A Byte of my 2.2-lb Brain

Just sharing stuff…

Twitter Users as Seismographs

The idea that Twitter users generate volumes of earthquake-related tweets at the onset of an earthquake is hardly new [link], but it’s always fun to work with data from home (the Philippines) and report the findings. In October 2013, for example, I also collected tweets on the Bohol earthquake; but my focus then was more on the contents of the tweets.

According to DOST-PHIVOLCS, a Ms 5.4 earthquake struck parts of the Philippines, 019 km N 51° E of Looc (Occidental Mindoro, at 9:50 p.m. on 19 October 2015. In this  CNN Philippines news article, the writer mentions that “netizens noted the quake was felt in Tagaytay, Orani in Bataan, and Metro Manila.” That’s one interesting note there—“netizens” effectively operating as seismographs.

So, how fast do Pinoys respond to earthquakes on Twitter? The graphs below indicate that we’re pretty fast. The DOST-PHIVOLCS recorded the onset of the earthquake at 09:50:41 PM in their clock; on the other hand, the first tweet that my listener “detected” was at 9:51:19 PM—that’s less than a minute apart (38 seconds to be exact)! The left graph shows the number of tweets that contain at least one of the two terms [“lindol”, “earthquake”] at 15-min intervals, while the one on the right, every 10 seconds. “Lindol” is the Filipino term for “earthquake”.

earthquake-temporal

Temporal dynamics of tweets that contain the terms “earthquake” or “lindol”. The left image shows tweet counts at 15-min intervals. The one on the right counts tweets at 10-sec intervals.

Curious to see some of the very first few tweets? [**Warning** FOUL language]

lindol

 

Finally, the heatmap below shows where the tweets came from. Seismographs, right? 🙂

index


EXTRA. The video below shows tweet bursts; note the timeline at the lower-left corner of the video. Apologies for the datetime shift in the video clip; I didn’t notice it right away that cartoDB altered something in my dataset (should subtract 8 hours).


Acknowledgement

I’d like to thank @ianalis for pointing us to the article mentioned above. I’d also like to thank @baracomama and @avsolatorio for reminding me about this dataset. 🙂


Other Details

  • I only have the Twitter spritzer access, which only allows mining of 1% of randomly sampled public tweets for free. The other two types of access are gardenhose (10%) and firehose (100%).
  • I wasn’t specifically tracking the terms “earthquake” and “lindol”; but I have been continuously mining streaming tweets originating from the Philippines (thanks to DigitalOcean for the paid cloud service and to Ed David for helping me set up and maintain the listener).

One comment on “Twitter Users as Seismographs

  1. primehunter
    October 25, 2015

    erika, what accounts for the smaller heat spots (and their locations)? I can see the major urban centers- yung mga in touch as news, pero curious ako dun sa dulo sa ibaba, off saranggani.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Information

This entry was posted on October 24, 2015 by in Philippines and tagged , , .
%d bloggers like this: