Machine Learning System Reveals Thousands of Microquakes in Alaska
June 8, 5:30 PM – A machine learning system has identified thousands of previously undetected microearthquakes in Alaska, refining the boundary of the Yakutat microplate as it subducts beneath the North American plate. The research was led by Megan Miller from the Australian National University.
Using data from permanent seismic stations and a temporary network active from 2018 to 2021, the research team built an expanded earthquake catalog with machine learning algorithms. This effort uncovered a chain of roughly 1,750 small earthquakes stretching nearly 250 kilometers from northwest to southeast. This seismic line delineates the edge of the Yakutat microplate, situated beneath the most sharply curved section of the Alaska Range.
Study Findings
The Yakutat microplate is linked to the Denali fault system, which produced a magnitude 7.9 earthquake in 2002. The newly identified plate boundary aligns with previous studies that analyzed tectonic tremors.
'The linear structure we found matches the location where tectonic tremors stop.' – Megan Miller
She further emphasized that 'such a structure had not been observed before, but it fits perfectly with other independent seismic data.' The study revealed that rocks west of the plate boundary behave more plastically, while those at the boundary are more brittle.
The scientists plan to analyze data from before 2018 to further clarify geological processes in the region.
This discovery is crucial for understanding tectonic activity in Alaska, where earthquake risks can impact public safety and infrastructure. The use of advanced technologies like machine learning enables researchers to obtain more detailed information about seismic events, potentially aiding future earthquake forecasting. Expanding knowledge of the Yakutat microplate will also deepen insights into regional geological processes and their environmental effects.