Napa earthquake: An earthquake in a highly connected world

The Napa earthquake recently occurred close to Silicon Valley. This makes it a good candidate to study what social networks, wearable objects and website traffic analysis (flashsourcing) can tell us about the way eyewitnesses react to ground shaking.

In the first part, we compare the ratio of people publishing tweets and with the ratio of people visiting EMSC (European Mediterranean Seismological Centre) real time information website in the first minutes following the earthquake occurrence to the results published by Jawbone, which show that the proportion of people waking up depends (naturally) on the epicentral distance. The key question to evaluate is whether the proportions of inhabitants tweeting or visiting the EMSC website are similar to the proportion of people waking up as shown by the Jawbone data. If so, this supports the premise that all methods provide a reliable image of the relative ratio of people waking up.

The second part of the study focuses on the reaction time for both Twitter and EMSC website access. We show, similarly to what was demonstrated for the Mineral, Virginia, earthquake (Bossu et al., 2014), that hit times on the EMSC website follow the propagation of the P waves and that 2 minutes of website traffic is sufficient to determine the epicentral location of an earthquake on the other side of the Atlantic. We also compare with the publication time of messages on Twitter. Finally, we check whether the number of tweets and the number of visitors relative to the number of inhabitants is correlated to the local level of shaking.

Together these results will tell us whether the reaction of eyewitnesses to ground shaking as observed through Twitter and the EMSC website analysis is tool specific (i.e. specific to Twitter or EMSC website) or whether they do reflect people’s actual reactions.