Mark F. Britt* and Fred H. Glass

St. Charles, Missouri


Clearly the most important task facing a NWS field office forecaster is the prediction and warning
of severe local storms.  At the very foundation of the forecast and warning decision process
is a knowledge of the climatology of the severe weather events.  Knowing the annual, monthly
and hourly distribution of tornadoes can give a warning decision maker greater insight into
the commonalities and probabilities of an event occurring, even before the first thunderstorm forms.

County Warning Areas (CWAs) were redrawn in the 1990s as part of the Modernization and 
Reorganization of the National Weather Service (NWS).  The newly reorganized CWAs are larger
and encompass geography never previously linked together.  This paper examines the climatology 
of tornadoes of the National Weather Service Forecast Office in St. Louis, Missouri (NWS LSX).
The NWS LSX CWA encompasses 46 counties (Figure 1, below) in central and eastern Missouri
and portions of western Illinois, with an approximate population of 3.4 million.


Figure 1 -
WFO LSX County Warning Area

2.  DATA

2.1 Data Sources and Methodology

The Storm Prediction Center's (SPC, formerly the National Severe Storms Forecast Center) severe 
weather database was used for this study (Schaefer and Edwards, 1999).  This database contains 
tornado statistics from 1950 to 1995.  Statistics for 1996 to 1999 were extracted from Storm Data
(NOAA 1959-1999) for inclusion in this study.

Since tornado warnings are issued and verified by county, the SPC database contains information
on tornado segments rather than tornadoes.  A segment is counted each time a tornado touches
down or crosses a county line, therefore a single tornado which moved across three counties would 
be recorded as three separate tornado segments.  After close examination of all tornado segments
using the Severe Plot software (Hart 1993) and Storm Data, the authors were able to merge
individual segments back into individual tornadoes.

Tornado intensity is determined by using the F-Scale (Fujita 1981), as listed in Storm Data.  This 
paper follows the accepted nomenclature that F2 and F3 tornadoes are strong and F4 are violent
Tornado times are in Central Standard Time (CST) and refer to the hour in which they occurred.
For example, 14 refers to any tornado that occurred between 1400 to 1459 CST. A tornado day is
a traditional 24-hour day (0000 to 2359 CST) in which at least one tornado occurred.

2.2 Data Limitation
Kelly et al. (1985) and Hales (1993) have documented a number of limitations inherent to the SPC
database.  They found that the number of severe weather reports increases in areas with greater
population density and/or a greater concentration of trained spotters and well-educated people.
Highway distributions, distance to an official reporting station, time-of-day, the warning office's
motivation in verifying warnings, lack of appropriate measuring devices, intervening clouds, and
topographic features are additional factors which affect severe weather reporting.  Finally, the
perception and motivation of the person witnessing the event must be considered.  For example, a 
farmer may not feel obligated to report a tornado that touches down in an open field.

Ostby (1993) found that the occurrence of weak tornadoes (F0-F1) has shown a dramatic increase
since 1980, while the number of strong and violent tornadoes has either remain steady or 
decreased.  Reasons for this include improved verification efforts by local NWS offices and the
marked increase in storm chasing.  Since strong and violent (or significant herein) tornadoes
produce a more stable long-term data set, these categories were the main focus of this study.


3.1 Overall and Yearly Distribution

During the 50 year period from 1950 through 1999, a total of 539 tornadoes (all intensities) were 
identified in the LSX CWA.  Significant tornadoes accounted for 168 of the tornadoes (or 31% of
the total), resulting in an average 3.32 strong or violent tornadoes per year.  Significant tornadoes
have occurred in the LSX CWA in 8 out of every 10 years (Figure 2, below).  The most active year
was 1957 with 12, mainly because of two days with large outbreaks across the CWA: May 21 and
December 18.  The database indicates there has been a gradual decline in the number of significant
tornadoes in the past 50 years.  The 1950s had a total of 44 tornadoes, while the 1990s only had 24.


Figure 2 -
Yearly significant tornado distribution

The LSX CWA has experienced 106 significant tornado days (days with one or more strong or
violent tornadoes) from 1950-1999, an average of 2.12 days per year (Figure 3, below).  The
number of days has also shown a gradual decline.

(missing graph)

Figure 3 -
Significant tornado days distribution

3.2 Monthly/Seasonal Distribution
The most volatile time of year for tornadoes is the middle to late spring.  Figure 4 (below) depicts
May having the highest incidence of significant tornadoes at 38, with April a close second at 35.
The four-month period from March to June contains 65 percent (109 out of 168) of all the significant
tornadoes.  Summer and early Autumn are relatively inactive, with August only having two strong
tornadoes observed in the 50 year period.  There is sound evidence of a secondary maximum from
late autumn into early winter as the number of significant tornadoes increases to a peak of 16 in
December; 21 percent of all significant tornadoes occur between October and December.


