in the late afternoon and early nighttime. A drop in large hail events occurs between 1600-1800 CST. Reports of hail greater than or equal to 1 3/4 inch diameter approximate the same trends as all large hail reports including the dip in events between 1600-1800 CST. The relationship between time and large hail events suggests that, like significant tornadoes and convective wind damage events, large hail events are tied to the diurnal heating cycle.
Figure 16. Large hail reports by hour (CST), 1955-1993, WFO Quad Cities modernized CWA.
This study was undertaken to determine the climatology of severe thunderstorms in the WFO Quad Cities, Iowa-Illinois, modernized County Warning Area of eastern Iowa and northwestern Illinois. This climatology used a 122-year database for significant tornadoes (tornadoes producing F2 and/or greater damage or at least one fatality) and a separate 39-year database for convective wind damage (speed greater than or equal to 58 mph) and large hail (hail greater than or equal to 3/4 inch diameter) events.
Results indicate an average of 17 significant tornadoes, 11 significant tornado days, 3 killer tornadoes, 9 tornado-related fatalities, 378 convective wind damage events, and 143 large hail events occur each decade. The typical significant tornado occurs from mid-April to late May between 1400-1800 NST (1500-2000 CST) across the northwest to southeast third of the CWA. Almost all significant tornadoes occur from mid-March to mid-November between 1200-0000 NST (1200-0000 CST) across the southwest two-thirds of the CWA. Five peaks in significant tornado activity were found: a well-defined peak in early May, and secondary peaks in early January, late June, late September, and early November. Convective wind damage typically occurs from June through August between 1500-2200 CST. Over three-fourths of all events occur between 1200-0000 CST. A primary peak in convective wind damage events occurs in June and a secondary peak occurs in November. Large hail events typically occur from April through June between 1400-2100 CST. A primary peak in large hail events occurs in June and a smaller secondary peak occurs in August.
Future research efforts could include severe thunderstorm climatologies for additional WFOs or the comparison of severe thunderstorm activity from one WFO to another. According to Prentice (1991), such efforts would "define the true tornado [or general severe thunderstorm] threat for each WFO and allow time for adequate preparation and qualified staffing." Also, future research should be conducted to determine the possible relationship between population and tornado occurrence.
This climatology reveals other areas of interest as well. Among these areas are the secondary maximums for tornado activity in mid-June, late September, and early November, for convective wind damage events in November and December, and for large hail events in August. These occurrences are not what a "normal" severe thunderstorm climatology with a steady decrease in activity from a spring maximum would indicate. Statistical analyses may show if these occurrences are real or statistical artifacts. Due to the high variability of severe thunderstorm distribution, such event clustering is unlikely to occur by chance. Thus, the question becomes one of the types of synoptic weather situations favorable for these activity peaks, and how those situations are different from those for earlier events.
One particular area of interest is the sharp decrease in large hail events from 1600-1800 CST. This occurrence is not what a "normal" hail climatology with a steady increase in large hail activity from 1600-1800 CST would indicate. Most of the possible explanations for this drop are ones of data collection such as possible "missing" data. Future research should be conducted to determine the cause of this activity decline.
This climatology suggests other investigation avenues too. For example, why are severe thunderstorms apparently more common in the early rather than late morning? Also, how could the severe thunderstorm preparedness program be modified to reduce or eliminate the increase in late evening fatality rates? Finally, since most significant tornado, convective wind damage, large hail, and flash flood events are produced by supercells, the deployment of the WSR-88D network may allow for the initiation of supercell, and thus more thorough severe thunderstorm, climatologies.
I would like to express my gratitude to Kurt Van Speybroeck, Storm Prediction Center, for providing convective wind damage and large hail events data for the WFO Quad Cities modernized County Warning Area from the official National Severe Storms Forecast Center severe weather database. Special thanks also goes to Alan R. Moller, National Weather Service, Fort Worth, Texas, for providing a copy of his Master's thesis. I would also like to thank Ray Wolf, National Weather Service, Quad Cities, Iowa-Illinois; Dr. Eugene S. Takle, Iowa State University; and Dr. William R. Gallus, Iowa State University, for their helpful suggestions and reviews during the preparation of this paper. Finally, the suggestions and materials provided by the Quad Cities, Iowa, and Des Moines, Iowa, National Weather Service office staffs are gratefully appreciated.
