A few days ago I found a wonderful story about polar ice caps melting that led me to some wonderful data. I thought I could potentially make a viz that showed the changing extent of the sea ice at the poles – some line charts for the temporal view and a map with the shape files. I figured some animation would be even sexier.
I pulled the shape files down from the web and used Alteryx to union them together into a single file (+1 vote for shape file union). I loaded it into Tableau and drew my map but I got this:
WTF!?! Something is very broken. I decided to take a closer look at a single shape file:
Yep – definitely borked somewhere. My initial thought was that maybe there was a problem with the polygon data – that perhaps the polygons weren’t being closed properly. Because look here…
But surely not. I mean, these people are professionals. I’m sure their data is used all the time and an error like this would certainly be flagged. I downloaded and installed QGIS, pointed it at the file (and the Tableau tile service) and voila! One sea ice polygon:
BTW – check out how QGIS takes our Tile Service and projects it nicely. Very cool!
Anyhow – it turns out the problem is with the projection in the data. The shape file has the data in EPSG:3412 (NSIDC Sea Ice Polar Stereographic South). However, Tableau only understands WGS 84 (Web Mercator) and so it is doing on-the-fly transformations. Here’s what happens when I transform the data into a Mercator projection in QGIS:
BOOM! Also borked. I’m no GIS expert (looks around for Sarah Battersby) but it looks like Mercator can’t handle polygons that cross the +/- 180 degree meridian. So until Tableau can support more projections, I’m going to have to park this project. Or like Sarah suggested, map it to completely different coordinates somewhere else on the globe.
Learnin’ every day, folks.