This is a study of the spatial
distribution of tornadoes in Kansas and Oklahoma, the middle of "Tornado
Alley". Tornadoes do not occur in just part of this region, but rather
their distribution changes with the progression of time. In order to best to
determine what areas are most at risk, it is necessary to investigate where
tornadoes have occurred and if the locational trends will allow for the
prediction of future tornado occurrences. The States' of Kansas and Oklahoma
desire to build storm shelters in areas likely to encounter more tornadoes. In
order to determine those areas, it is necessary to weigh tornado frequency,
tornado size, as well as changing weather patterns throughout time.
To determine the locations of future
storms, several points and distances will be calculated. The Mean Center of The
tornado locations will be calculated, to determine the average location for all
tornadoes in a time period. This will provide a simple midpoint from which
future observations will be based. Another point to be calculated is the
Weighted Mean Center; a more complex average of the tornado locations based off
of their widths. After the mean center and weighted Mean center are
calculated, the standard distance and weighted standard distance may be calculated.
The standard distance is the spatial equivalent to the standard deviation, and
allows the map-reader to more closely identify the spatial trends of the data
around the mean center. The weighted Standard distance shows how data is
distributed around the weighted mean center.
These values will be calculated from
data sets of tornado locations between 1995-2006, and 2007-2012. A data set
showing the number of tornado occurrences per county between 2007-2012 will
also be used.
As time progressed, the distribution of
tornadoes varied, but not by much. In figure 1, tornadoes are seen in three
groups: a band running from east to west across the middle of Oklahoma, a bear
and running from north by northeast to south by southwest from Kansas to
Oklahoma, and a third band and running from Northeast to Southwest across
Kansas to the panhandle of Oklahoma. In figure 2, only two major groups are
seen. There is a line of large tornadoes that occurred across Oklahoma in a
pattern following east by north east to west by Southwest pattern. The second
line of storms stretched almost perfectly east to west with the epicenter of
storm activity directly underneath the weighted mean center. Figure 3 indicates
that when will the locations of tornadoes they have changed throughout the
different time periods, the location around which they are centered had not
changed. Comparison between maps four and five shows how the distribution
of tornadoes has become more centralized in recent years. In figure 6 this more
visibly clear, as a majority of all major tornadoes occurred in the region of
overlap between The 90s and 2000s weighted standard distances. Figure 7 shows
how the frequency of tornado occurrences compares to the mean of 4.0, with a
standard deviation of 4.3. Figure 7 shows that the concentration of tornado
occurrences is most centered in Kansas, just north of the border with Oklahoma. The calculations suggest it is highly unlikely
any counties will sustain the occurrence of more than 8.3 tornadoes within a
five year period, but most will encounter at least four within the same time
period.
The result of this analysis shows an
extremely varied distribution of tornado locations throughout the region of
Oklahoma and Kansas. The patterns of storm system occurrences have varied
significantly, but within a limited area. When collating the weighted standard
distance of the 1990s tornado locations with the weighted standard distance of
the 2000s tornado locations, it becomes more evident that the area where the
two intersect is a hotbed of storm activity. While many of the counties do not
wish to build storm shelters, it would seem prudent to require them in the
counties with tornado counts well above average, and strongly suggest the
construction of them in the other counties within both weighted mean centers.
 |
| Figure 1 |
 |
| Figure 2 |
 |
| Figure 3 |
 |
| Figure 4 |
 |
| Figure 5 |
 |
| Figure 6 |
 |
| Figure 7 |
Mean: 4.0
SD: 4.3
Russell County: Count – 25,
Z-Score – 4.0837, Probability - outlier
Caddo County: Count – 13,
Z-Score – 2.093, Probability – 1.83%
Alfalfa County: Count – 5, Z-Score
- .2325, Probability – 40.9%
70% of the time more than 1.764 tornadoes will occur.
20% of the time more than 7.612 tornadoes will occur.
