The world’s leading publication for data science, AI, and ML professionals.

Will The Next Hurricane Hit My Home?

A Data Analysis Based On Historical Storm Trajectories

Photo by Shashank Sahay on Unsplash
Photo by Shashank Sahay on Unsplash

We’re in the midst of a very active Hurricane season with hurricane Sally making landfall last night and several other tropical storms brewing in the Atlantic Ocean. The big question on everyone’s mind is always: "Will the next hurricane hit close to my home?". Despite many excellent advances made in meteorology, tropical storms remain very difficult to predict. Both the path and the intensity can change significantly throughout the lifetime of a storm.

Another question one can ask is: Of all the past tropical storms that have ever passed through a geographical region (like a city) what did the track look like? In other words, take a city like New Orleans and make a geographical plot of all the hurricanes that passed through it or came very close to it:

(image by author)
(image by author)

To answer this question in full generality, we need a dataset that is as complete and accurate as possible for all tropical storms. NOAA maintains such a dataset as a collection of shapefiles, which the Wolfram Language can easily import. The code that transforms the raw data from NOAA into a computable dataset for the Wolfram Language. It can be queried just like any old dataset. For example, here I am asking for storms named "KATRINA" in the "2005" season and it returns all the rows that are relevant:

(image by author)
(image by author)

To get all storms that came close to New Orleans, we can look up the latitude and longitude for that city and select the rows of the dataset that have a matching coordinate:

Interpreter["City"]["New Orleans, Louisiana"]["Position"]

This returns with GeoPosition[{29.9728, -90.059}]. Next, we can look for paths that come close to that position and return the list of named storms and seasons. The resulting dataset lists all the storms that passed through New Orleans:

(image by author)
(image by author)

With these names, you can then plot the paths for each storm. This is what is shown in the image at the beginning of this story. A more zoomed-in map shows how many storms have crossed paths with New Orleans:

(image by author)
(image by author)

There are of course no simple conclusions to draw from this data. A few storms that hit New Orleans started very far away in the Atlantic Ocean. But many more started much closer in the Caribbean and the Gulf of Mexico.

The code can easily be adapted for any location, such as Miami. In that case, the geographical plot is as follows:

(image by author)
(image by author)

To get the full code, you can open this Wolfram notebook in the cloud.

Photo by Morgan Petroski on Unsplash
Photo by Morgan Petroski on Unsplash


Related Articles