Earthquakes and faulting are intricate processes that require multidiscipline approaches to fully characterize fault interaction, stress perturbations, and the associated surface deformation. My research is at the intersection of seismology and geodesy and focuses on the Earth’s response to transient deformation. I explore these problems using seismic (e.g. waveforms, focal mechanisms, seismicity), geodetic (InSAR and GPS), and modeling tools to estimate stress changes and explore variations in seismicity rates.
Seasonal stress modulation on active California fault structures
In California the accumulated winter snow pack in the Sierra Nevada, reservoirs and groundwater water storage in the Central Valley follow an annual periodic cycle and each contribute to the resulting surface deformation, which can be observed using GPS time series. The ongoing drought conditions in the western U.S. amplify the observed uplift signal as the Earth’s crust responds to the mass changes associated with the water loss. The near surface hydrological mass loss can result in annual stress changes of ~1kPa at seismogenic depths. Similarly, small static stress perturbations have previously been associated with changes in earthquake activity. Periodicity analysis of earthquake catalog time series suggest that periods of 4-, 6-, 12-, and 14.24-months are statistically significant in regions of California, and provide documentation for the modulation of earthquake populations at periods of natural loading cycles. Knowledge of what governs the timing of earthquakes is essential to understanding the nature of the earthquake cycle. If small static stress changes influence the timing of earthquakes, then one could expect that events will occur more rapidly during periods of greater external load increases. To test this hypothesis we develop a loading model using GPS derived surface water storage for California and calculate the stress change at seismogenic depths for different faulting geometries. We utilize the UCERF3 fault geometry for all faults in California and develop a stress time series for each. We then evaluate the degree of correlation between the stress models and the seismicity taking into consideration the variable amplitude of stress cycles, the orientation of transient load stress with respect to the background stress field, and the geometry of active faults revealed by focal mechanisms. Our results indicate more earthquakes are occurring on reverse faults during periods of peak unloading. The reverse structure orientation is favorably oriented for a maximum Coulomb stress change ~1k Pa. A less pronounced signal is observed for strike-slip faults.
Delayed dynamic triggering: Local seismicity leading up to three remote M≥6 aftershocks of the 11 April 2012 M8.6 Indian Ocean earthquake
The 11 April 2012 M8.6 strike-slip Indian Ocean earthquake (IOE) was followed by an increase in global seismic activity, with three remote M≥6.0 earthquakes within 24 hours. We investigate delayed dynamic triggering by systematically examining three offshore regions hosting these events for changes in microseismic activity preceding the IOE, and during the hours between the IOE surface-wave arrival and the triggered-event candidate. The Blanco Fault Zone, USA and the Tiburón Fault Zone, Mexico each host a strike-slip event and the Michoacán Subduction Zone, Mexico hosts a reverse event. At these locations we estimate transient Coulomb stresses of ±1-10 kPa during the IOE. Each study area contains a regional seismic network allowing us to examine continuous waveforms at one or more nearby stations. We implement a short- /long-term-average algorithm and template matching to detect events and assess the seismicity with the β-statistic. Our results indicate low-magnitude seismicity in the days prior to the IOE and the occurrence of earthquakes during the surface-wave passage after more than 2-hours of transient loading. We find both transtensional tectonic environments respond to the transient stresses with a substantial increase observed in the seismicity rates during the hours after the surface waves passage. In contrast, seismicity rates remain constant in the subduction zone we investigate during the 14-hour delay between the IOE and the large-magnitude earthquake. The seismicity rate increases we observe occur after many hours of dynamic stresses and suggest the long duration of transient loading initiated failure processes leading up to these M≥6.0 events.
Global catalog analysis shows that dynamic triggering of remote M≥5.5 earthquakes is rare
Probing the effects of a transient stress on the timing of an earthquake occurrence is necessary for understanding the remote interaction of large-magnitude events. Global catalog data containing 35 years of M≥5.5 earthquakes allows us to explore for periods of enhanced or suppressed seismic activity. We consider 113 M≥7.5 mainshocks between 1977-2012 and focus on seismic activity on time scales from seconds to days following these mainshocks. We search for evidence of dynamic triggering of large-magnitude events similar to the previously observed global increase during the first few days following the 2012 M8.6 Indian Ocean mainshock. We restrict the analysis to regions of elevated strain during the passage of surface waves. Using a threshold of 0.1 µstrain (~3 kPa) and a temporal window of ±1-year, we stack daily seismicity rate curves using the exclusion-zone declustered M≥5.5 pre- and postshocks in order to resolve deviations from the background rate. Our results do not indicate a significant change in activity for at least 10-days when considering the collective set of 113 mainshocks and subsets at M8.0 and M8.5 thresholds. The results also do not indicate immediate triggering of M≥5.5 events. We do find two instances of increased seismicity in the elevated strain region within 10-days. These increases are subsequent to two mainshocks, the 1977 M8.3 and 2012 M8.6, both located in the Indian Ocean. We conclude that a global change in M≥5.5 earthquake rates following a transient stress from distant earthquakes is a rare occurrence.
Depth Migration of Seasonally Induced Seismicity at The Geysers Geothermal Field
Seismicity from injected fluids provides insight into the hydraulically conductive fracture network at The Geysers, California (TG) geothermal reservoir. Induced earthquakes at TG result from both thermo- and poroelastic stresses as injected fluids cool the rocks and increase pore pressure. The spatio-temporal evolution of M≥1.5 seismicity is characterized as a function of depth in the northwest and southeast regions of TG to develop time-dependent earthquake rates using an Epidemic Type Aftershock Sequence model. The seismicity and injection follow an annual cycle that peaks in the winter months and is correlated by depth. The results indicate a time lag of ≤6 months for fluids to migrate >3 km below the injection depth. Water injection is the main cause of seismicity as fluids penetrate into the reservoir. Our results suggest a steeply dipping fracture network of hydraulically conductive faults allows fluid migration to a few kilometers below the point of injection.
