Can Big Data Analytics And GPS Sensors Predict Earthquakes And Save Lives?

  • Blog
  • Smart Buildings

Can Big Data Analytics And GPS Sensors Predict Earthquakes And Save Lives?

After a devastating 7.8 magnitude earthquake struck Turkey and Syria, killing over 40,000 people, it’s worth asking if it will ever be possible to predict earthquakes and mitigate their harrowing consequences. Though it’s still early, there is enough progress to interest leaders in real estate and EHS.

This blog describes the state-of-the-art approaches to using big data to predict earthquakes. For years, researchers, seismologists, and statisticians have been attempting to combine big data analytics and the use of seismometers, satellites, and even animal behaviour to forewarn public and private leaders of imminent earthquakes. What may have once seemed impossible may soon be achievable.

Most recently, researchers have sought to forecast the likelihood and severity of earthquakes in more detail. Through analysis of seismic data, seismologists can monitor and closely examine the changes in frequency, scale and duration of seismic signals that could indicate a looming earthquake. While the United States Geological Survey (USGS) stresses that earthquakes cannot yet be predicted, existing faults and historic earthquakes can, in fact, provide critical information about future earthquakes. A team of seismologists and statisticians at Northwestern University have recently announced a model that uses long-term fault memory to predict when and where the next earthquake might strike – taking into consideration the exact order and timing of past earthquakes.

Furthermore, research teams are using satellite imagery and GPS sensors to understand the complexities of the Earth’s surface and generate early warnings. Geophysicists at the University of Bristol have tested interferometric synthetic aperture radar (InSAR) via satellites in order to detect changes in the Earth's surface around East African volcanoes. This method can monitor ground changes and deformations that might be signs of an impending earthquake.

Case studies from around the world reflect advances in the use of geodetic data for earthquake monitoring. For example, the USGS’s Prompt Assessment of Global Earthquakes for Response (PAGER) system used pixel tracking results from satellite imagery to raise the alert level for a large earthquake in Palu, Indonesia from yellow to red due to the area’s high shaking exposure. The system can raise alarms within 20 minutes about the impact of earthquakes of magnitude 5.5 or higher.

Scientists are also assessing whether changes in animal behaviour can provide warning of earthquakes. Snakes, fish, birds, dogs, cats, and elephants are all reported to behave differently before an earthquake. These responses may be triggered by changes in the Earth’s magnetic field or other environmental changes such as the ionization of the air caused by large rock pressures in earthquake zones.

Researchers from the Max Planck Institute of Animal Behaviour, in collaboration with the Centre for the Advanced Study of Collective Behaviour, have investigated this in more detail. In an earthquake-prone area of Italy, animals that had already exhibited unusual behaviour prior to earthquakes had accelerometers fitted to their collars. Then, for several months, researchers regularly documented their movements. In this time, the region had about 18,000 earthquakes. The 2020 research paper on this study revealed that a “collective of domestic animals repeatedly showed unusually high activity levels before earthquakes, with anticipation times (1–20 hr) negatively related to distance from epicentres (5–28 km)”.

An animal earthquake early warning system could resemble that developed by the Italian team: a chip on the collar channels movement data to a centralized database every three minutes. If it detects significantly increased animal activity for at least 45 minutes, a warning signal is sent. Further animal observations must be made over a longer length of time in more earthquake-prone regions before animal behaviour may be adopted in earthquake forecasting.

Many platforms are emerging to assess earthquake trends, such as DeepShake, which uses machine learning to assess seismic signals in real time and issue advanced warnings. Other examples include Early Warning Labs (EWL), a US based startup that has partnered with the USGS. EWL is currently developing an earthquake early warning system (EEW) for the West Coast of the US called ShakeAlert, aiming to inform citizens, first responders, and engineers about where earthquakes occur, the ground shaking intensity in different locations, and likelihood of major ground shaking. EEW can provide alerts before the arrival of the S-wave: the strong shaking that often causes most of the damage.

Earthquake prediction and mitigation is still a relatively nascent and complex space that requires ongoing research, development and investment; that R&D is increasing. EEW systems could revolutionize responses to earthquakes, enabling businesses, policymakers, and international institutions to reduce their devastating impacts and save countless lives.

Maya Hilmi


Maya is a Net Zero, Climate Risk Analyst. She is currently specialising in carbon management, ESG regulations, and identifying climate risk solutions. Prior to joining Verdantix, Maya interned at Cardano Advisory where she gained experience in covenant, sustainability, and pensions corporate finance matters. Maya holds a master's degree in Conflict Resolution in Divided Societies with Distinction from King's College London, and an undergraduate degree in International Relations from SOAS, University of London.