Almost immediately after finishing the last post I began to wonder if maybe there might be a correlation between reports of explosions and similar incidents. For example, if there are many reported gas leaks in an area is there also an increase in the number of reported explosions?

The first thing I did was to give the map an edgy “night” look. Then I figured out how to make heat maps because using points with some data sets just pollutes (covers) the map.

The first result was a white-to-blue heat map of reports of natural gas leaks overlaid with yellow-ish points indicating reports of explosions, not unlike the previous map.

Nothing obvious stood out to me.

Next I overlaid a green heat map of chemical hazard reports with the same yellow-ish points.

This time there seemed to be a cluster of explosions and hazard reports on the East end of the city’s core.

Mildly interesting.

Finally, I made a caustic black-to-yellow heat map of industrial accidents and added the same explosion locations on top.

Two or three explosions sit comfortably within clouds of industrial accident reports but otherwise I don’t see anything obvious.

So right now, chemical hazard reports on the downtown east end seem to be only gently correlated to reports of explosions. I guess it’s a little more than nothing.

Why did I choose gas leaks, chemical hazards, and industrial accidents? Simply because they make good heat maps and they’re the only incident types out of the 90 that appear in the data that seem directly related to stuff exploding.

There are lots of fire and vehicle collision reports that produce very busy maps so I think in these cases it might be more useful to get a little more refined, a little more fancy. I’m thinking along the lines of reports that are similar not only in location but also in date and time.

As you can see, the data displayed below is available as CSV-formatted files. They’re pretty straightforward, just be aware that dates and times are in the ISO 8601 format at GMT/Zulu time so they won’t match what you see displayed on this page until adjusted for timezone and DST.