Real-Time Mapping of Earthquake Perception Areas in the Italian Region from Twitter Streams Analysis

Abstract

The mapping of earthquake perception area is useful to determine how many people have felt it. This is an important issue because even moderate magnitude earthquakes can affect critical communication infrastructures. However theoretical estimates of instrumental intensity distribution, derived from ground motion parameters (e.g. ShakeMaps), maybe poorly correlated with the actual earthquake perception. Furthermore, the number of people who felt the earthquake depends strongly on the spatial distribution of the population density. In recent years there has been a growing interest in the data mining of citizen-provided information from social networks, Internet accesses and web-based macroseismic surveys aimed at detecting, locating and characterizing the macroseismic field of moderate and strong earthquakes. Here we propose a strategy to retrieve in real-time useful information about the area where an earthquake has been perceived and how many people felt it, using data mining of Twitter streams. We show that using a proper normalization of these data allows a quantitative definition of an Earthquake Perception Index based on Twitter posts (EPIT). This index shows a good correlation with ground motion parameters and macroseismic data and hence allows a rapid but realistic mapping of the perception area.