Exciting news! We're transitioning to the Statewide California Earthquake Center. Our new website is under construction, but we'll continue using this website for SCEC business in the meantime. We're also archiving the Southern Center site to preserve its rich history. A new and improved platform is coming soon!

Poster #207, Seismology

New insights into the Rangely earthquake control experiment

Kaiwen Wang, Gregory C. Beroza, & William L. Ellsworth
Poster Image: 

Poster Presentation

2021 SCEC Annual Meeting, Poster #207, SCEC Contribution #11426 VIEW PDF
The Rangely experiment was the first experiment to control the occurrence of earthquakes deliberately by varying fluid injection. It was conducted in an anticlinal oil field at Rangely, Colorado in 1970s. Due to its controlled nature and to the close correspondence between local earthquake activity and reservoir pressure, it provides a unique opportunity for studying injection induced earthquake mechanisms. Although our understanding of the mechanisms responsible for inducing earthquakes has advanced significantly since the time of the Rangely experiment in the 1970s, few attempts have been made to reanalyze the Rangely earthquakes (Byrne et al., 2020, Silva et al., 2021). This is largely ... attributed to the loss of original earthquake catalog. Recovery of the catalog using standard methods would be both time consuming and challenging, as all that survives are the original Develocorder microfilms. We developed an efficient machine-learning-based system for analysis of the films that efficiently processes analog seismograms at scale. We build on our reconstructed catalog to conduct a geomechanical analysis and investigate the physical mechanism inducing the Rangely earthquakes. Our results support the effective stress hypothesis as originally proposed in the Raleigh et al. (1976) study. We found that events are more continuously distributed between inside and outside the oil field than previously reported, which supports a direct fluid connection across the bounding thrust fault. The spatiotemporal distribution of seismicity and its correspondence with reservoir pressure also supports the conclusion that the earthquakes were induced by pore pressure increase rather than poroelastic effects. Additionally, we performed earthquake clustering and found that the Rangely earthquakes are characterized by short-lived sub-sequences. We inferred from the sub-sequence propagation that they are likely driven by fast stress transfer processes such as earthquake interaction, fluid diffusion or aseismic slip. These results offer new insights on the physical mechanisms of the Rangely induced earthquakes. With the Rangely example, we demonstrated that we can efficiently perform seismic analysis on analog data. This demonstrates the potential of the approach to study other important pre-digital earthquake sequences.