The aim of an earthquake early-warning (EEW) system is to provide alerts to people, devices and automated systems that shaking caused by an earthquake is about to occur, ideally with an indication of the expected shaking intensity and time of arrival.
Unlike earthquake prediction, which aims to specify details of an earthquake before it occurs, but is not considered possible, an EEW system uses a network of sensors to detect an earthquake once it has already begun and sends alerts to locations before the waves caused by the earthquake reach them.
EEW systems can provide from a few seconds to a couple of minutes of advance warning, depending on the distance between where the earthquake is detected and where the alert is sent to. This helps to save lives by giving people time to seek cover and also enables automated systems to be triggered that reduce the risk of damage, such as turning off machinery and slowing down trains.
As discussed in our article The Earthquake Bell, the concept of such an EEW system was first proposed in 1868 for the San Francisco area by Dr JD Cooper, but it was not until 100 years later that they started to become a reality.
Following the devastating 1985 earthquake in Mexico City, an EEW system was developed for Mexico City that began issuing public alerts in 1993, making it the first EEW system in the world to do so. Since then it has been expanded to cover areas of central and southern Mexico.
Mexico’s EEW system was inspired by Japan, which has pioneered the development of EEW systems since the 1960s when its national rail company began developing an earthquake alarm as a way to reduce the risk of derailment during earthquakes. Despite this, Japan’s nationwide EEW system only began issuing public alerts in 2007. A notable success came on 11 March 2011 when the system successfully provided alerts for the magnitude 9 earthquake that struck off the coast of Tohoku.
Besides Japan, the only other country with a nationwide EEW system is Taiwan, which began issuing alerts to key agencies in 2014 and then to the general public in 2016.
An EEW system is currently being developed for the west coast of the USA, with the first alerts being available for the LA area since the start of 2019 by means of a smartphone app.
Despite the potential of EEW systems to save lives and reduce the damage caused by earthquakes, the majority of seismic regions in the world do not have one. With a price tag ranging from tens to hundreds of millions of dollars, if not more, an EEW system like those described above is a luxury that many earthquake-prone countries cannot afford.
However, recent advances in Internet of Things (IoT) and cloud technology have opened up the possibility to develop lower-cost and rapidly-deployable alternatives that can significantly expand the reach of EEW systems around the world.
Our mission at Grillo is to turn this possibility into a reality. During the past two years, we have developed our own IoT- and cloud-based EEW systems in Mexico and Chile that have been issuing alerts since March 2018. Our sensors, which are small in size and easy to install, are constantly sending unprocessed accelerometer data to the cloud, where detection algorithms check for earthquakes and send notifications via our smartphone app and Twitter.
The reduced cost of our sensors compared to traditional seismic sensors, their ease of deployment and our use of the cloud mean we are able to set up EEW systems in new regions quickly. Indeed, Grillo’s EEW system in Chile emitted its first alerts within weeks of its sensors arriving there.
Our experience so far has confirmed the potential of an IoT-based approach to EEW systems to bring life-saving alerts to new regions around the world. To expedite this process, Grillo will soon be launching OpenEEW, an initiative to share our data, sensor technology and detection algorithms.
The first step will be to open up our entire archive of unprocessed accelerometer data, including a magnitude 7.2 earthquake, enabling people to develop their own detection algorithms using cutting-edge machine learning models.