Figure 4 -
Monthly significant tornado distribution

When the monthly distribution of all tornadoes (all F scales) is plotted versus significant tornadoes
(Figure 5, below), some noteworthy observations are evident.  While the four-month period from
March-June contains the greatest percentage of significant tornadoes (65 percent; 109 out of 168)
and of all tornadoes (66 percent; 355 out of 539), only 31 percent of all tornadoes which occur
during this period are significant (109 out of 355).  In contrast, the four-month period from
November-February contains only 23 percent of the significant tornadoes (38 out of 168) and only
15 percent of all tornadoes (78 out of 539), however 49 percent of all tornadoes which occur during
this period are significant (38 out of 78).  Thus for any tornado occurrence, the greatest probability
of it being significant, is during the late autumn and winter.

Figure 5 -
Monthly distribution comparison between all tornadoes
 and significant tornadoes


3.3 Hourly Distribution

Tornadoes are mainly a late afternoon and evening phenomena in the LSX CWA (Figure 6, below),
with two thirds (113 out 168) of the significant tornadoes occurring between 300 and 1100 pm CST.
Only 5 percent (8 out of 168) occurred between 300 and 1000 am.

Figure 6 -
Hourly significant tornado distribution

A scatter plot showing the date/time distribution of strong-violent tornadoes yields some additional
interesting results (Figure 7, below).  Tornadoes that occur in the late afternoon and evening hours
(300 to 1100 pm CST) during the spring (March through June) account for 49% of significant tornado
occurrences (82 out of 168).  Occurrences outside this window tend to be spread out.  What is 
particularly disturbing is only 4 of the 11 violent tornadoes (F-4) are in this cluster.  Six of the 
remaining violent tornadoes occurred in the winter months (from four separate severe thunderstorm
during the late evening and overnight hours.  The most notable event struck the city of
St. Louis at 140 am on February 10, 1959.  This violent tornado resulted in 21 fatalities and 345

Figure 7 -
Month vs. Hour distribution

3.4 Multiple Tornado Episode

Table 1 (below) lists the number of severe weather episodes that produced two or more significant
tornadoes.  An episode was defined similar to Galway (1977) in which individual tornadoes are not 
separated by more than 6 hours.  For this study, being in the same CWA was considered spatially
linked.  Results indicate that 94 of the 168 significant tornadoes in the data base (56%) occurred in
these events.

























Table 1 -
Number of multiple tornado episodes per month


The year-to-year comparison of tornado data shows a gradual decrease in the occurrence of 
significant tornadoes observed in the LSX CWA in the past 50 years.  As expected, most of the
tornadoes occurred in the late afternoon and evening hours of the spring months.  Significant 
tornadoes however comprise a larger percentage of all tornadoes that occur during the late autumn 
through winter.  Violent tornadoes appear to have the largest temporal disparity in the occurrence,
with a cluster centered in the early evening hours and another cluster centered just after midnight.
For the immediate St. Louis metropolitan area, the climatology shows a decided preference for 
violent tornadoes during the late evening-early morning hours of the winter.

The desired goal in developing this climatology is to give forecasters greater insight into the 
commonalities and probabilities of an event occurring, and prompt a higher state of awareness even
before the first thunderstorm forms.  This is the first step toward developing a synoptic climatology
of severe weather events across central and eastern Missouri and portions of western Illinois.


The authors would like to thank Doug Speheger from WFO Norman for his computerized help in
deciphering the SPC database.  Additional thanks are extended to Ron Przybylinski (SOO/WFO 
LSX) and Steven Thomas (MIC/WFO LSX) for support in this study.

Fujita, T., 1981: Tornadoes and Downbursts in the context of generalized planetary scales. J.
Atmos. Sci.
, 38, 1511-1534.

Galway, J. G., 1977: Some Climatological Aspects of Tornado Outbreaks. Mon. Wea. Rev., 105, 
Hales, J.E., 1993: Biases in the severe thunderstorm data base: Ramifications and solutions. 
Preprints, 13th Conf. Wea. Forecasting and Analysis, Vienna, VA, AMS (Boston), 504-507.
Hart, J.A., 1993: SVRPLOT: A New Method of Accessing and Manipulating the NSSFC Severe
Weather Data Base. Preprints, 17th Conf. On Severe Local Storms, St. Louis, AMS (Boston), 40-
Kelly, D.L., J.T. Schaefer, and C.A. Doswell, III, 1985: Climatology of Nontornadic Severe
Thunderstorm Events in the United States. Mon. Wea. Rev. 113, 1997-2014.
National Oceanic and Atmospheric Administration, 1959-1995: Storm Data. Vols. 1-37, Nos. 1-12,
National Climatic Data Center, Asheville, NC.
Ostby, F. P., 1993: The Changing Nature of Tornado Climatology. Preprints, 17th Conf. On Severe
Local Storms
, St. Louis, AMS (Boston), 1-5.

Schafer, J. T. and R. Edwards, 1999: The SPC Tornado/Severe Thunderstorm Database. Preprints,
11th Conf. On Applied Climatology
, Dallas, AMS (Boston), 215-220.


*Corresponding author address:
Mark Britt, National Weather Service, 12 Research Park Drive, St. Charles, MO 63304; e-mail: is the U.S. government's official web portal to all federal, state and local government web resources and services.