Even when we are resting, laying down, or sitting, our little planet and the ground beneath us are constantly moving. Many of us even forget the fact that without a second’s delay she is doing her job so impeccably.
The ground underneath our feet is not so stable as we think, there are tectonic plates moving past each other. Sometimes colliding with the neighbouring plates causing cracks on the surface and bursts which could even lead to natural disasters. It is occasionally so powerful that these quakes are capable of demolishing everything on the surface. It also vanishes out quickly by shaking out human life and things. A mild variation in the earth’s layers which happen kilometres down might even come to us like a huge wave of destruction. Our whole savings for a lifetime could vanish within a blink of time. And it’s not all about that these motions could cause after-effects and passes on to the generation without getting noticed. The fear and pain of the unknown are indeed a big threat to our own existence.
Background history behind the study of earthquakes:
Oklahoma was also facing similar problems by 2015. There had been a rapid increase in the number of earthquakes happening and many were so destructive in nature. So the researchers were thinking about the reasons for these cases. They were also concerned about the ways to measure these quakes beforehand to take necessary precautions. The big waves are easy to detect and to learn. But there are about 5 times more waves left undetected because of their feeble intensity. In order to learn about these, they needed a catalog of these seismic waves and their data.
Previously the members of Harvard university’s department of engineering and earth sciences had used artificial intelligence to solve this problem. They had tried to amplify the seismic signals using a method similar to the voice recognition used in Alexa and Siri. In which the sound waves got from the source are recognized using the inherent data. The same technique is possible for the detection of earth waves, because of the similar nature of both these waves. The seismic waves travel through hard rock, soil, etc. whereas the sound waves travel through the air. For doing so they had made a neuronal network to eliminate the background noises from the surface. These are caused due to environmental issues, constructions, and other ambient noises. And then these filtered waves could be fed as the input to match the previous data of other similar waves.
DEEP LEARNING MODEL TO FIND HIDDEN EARTHQUAKES:
Recently an article was published in Nature Communications under the efforts of S. Mostafa Mousavi, William L. Ellsworth, Weiqiang Zhu, Lindsay Y. Chuang, Gregory C. Beroza. Mousavi and co-authors describe a new method for using artificial intelligence to bring into focus millions of these subtle shifts of the Earth. “By improving our ability to detect and locate these very small earthquakes, we can get a clearer view of how earthquakes interact or spread out along the fault, how they get started, even how they stop,” said Stanford geophysicist Gregory Beroza, one of the paper’s authors.
Mousavi was working daily by analyzing seismographs in Memphis. But this cumbersome process had forced him to automate the earthquake detection. A few years later, after joining Beroza’s lab at Stanford in 2017, he started to think about how to solve this problem using machine learning. He had tried to find alternatives using the algorithms. But his models struggled to eliminate the noise inherent to seismic data. So they had found a series of powerful detectors to measure these feeble signals from the surface.
METHODOLOGY :
PhaseNet was developed by Beroza and graduate student Weiqiang Zhuin 2018 by adapting algorithms from medical image processing to excel at phase-picking, which involves identifying the precise start of two different types of seismic waves. Another machine learning model was released in 2019 and dubbed CRED, was inspired by voice-trigger algorithms in virtual assistant systems, and these proved effective at detection of the seismic signals. The fundamental patterns of earthquake sequences were extracted from relatively few data sets recorded in the seismographs in northern California. According to Mousavi, the model builds on PhaseNet and CRED, and “embeds those insights I got from the time I was doing all of this manually.” Specifically, Earthquake Transformer mimics the way human analysts look at the set of wiggles as a whole and then hone in on a small section of interest.
The attention Mechanism in the outlook of authors:
We often tune out less important details to focus on our priority tasks. Computer scientists call it an “attention mechanism”. They frequently use it to improve text translations. But it’s new to the field of automated earthquake detection. According to Mousavi “I envision that this new generation of detectors and phase-pickers will be the norm for earthquake monitoring within the next year or two,”. “The more information we can get on the deep, three-dimensional fault structure through improved monitoring of small earthquakes, the better we can anticipate earthquakes that lurk in the future,” Beroza said.”Earthquake Transformer gets many more earthquakes than other methods, whether it’s people sitting and trying to analyze things by looking at the waveforms, or older computer methods,” Ellsworth said. “We’re getting a much deeper look at the earthquake process, and we’re doing it more efficiently and accurately.”
The authors also report about the new model that they had developed to detect the small earthquakes and weak signals which the present methods ignore. It is called the Earthquake Transformer which was used to pick out the precise timing of the seismic phases using earthquake data from around the world.
They are reaching toward their goal with the hope to predict the waves even before they leave their point of occurrence. It’s not just exciting, but it would be a great boon that would benefit the human race. Now the aim is to predict the earthquakes and the team is after it.
Journal reference: Earthquake transformer—an attentive deep-learning model for simultaneous earthquake detection and phase picking S. Mostafa Mousavi, William L. Ellsworth,
Weiqiang Zhu, Lindsay Y. Chuang &Gregory C. Beroza.
https://www.nature.com/articles/s41467-020-17591-w
PhaseNet: a deep-neural-network-based seismic arrival-time picking method
Weiqiang Zhu, Gregory C Beroza.
https://academic.oup.com
Author
