Earthquake Prediction Technology: An Innovation of American Scientists; Data being recorded for 31 years helped

Mail Us For Business inquiry
- Advertisement -

The US Geological Survey is often asked whether earthquake prediction is possible. The agency’s website states that no scientists have predicted any major earthquakes. But soon this answer of the agency is going to change. After decades of attempts and failures, a team of geophysicist Dr. Paul Johnson at Los Alamos National Laboratory has developed a tool that can make earthquake predictions possible. Machine learning technologies that range from converting speech to text to cancer detection are now being applied to seismology. Dr. Johnson explains that 10 cycles of earthquake data are needed to train the machine learning system. This is difficult work. For example, the San Andreas Fault in California produces a major earthquake every 40 years. But to understand this, sufficiently useful data is available only for the last 20 years (ie half-cycle). On this, Dr. Johnson and his team applied machine learning to different types of activities (slow-motion phenomena, also known as silent quakes) in 2017. Earthquakes usually end in a few seconds. But silent quakes can take hours, days or even months. It is better from machine learning perspective, as it gives massive amount of data. Anyway, Dr. Johnson’s lab is in San Andreas, a region of tectonic diversity. There is a slight slip in the plates every 14 months, scientists have been recording them since 1990. That is, there are lots of full cycles of data. The machine learning system was able to predict when they would happen again.

Could have saved 45 thousand lives: According to the US Geological Survey (USGS), more than 45,000 deaths have been caused by earthquakes worldwide in the past decade. These include the 2011 Japan earthquake, the 2015 Nepal earthquake, and the 2018 Indonesia earthquake. Experts believe that if predictions about them were possible, then these 45 thousand lives could have been saved.

Used in sensitive zones to estimate actual earthquakes
Lab quakes are miniature earthquakes that are produced in the lab by light pressure on glass beads. The team has also prepared a simulation (computer model), which is capable of capturing these movements. The team is training its machine learning system for earthquake prediction on this. It will be planted in the San Andreas Fault to find the actual geological fault. There was an earthquake of 6 Richter scale in 2004. Results will be available in three to six months. Johnson says it will be a revolutionary discovery.


- Advertisement -
Mail Us For  DMCA / Credit  Notice

Recent Articles

Stay on top - Get the daily news in your inbox

Related Stories