Dashboard Week: Day III - Tornadoes

by Lukas Jennings

We're now 3 days into dashboard week and the data sets just keep on coming! To continue with the theme of environmental data, coach Andy set us the task to visualise some tornado data.

My Approach

After an initial scan through the data set, I could see that there was not much data prep involved here. After a quick formula tool and datetime conversion, the data was ready to be outputted into a hyper file. The workflow can be seen below:

Alteryx Workflow

The first element of comparison I wanted to be able to draw from was the frequency of tornadoes occurring during the day time vs night time. I did this by creating a dual axis line/area chart with a trend line to see how this developed over time, both at night and during the day. Instead of placing both night and day dimensions on the same chart, I opted to use a parameter action with shape buttons to switch between the graphs. Both these charts highlighted an increase in the number of tornadoes over time.

This parameter action was also applicable to the graph directly underneath, which looked at the correlation between magnitude and damage caused by tornadoes in each state. In this case, there was a particularly strong positive correlation between damage and magnitude at night time.

I then created a tornado tracker using the pages feature in Tableau. I found this interesting because it clearly showed how central states, particularly Texas, have been constantly affected by tornadoes since the 1950s.

I found out that this area of the US is heavily affected by tornadoes due to a clash of warm, moist air from the gulf of Mexico and colder, drier air stemming from the Rocky Mountains. I wanted to be able to visualise this with additional data, but I was unable to do so.

The final chart was aimed at finding out whether the magnitude of tornadoes had been increasing over time. The answer is, no it hasn't. The line/area chart shows a downward sloping trendline to demonstrate that the severity of tornadoes has actually been declining since the 1950s.

What could I have improved?

  • Ideally, I would have drilled down further into why these tornadoes were occurring so much in the US, and what was driving the steady increase as well as why magnitude had declined over time. I decided from the start of the day that I wanted to add supplementary data in order to answer these questions.
  • Despite finding temperature data from the Gulf of Mexico, I was beaten by the deadline and was unable to add this into the dashboard and I lacked the context to be able to draw enough insights as to be able to answer the appropriate 'Why?' questions.

My Dashboard

A link to this dashboard on Tableau Public can be found here...