Automated Microseismic Event Detection with Machine Learning

The research aims to speed up the migration based passive seismic source location algorithm such as passive seismic emission tomography and semblance weighted stacking using massive parallel computers. This will be achieved through both single-GPU and multi-GPU implementations of parallel algorithms on the high-performance supercomputer Savanna. Both synthetic and field passive seismic datasets will be processed and varieties of forward models with different geophone arrays will be tested for performance comparison. This research will facilitate the on-site real-time monitoring of hydraulic fracturing, which entails massive data volume and high-speed data processing especially in surface based microseismic monitoring.

Funding: Research Training Program (RTP)
Advisors: Associate Professor Lutz Gross