Data Storytelling and Data Viz Approach
I really enjoyed this dataset, and probably went a bit overboard. I've always found the hurricane track maps, and especially the forecast models to be very interesting. The cone of probability and multiple spaghetti models with their underlying algorithms and forecasts make for an engaging, albeit busy, visualization. Here, we're only looking at the historical tracks and details of past storms, but the visuals can still be quite stunning.
It shouldn't be a surprise that I've incorporated maps into this visualization. I wanted to include the typical storm tracks map, but also wanted to answer some questions I had when looking through the dataset - where do most storms originate, and where do most storms make landfall? We'll talk about the storm origin first - or maybe it's better to call it the first observation locations. This was relatively easy to accomplish. I used the latitude and longitude columns, and plotted only the details for the first observation.
Courtesy of satellites and other technology, in recent years, the first observation is most likely the origin of the storm. However in the older data, the first observation truly is the first time that the developing storm system was chanced upon by humans. By choosing different years across the decades (and maybe zooming out a bit), you can see that more recently, we detect storms as they develop off the coast of Africa. In the late 1800's, most of the first observations are much closer to North and Central America.
I also thought it would be interesting to see where storms made landfall. Once again, this was pretty easy to accomplish by plotting points on the map and filtering for an Event Type of 'L' for landfall. On both of these maps, I filtered the values down to the decade, using color and size to highlight the data tied to the selected year, but still showing the relationship between the storms in the selected year and the others in the decade. I used a parameter as a sheet selector (in conjunction with some filters) to allow the user to choose which map type they want to view.
To drive some additional interactivity with the maps, I added a very small viz with a listing of all of the storms for the selected year. When the user clicks on a storm name, it highlights the pertinent data on the chosen map. Framing the maps, I used a set of BANS to show the number of storms for the chosen year by the maximum storm type. To do this, I created a set of calculated fields to work through a hierarchy of the storm types to count the storm at its strongest - the maximum type for that storm. Completing the frame, I used simple bar graph to show the relation between the total number of storms by year for the selected year and the rest of the decade.
I created several calculated fields, mostly to deal with converting the data that was in the original dataset into more meaningful data elements - extracting dates, accounting for '999' where observations didn't exist, determining the hurricane categories, etc. Overall, this viz was a lot of fun to create!
This definitely is not the first time that hurricane data has been visualized in Tableau, but we're curious what new insights or visualizations you can find and create with this dataset. Take some time to post your work to Tableau Public and Twitter with the hashtag #ThrowbackThursday, tagging @TThrowbackThurs. We'd really love to see what you can come up with!
This week's dataset comes from the National Hurricane Center. Check out the Data Dictionary for column definitions! Please be sure to cite the source on your viz.