MKTG 7213 - Project 1

PROJECT 1: TORNADO ACTIVITY

MKTG 7213 - Dr. Greg Boller

Submitted by Indradi Soemardjan on February 12, 2001

Objective

To explore possibilities regarding further development of advanced warning systems with modern communication technology that would benefit the Southern region of the United States.

Scope of Work

Using historical data provided by National Severe Storms Forecast Center, the followings tasks are to be accomplished:

To compare tornado "casualty” numbers during the hours of the day when most people are likely to be awake versus the hours of the day when most people are likely to be asleep.
To describe any tornado casualty numbers’ differences across 11 states in the Southern region of the United States.
To compare “tornado” appearances during the hours of the day when most people are likely to be awake versus the hours of the day when most people are likely to be asleep.
To compare the number of appearances of each tornado strength classification and the number of casualties each may have caused during a specific time of the day.

Data Source

CODED RAW TORNADO DATA - by U.S. State and Territory

http://www.spc.noaa.gov/archive/tornadoes/index.html

DECODING SPC TORNADO HISTORY DATA

http://www.spc.noaa.gov/archive/tornadoes/tdecode.html

Data Analysis

Using selected raw tornado data between 1985 and 1995 for 11 Southern states (Alabama, Florida, Georgia, Indiana, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee and Texas), describing date and time of occurrences, damage levels, strength classifications and numbers of fatalities and injuries in each state, we were able to calculate the number of casualties based on this premise:

Fatalities + Injuries = Casualties

With basic arithmetic and spreadsheet formulas, we simply grouped the data as follows:

Just by looking at the graphs, we simply cannot make any inference about the fluctuations of the data. Texas, for example, had a total of 1128 casualties because 1740 tornadoes appeared in the area between 1985 and 1995, whereas Alabama had 1302 casualties (the highest total number of casualties of all 11 states) with only 247 tornado appearances. Therefore, to make the data more presentable and more meaningful, we calculated the casualty numbers for every 100 appearances of tornadoes.

State	Casualties per 100 appearances
AL	527
FL	68
GA	326
IN	183
KY	60
LA	103
MS	204
NC	277
SC	121
TN	241
TX	65

Clearly, between Alabama and Texas, there is something more to explore other than just describing the total casualties per year or per decade, because the numbers of casualties per 100 tornado appearances in both states are significantly different. Below, we attempted to understand when they mostly occured and when they actually caused the most casualties. Once the number of casualties per occurrence is calculated, we were then able to separate it into two different conditions:

C_awake: casualties that occurred during the hours of the day when most people are likely to be awake, that is between 07:01 and 20:59 (awake hours).

C_asleep: casualties that occurred during the hours of the day when most people are likely to be asleep, that is between 21:00 and 07:00 (asleep hours).

Now that we know most casualties occurred between 07:01 and 20:59 (awake hours), we felt that it was important also to know how many tornadoes actually appeared during that same period of the day.

Using the information above, we can see how these two findings have a direct relationship with each other because the high number of casualties caused by tornadoes in the awake hours is proportional to high number of tornado appearances during the same period of the day. Thus, this relationship has led us to form this first hypothesis:

“Most tornadoes appeared between 07:01 and 20:59 (awake hours) and most tornado casualties also occurred during awake hours, thus, we assume that tornadoes that appeared during the day have been responsible for most of the tornado casualties.”

To confirm the validity of the above hypothesis we attempted to find a relationship between the appearances of each F-scale and the number of casualties that is caused by each F-scale. The number of appearances of each tornado strength classification (also known as F-scale) and the number of casualties each F-scale caused can be best described by the following graphs:

From the graphs above we can conclude that, even though tornadoes with F-scales 3 and 4 did not appear very often, they have caused significantly much greater number of casualties compared to the ones with lower F-scales. To help quantify the number of casualties each F-scale could cause, the table below projects the possibility of the number of casualties for every 100 appearances of each F-scale:

F-scale	Casualties per 100 appearances
0	5
1	61
2	288
3	909
4	2640
5	N/A

A linear relationship does not exist between the F-scale and the number of casualties, but we do know that there is much greater safety risk involved when the F-scale reaches 3 and 4. Thus, this relationship has led us to form the second hypothesis:

“The higher the F-scale a tornado has, the higher the number of casualties.”

To relate to our findings earlier regarding the time of the day when most casualties occurred, we also analyzed the data to find the number of times each tornado with a specific F-scale appeared during the awake hours and summarized them as follows:

Our finding above shows that most of them actually did appear during the awake hours and thus enabled us to form the third hypothesis:

“Most tornadoes that appeared between 07:01 and 20:59 (awake hours) with an F-scale of 3 or higher have been responsible for most of the tornado casualties.”

Recommendation

Our analysis shows that between 1985 and 1995, most of the deadliest tornadoes occured during the day when most people are likely to be awake and such tornadoes had resulted in such a high number of casulties during the awake hours as well. Since we have only focused on the time when each F-scale type mostly occured, we are not able to actually come up with an idea for a new warning device product development until a further study on how and where most tornado casualties mostly occured in a specific areas of the 11 states are conducted. By knowing when, how and where high F-scale tornadoes mostly occured, hopefully, we will be able to understand if the current warning signals have been effective in saving people's lives.

* Link to MS Excel spreadsheet file in a WinZIP self-extracting format: tornado-work2.exe