Following the idea of a Twitter wave from xkcd (
http://xkcd.com/723/), I decided to make a visualization of the earthquake in the Touhoku region of Japan to see what credibility this idea had.
I search for the words for "shake", "tsunami" and "burn", and then GeoLocate the Tweets using Google. I don't use the words earthquake nor hash tags because it was so soon after the earthquake, and many tweets that used the full word "earthquake" were written by people merely observing that there was an earthquake.
An interesting point to note is that in Japan, messages are shown on all major networks across the country when an earthquake has happened, or a tsunami warning is in place. An interesting video to watch is (
http://www.youtube.com/watch?v=eOrAwvJLKxo&feature=player_embedded), which shows how the major 6 networks reacted to the earthquake. The national public broadcaster NHK had shown the earthquake info 8 seconds after the quake, the other networks took a few minutes.
Another point to note is the geographic distribution of Tweets involving the word "shake", versus "tsunami". While "shake" is centered around areas that were actually shook, "tsunami" is more widely spread. This is of course merely a hypothesis by myself, but I believe this is the effect of people only tweeting about the "tsunami" after a national tsunami alert went out, as opposed to feeling the effect of one. This is especially interesting given how soon after the earthquake this phenomenon manifests itself.
Special thanks to Facebook Hackathon and Rasmus Lerdorf for helping me with this code.