Anthony, R. W., 1994: Severe thunderstorm climatological data for the new Jacksonville, Florida, County Warning Area. NOAA Technical Memorandum NWS SR-155. U.S. Department of Commerce, NOAA, NWS, Southern Region, Scientific Services Division, Fort Worth, TX, 19pp.
DiPlacito, L. R., and J. J. Kwiatkowski, 1993: A brief climatology of significant (F2-F5) tornadoes in the Goodland, Kansas, Modernized County Warning Area. Central Region Applied Research Paper 13-01. U.S. Department of Commerce, NOAA, NWS, Central Region, Scientific Services Division, Kansas City, MO, pp. 1-8.
Doswell, C. A., III, 1985: The operational meteorology of convective weather. Volume II: storm scale analysis. NOAA Technical Memorandum ERL ESG-15. U.S. Department of Commerce, NOAA, NWS, Environmental Sciences Group, Boulder, CO, 240pp.
Fujita, T. T., 1971: Proposed characterization of tornadoes and hurricanes by area and intensity. Satellite and Mesometeorology Research Project (SMRP) Research Paper, No. 91, The University of Chicago, 42pp.
Funk & Wagnalls, 1995: The World Almanac and Book of Facts, 1995. Funk and Wagnalls, Mahwah, NJ, 976pp.
Galway, J. G., 1977: Some climatological aspects of tornado outbreaks. Mon. Wea. Rev., 105, 477-484.
Grazulis, T. P., 1993: Significant Tornadoes 1680-1991. A Chronology and Analysis of Events. Environmental Films, Tornado Project, St. Johnsbury, VT, 1326 pp.
McNally, R., 1995: 1995 Commercial Atlas and Marketing Guide. 126th Edition. Rand McNally, United States of America, 572pp.
Moller, A. R., 1979: The climatology and synoptic meteorology of Southern Plains' tornado outbreaks. M.S. Thesis, University of Oklahoma Graduate College, Norman, OK, 70pp.
Prentice, R. A., 1991: A climatology of significant (F2-F5) tornadoes 1880-1989 in the WFO Des Moines Modernized County Warning Area. Central Region Applied Research Paper 10-3. U.S. Department of Commerce, NOAA, NWS, Central Region, Scientific Services Division, Kansas City, MO, pp. 36-43.
Sammler, W. R., 1994: An updated climatology of large hail based on 1970-1990 data. NSSFC Operational Notes, March 1994. U.S. Department of Commerce, NOAA, NWS, NSSFC, Kansas City, MO, 4-7.
U.S. Bureau of the Census, 1991: 1990 Census of Population and Housing. Summary Populations and Housing Characteristics: Illinois. U.S. Department of Commerce, Economics and Statistics Administration, Bureau of the Census, Washington, DC, 666-860.
____________, 1982: 1980 Census of Population. Volume 1: Characteristics of the Population. Chapter A: Number of Inhabitants. Part 15: Illinois. U.S. Department of Commerce, Bureau of the Census, Washington, DC, 81pp.
____________, 1982: 1980 Census of Population. Volume 1: Characteristics of the Population. Chapter A: Number of Inhabitants. Part 17: Iowa. U.S. Department of Commerce, Bureau of the Census, Washington, DC, 55pp.
____________, 1921: 14th Census of the United States Taken in the Year 1920. Number and Distribution of Inhabitants. Bureau of the Census, Government Printing Office, Washington, DC, 695pp.
Wolf, R., R. Pryzbylinski, and P. Berg, 1996: Observations of a merging bowing segment and supercell. Preprints, 18th Conf. On Severe Local Storms. San Francisco, CA, AMS (Boston)., 740-745.