Thinking about the hurricanes in the US right now, and watching like everyone else the projected paths they are supposed to take. The overlap of the lines got me thinking if there was data for the path of previous hurricanes available.
I found what I was looking for by first visiting Historical Hurricane Tracks which led me to data.noaa.gov and finally to some CSVs. The file I looked at (500MB) had some 12,000 tropical cyclones and their data going back to the 19th century.
I knew I wanted to draw their paths, the file had their lat/long decimal degrees for each storm to the second decimal place. I did not want to draw them onto a map though. I wanted something more abstract, the paths disconnected from their geographical coordinates. I found someone had converted a google earth function to go from lat/long to pixels at a specific zoom level to python. This let me translate all the storms paths into pixel paths and store that as reduced (4BM) JSON file.
Using some simple HTML5 Canvas commands I draw the paths onto a 1000x1000 canvas. The line color is based on the decade of the storm while the starting position, thickness and number of storms painted is random.
Take a look at the results: https://thisismattmiller.github.io/hurricane-track-draw/