Yes, quakes can be forecast

Earthquakes would be less scary if they could be forecast, like the weather. Small quakes would no longer be startling, and there could be early warning of bigger quakes, preventing a greater loss of life.

In seismic zones like Nepal, it is not a question of ‘if’ but ‘when’ an earthquake will strike. And most gadgets that detect earthquakes are either false alarms or come just a few seconds before the violent shaking starts.    

Earthquake Early Warning Systems (EEWS) are instruments which can buy people 5-60 seconds before an earthquake hits, depending on distance from the epicenter. This is not a lot of time, but often enough to get out of a building and head to safety.

EEWS do not forecast an earthquake, but the sensitive instruments detect the initial P-waves of an earthquake and send out the warning before the more destructive S-waves arrive. Such systems are used in Japan to stop bullet trains and to make lifts stop at the next floor. Installing the relatively inexpensive systems in schools and other public buildings, can provide critical seconds for building evacuation. 

Although they do not provide much time, the most reliable method is the EEWS. How much early warning these systems can buy depends on the density of the sensor network, how fast the data is transmitted, and distance from the epicenter. As seismic waves travel through speeds of 1-5 km/second, EEWS is not much use in the areas right around the epicenter, but can be life-saving further away.

Such a system is already in place now in many countries, but has not been used much in Nepal. If a proper system had been in place during the 2015 earthquake epicentered in Gorkha, Kathmandu could have received a warning 30-40 seconds prior, says earthquake hazard expert Prashant Rawal of the National Society for Earthquake Technology - Nepal (NSET).

“Before we can implement a functioning warning system, we need to have a denser network of sensors,” says Rawal, “Right now, we only have about 40 sensors setup, so we are still in the first stage of network establishment.”

Japan’s EEWS system, for example, features over 1,000 sensors, while China has more than 150,000 monitoring stations. The sensors detect the early P-waves of an earthquake, and then use later S-waves to update the magnitude. Alerts are then broadcast through multiple channels, including mass media, phones, apps, and sirens.

Upon receiving a signal, trains are programmed to stop at the nearest possible station, and elevators at the closest floor. The system is not perfect, sometimes underestimating and other times overestimating the strength of a quake, but it worked well enough to give Tokyo 30 secods in the 9.0 Sendai earthquake in 2011, and 10-20 seconds in the 2016 Kumamoto Earthquake.

In Nepal, there is no clear policy from the government for such a project, which makes it difficult to fund -- although NSET has worked with the local and federal governments to place sensors.

Other problems are largely technical. In regions like Karnali and Surkhet, fluctuations in voltage and lightning strikes often damage sensors, and even when they do detect shaking, there is no reliable internet to transmit these signals.

“We have to establish these sensors in hill or mountain areas, and make long trips in order to maintain or repair them. There is often a lack of budget to make these trips,” says Rawal.

Duke University and Tribhuvan University are collaborating to set up a smart seismic sensing network in Kathmandu, create a mobile messaging system for early warnings, and develop data science models from previous seismic data.

There is now promise of even more accurate prediction systems that use Artificial Intelligence. Researchers from the University of Texas at Austin trained AI and were able to develop a model that predicted 70% of earthquakes during a seven-month trial in China, with accurate forecasts of magnitude and epicentre. 

Another traditional prediction method is to observe animals which have a sixth sense to feel earthquakes before they happen. Other theories include measuring the emission of the radioactive gas radon which is trapped underground and is released as the ground shifts before an earthquake. But there is the issue of false positives: the release of the gas may be triggered by rainfall. 

However, while the developers of these theories have always claimed to have predicted certain earthquakes, none have been able to do so reliably. Some have been debunked, after failed forecasts, or due to shoddy methodology.

EEWS sensors use accelerometers which detect how fast the ground moves. If three to four accelerometers spaced at a distance apart all detect the movement of the ground, then it is likely that an earthquake is happening. 

The magnitude and approximate location of the earthquake is then calculated using computer algorithms, and a message would be sent out if the quake is deemed to be large enough. It is not a lack of budget that is holding back using EEWS: the sensors cost only $200-500. “The system could be certainly implemented if the necessary political will was there,” says Rawal, presumably from the government.