Introduction:
I have
been hired by Tornado Valley Research Institute to study the geography of
Tornadoes in Kansas and Oklahoma. Recently there has been some arguments over
the distribution of Tornados and if locals should be required to build storm
shelters. Proponents of the storm shelters argue that spatial patterns of
Tornadoes have not changed that much over the last twenty years and because of
that, they should build more shelters in areas that receive more tornadoes.
Others argue that shelters are a waste of money and not every “place” ever sees
tornadoes.
So the
focus of my study will be to analyze the placement of Tornadoes over two separate
time periods (1995-2006 and 2007-2012). I
will look at the varied placement of these Tornadoes and how large they were
when recorded (Not the Fujita Scale but rather width in feet). This will give
me a good idea of specific locations that have received not only a large amount
of tornadoes but also those that were rather large in size and therefore pose
more of a threat. Additionally I will look at the placement of tornadoes at the
county level of these two states.
Methodology:
To do this study I used
ArcGIS to map out these tornadoes how we can best understand them spatially. I
was given data of both Kansas and Oklahoma and how many tornadoes each of their
counties recorded from 2007-2012. I also have the individual data of each
Tornado and its width in feet from 1995-2012. Those dates have been broken up
into 1995-2006 and 2007-2012.
To help
me better understanding how the spatial pattern of Tornadoes has changed over
time I will be looking at the mean center of the Tornadoes over those two time
periods. The mean center gives me the geographic middle of all the data that
was collected. By looking at the mean center over the two time periods we can
see if and how the pattern has moved over time. I also looked at the weighted
mean center of the Tornadoes width. This is the same function as above but this
time I add another parameter to the tool so that it weights the wider tornadoes
more than the smaller ones. This is especially important as it shows how the
pattern has changed over time with respect to the larger, more damaging
tornadoes.
 |
| Figure one |
Another measure I took to study the
pattern of tornadoes was the Standard Distance of the Tornadoes. In its most
simple form, standard distance draws a circle around the mean center up to the
first standard deviation so that theoretically 68% of the tornadoes will lie
within that circle. If I were to go up another standard deviation it would draw
a circle around 95% of the tornadoes. The standard distance is a good indicator
of where in our study area we have the most data associated with our area. In
this study I based the standard distance upon the weighted mean center and not
the mean center to give more of a priority to the larger tornadoes.
 |
| Figure two |
The last item I looked at was the
standard deviation of the tornadoes at the county level. This was the most
basic map I created and shows which counties from 2007-2012 had the most
tornadoes and how they deviated from the mean. You can clearly see the counties
that drove the mean up as they are indicated by dark blue.
Results:
After running all the tools
in Arc, these are the various maps that I created.
 |
| Map one |
Map one shows the location of Tornadoes
from 1995-2006. I also showed the Mean Center and the Weighted Mean Center of
this dataset. You can see that once the mean center is weighted, it shifts to
the South due to the abundance of large tornadoes further south. Some patterns
that stick out to me are the cluster of tornadoes in the northern central
region of Oklahoma. This cluster however doesn’t draw my eye as much as the absence
of tornadoes in the Southeastern region of Kansas. From this data set I pick
out 7 counties that recorded zero tornadoes in this timespan of roughly 10
years and all 7 of those are in that absent region of Kansas.
 |
| Map two |
In map two we see the same components
as in map one but this data is from the tornadoes recorded from 2007-2012. What
we can decipher from looking at this map compared to the last map is that the
areas of clustered tornadoes have shifted elsewhere. I now see two separate areas
of larger tornadoes, one lateral stretch though the middle of Oklahoma and
another cluster in the central region of Kansas. I also infer from this map
that the severity of Tornadoes has relatively dropped from the first map to the
second. This is peculiar because you always hear how nowadays with climate
change that we will see bigger storms than in the past, which is not the case
here. I call attention to the Eastern region of Kansas where in the first map
we saw a lack of tornadoes. We still don’t see many tornadoes there but the
ones we do see are those that are larger than 280 feet in diameter.
 |
| Map three |
Map 3 illustrates both sets of data
put together. If you can look past the busyness of the map you will see that in
the past 20 years there has been tornadoes all over Kansas and Oklahoma and
every single county has experienced at least one. You can tell that there have
been more tornadoes recorded in the central region of the bi-state area but relatively
less in that eastern region of Kansas.
 |
| Map four |
 |
| Map five |
 |
| Map six |
Maps 4 and 5 both show us the
standard distance of both data sets while Map 6 shows us both overlaid onto one
another. If I may so dub each of the regions covered by the standard distance
in these maps the disaster zone, you can see the disaster zone over time has
shifted slightly to the northeast. I believe this is due to the large increase
of tornadoes in southern Kansas versus Northern Oklahoma and to the presence of
tornadoes in the once absent region of Eastern Kansas.
 |
| Map seven |
In my final map (Map 7) I have showed
the standard deviation for all the counties in the bi-state area. This data is
only for the 2007-2012 data set and you can see those blue counties in southern
Kansas as deviating from the mean in a drastic way. This map further reinforces
the abundance of tornadoes in the central regions and the lack of them in
eastern Kansas. From receiving the standard deviations of the counties I also
calculated the Z scores for three of those counties. Caddo County, OK, Alfalfa
County, OK, and Russell County, KS were the three. I found that over the
recorded period of time we can say that Caddo County will have 13 tornadoes
1.83% of the time, Alfalfa County will have 5 tornadoes 40.9% of the time, and
Russell County will have 25 tornadoes less than 1% of the time. Pretty much
this is saying that Russell County is likely to not record that many tornadoes
in 5 years ever again while Alfalfa County is somewhat likely at 40.9% to
record 5 tornadoes again in that time span.
Conclusion:
My findings have led me to
believe that even though some areas of the bi-state region experience a
dramatically lower rate of tornadoes, you cannot ever be sure that another
tornado will or won’t strike the same general location again. It is my
suggestion to these areas will high frequencies of tornadoes that they
immediately build shelters that can withstand a substantial amount of people
and a substantial tornado as those areas that experience the higher frequency
general experience larger tornadoes as well. To those areas with low
frequencies, I understand that you don’t record as many tornadoes as other
areas do but you cannot ever say for certain that one won’t hit your hometown,
your farmhouse, or your family. It is my recommendation that you too build
shelters, however, they do not have to be to the standard of the more prone
areas shelters, but should still be functioning in the event of a freak storm.
If I were you, personally, I would build the best shelter possible no matter
where I lived to protect my family in case of an emergency because when the
time comes, it will be too late to build one when it is actually needed and too
many good people have lost their lives to this beast of mother nature and
having said that, we should give these tornadoes the respect they deserve and
take the appropriate measures needed to keep our families safe.
I hope you find this study useful in making your final
decisions and I would like to thank the Tornado Valley Research Institute for
allowing me to conduct this study in the hopes that I could help you all make
the best economic decision and the safest one.
