| Abstract |
Evidence of increasing microseismicity during geothermal power plant outages has been observed worldwide. Seismic studies in these areas provide several hypotheses mostly based on fault slip induction by changes in pore pressure. However, geological dissimilarities between regions halt assumptions of a unique stress mechanism for this phenomenon. Past methods like microearthquake location mapping have proved useful in delineating subsurface structure in these systems and understanding pathways for injection flow. Nevertheless, the stress changes triggering these events remain unclear in some cases. We apply a data mining technique, called PageRank, to assess microseismic event connectivity and evolution at the Blue Mountain geothermal site during 2017, a crucial first step to understand the origin of increased seismicity on-site during the annual power plant shutdown (September 18th) and other power plant operations. Here we compute direct (CC greater than = 0.6) and indirect (CC less than 0.6, linkage stations greater than = 6) links to a reference event, a high PageRank event leading a seismic cluster. Cluster characterization results in several families comprised of unique microseismic events with similar waveform topology. In addition, traces from the indirect links within our identified families show a similar waveform pattern, confirming their belonging to the cluster. This study demonstrates the usefulness of PageRank to characterize microseismic events with similar physical properties, but also the advantages for a more robust cluster relocation analysis by affiliating low correlation events to the families, usually rejected by other methods. |