Interseismic coupling and refined earthquake potential on the Hayward-Calaveras fault zone
Interseismic strain accumulation and fault creep is usually estimated from GPS and alignment arrays data, which provide precise but spatially sparse measurements. Here, we use InSAR to resolve the interseismic deformation associated with the Hayward and Calaveras Faults (HF and CF) in the East San Francisco Bay Area. The large 1992-2011 SAR dataset permits evaluation of short and long-wavelength deformation larger than 2 mm/yr without alignment of the velocity field to a GPS-based model. Our time series approach in which the interferogram selection is based on the spatial coherence enables deformation mapping in vegetated areas and leads to refined estimates of along-fault surface creep rates. Creep rates vary from 0±2 mm/yr on the northern CF, to 14±2 mm/yr on the central CF south of the HF surface junction. We estimate the long-term slip rates by inverting the long-wavelength deformation and the distribution of shallow slip due to creep by inverting the remaining velocity field. This distribution of slip reveals the locations of locked and slowly creeping patches with potential for a M6.8±0.3 on the HF near San Leandro, a M6.6±0.2 on the northern CF near Dublin, a M6.5±0.1 on the HF south of Fremont, and a M6.2±0.2 on the central CF near Morgan Hill. With cascading multi-segment ruptures the HF rupturing from Berkeley to the CF junction could produce a M6.9±0.1, the northern CF a M6.6±0.1, the central CF a M6.9±0.2 from the junction to Gilroy, and a joint rupture of the HF and central CF could produce a M7.1±0.1.
Potential and limits of InSAR to characterize interseismic deformation independently of GPS data: application to the southern San Andreas Fault system
The evaluation of long-wavelength deformation associated with interseismic strain accumulation traditionally relies on spatially sparse GPS measurements, or on high spatial-resolution InSAR velocity fields aligned to a GPS-based model. In this approach the InSAR contributes only short-wavelength deformation and the two datasets are dependent, thereby preventing the evaluation of the InSAR uncertainties and the justification of atmospheric noise corrections. Here, we present an analysis using seven years of Envisat InSAR data for the characterization of interseismic deformation along the southern San Andreas Fault (SAF) and the San Jacinto Fault (SJF) in southern California, where the SAF bifurcates onto the Mission Creek (MCF) and the Banning (BF) fault strands. We outline the processing steps for using InSAR alone to characterize both the short- and long-wavelength deformation, and evaluate the velocity field uncertainties with independent continuous GPS data. InSAR line-of-sight (LOS) and continuous GPS velocities agree within ~1 mm/yr in the study area, suggesting that InSAR time series with a sufficient amount of data can be used to fully characterize interseismic deformation with a higher spatial resolution than GPS. We investigate with dislocation models the ability of this mean LOS velocity field to constrain fault slip rates and show that it can help distinguish between different slip-rate scenarios on the SAF and SJF (~35 km apart) but not on the MCF and BF (~12 km apart). Our results demonstrate that interseismic models of strain accumulation used for seismic hazards assessment would benefit from the consideration of InSAR mean velocity maps.
Interseismic coupling of major faults in the North San Francisco Bay Area from InSAR, GPS and seismic data
The San Andreas Fault System in the north San Francisco Bay area consists of four major sub-parallel strands, the northern San Andreas, Rodgers Creek - Maacama, West Napa, and Concord-Green Valley - Bartlett Springs faults. Several moderate to large earthquakes have occurred on these faults during the past century. In 1906, a Mw 7.8 earthquake ruputered on the NSAF causing significant damage to the city of San Francisco. In 1969, two damaging moderate magnitude (M5.6 and M5.7) earthquakes occurred on the Rogers Creek fault north of Santa Rosa. More recently, a M6 earthquake ruptured the west Napa fault on August 24, 2014, causing significant economic damage in the region. No major earthquakes have occurred during the historical record on the Maacama fault, the Concord Green Valley fault or the Bartlett Springs fault, despite the accumulation of a slip deficit that is capable of producing a M7 earthquake. Compared to the well-studied faults in the central Bay area, the spatial variations of strain accumulation and creep on major faults in the North Bay remain poorly understood. Here we rely on space geodetic and seismic data to determine the distribution of aseismic slip on North Bay faults and provide important information regarding their seismic potential.
Insights into distributed plate rates across the Walker Lane from GPS geodesy
Contemporary geodetic slip rates are observed to be approximately two times greater than late Pleistocene geologic slip rates across the southern Walker Lane. Using a dense GPS network, we compare the present-day crustal velocities to observed geologic slip rates in the region. We find that the Walker Lane is characterized by a smooth transition from westward extension in the Basin and Range to northwestward motion of the Sierra Nevada block. The GPS velocity field indicates that (1) plate parallel (N37°W) velocities define a velocity differential of 10.6 ± 0.5 mm/yr between the western Basin and Range and the Sierra Nevada block, (2) there is ~2 mm/yr of contemporary extension perpendicular to the normal faults of the Silver Peak-Lone Mountain extensional complex, and (3) most of the observed discrepancy in long- and short-term slip rates occurs across Owens Valley. We believe the discrepancy is due to distributed strain and underestimated geologic slip rates.