A document by Nader Shakibay Senobari. Click on the document to view its contents.

We search for repeating earthquakes (REs) in the northern San Francisco Bay Area in 1984-2016. By comparing over 670,000 waveforms from ~75,000 events, we identify candidate clusters of events whose waveforms have high cross-correlation coefficients at multiple stations. A key difference with our approach is that these ‘multi-station clusters’ do not require each event in a family be recorded at multiple common stations. We validate these candidate REs by estimating precise relative relocations for the events in each cluster. We identify 59 RE families whose relocated hypocenters are separated by less than one source radius. These are distributed throughout the Maacama fault zone, and along the northern Rodgers Creek and central Bartlett Springs faults, implying that widespread, pervasive creep occurs on those faults, at rates of 1-6 mm/yr. At either end of the Maacama fault, the RE pattern highlights structural complexity, suggesting that multiple subparallel strands may be active and creeping.

The 2019 Ridgecrest earthquakes pose interesting questions about the nature of intersecting conjugate ruptures, and also the possibility of re-rupture of fault segments. Aftershocks of the July 4th M6.4 event suggest the possibility of a secondary rupture along the fault that subsequently ruptured in the July 5th M7.1 event. Unfortunately, neither InSAR nor rupture mapping will be able to resolve this question, as no SAR acquisitions were made between the two earthquakes, and the critical ‘nexus’ of the two ruptures was located on the China Lake Navy base, and was not accessible between the events. Campaign GPS data and seismic data may provide clues to resolving these questions. We reoccupied 5 previously surveyed GPS benchmarks in the hours following the M6.4 event, meaning that we can separately measure the deformation from the two earthquakes at those locations. We construct a joint inversion of our campaign GPS data, along with the daily displacements of nearby continuous GPS stations, and ascending and descending Sentinel-1 InSAR data, in order to separate the fault geometry and slip of the two earthquakes, and address the question of re-rupture. This approach also allows us to precisely estimate the contribution of the early postseismic deformation to the InSAR data.

Partially creeping faults exhibit complex behavior in terms of which parts of the fault slip seismically versus aseismically. The specific geometry of creeping versus locked fault patches may pose constraints on rupture lengths on partially-creeping faults. We use the 3D finite element method to conduct dynamic rupture simulations on simplified partially-creeping strike-slip faults, to determine whether coseismic rupture can propagate into creeping regions, and how the presence and distribution of creeping regions affects the ability of rupture to propagate across the whole fault. We implement rate-state friction, in which locked zones are represented by rate-weakening behavior and creeping zones are assigned rate-strengthening properties. We model two simplified geometries: a locked patch at the base of a creeping fault and a creeping patch at the surface of a locked fault. In the case of a locked patch within a creeping fault, rupture does not propagate far past the edges of the locked patch, regardless of its radius. The case of a creeping patch within a locked fault is more complicated. The width of the locked areas around the creeping patch determine whether rupture is able to propagate around the creeping patch. Although rupture is always able to propagate at least a small distance into the creeping patch, if the width of the locked zone between the edge of the creeping patch and the end of the fault is too narrow, rupture stops. This simplified parameter study may be useful for understanding first-order behaviors of real-world partially-creeping strike-slip faults.

Interferometric synthetic aperture radar (InSAR) measures surface deformation from repeated passes of satellites or aircraft and has become an important tool to study geophysical phenomena such as earthquakes. However, InSAR data analysis is challenging due to atmospheric water vapor that can mimic the effects of Earth deformation and thus lead to wrong interpretations. We present preliminary results on how to differentiate between tropospheric effects and surface deformation from earthquakes using a convolutional neural network. As earthquake training sets are sparse, our approach leverages transfer learning techniques for tropospheric patterns from areas where deformations are known to be mostly absent over short time periods, and classifies specific areas of interferograms to reflect regions that are dominated by deformation or tropospheric noise. The applicability of the training set to a new area may depend on the similarity of the two climates. Examples of tropospheric delays are shown from interferograms constructed from Sentinel-1 data over part of southern California with short temporal baselines, and examples of deformation are taken from interferograms generated using Okada models. Our classifier is tested on data from the 2018 Oaxaca earthquake in Mexico from Sentinel-1. This work is a step towards using neural networks for a fine-granular tile-based validation of interferograms and automatically removing unwanted effects from InSAR signals, as well as towards enhancing the agility of disaster response programs. The open source code is available in the PyInSAR package on GitHub under the MIT license. We acknowledge support from NASA AIST80NSSC17K0125 (PI Pankratius) and NSF ACI1442997 (PI Pankratius).

The shallow 1971 Mw 6.6 San Fernando, California earthquake involved a complex rupture process on an immature thrust fault with a non-planar geometry, and is notable for having a higher component of left-lateral surface slip than expected from seismic models. We extract its 3-D coseismic surface displacement field from aerial stereo photographs and document the amount and width of the vertical and strike-parallel components of distributed deformation along strike. The results confirm the significant left-lateral surface offsets, suggesting a slip vector rotation at shallow depths. Comparing our offsets against field measurements of fault slip, we observe that most of the offset was accommodated in the damage zone, with off-fault deformation averaging 68% in both the strike-parallel and vertical components. However, the magnitude and width of off-fault deformation behave differently between the vertical and strike-parallel components, which, along with the rotation in rake near the surface, can be explained by dynamic rupture effects.

Recent geological observations imply that slow slip events (SSEs) occur in fault zones with a finite thickness of ~100s of meters. The bulk matrix of the fault zone deforms viscously, while pervasive frictional surfaces are distributed in the viscous matrix. In this theoretical study, we investigate the rupture behaviors of a frictional-viscous-mixing fault model and explore its potential to generate a single SSE. To simultaneously consider both the 10s-kilometer-scale rupture propagations and the 100s-meter-scale “frictional-viscous” features in the same model, we treat a fault zone as a zero-thickness “surface” embedded in an elastic medium. The “frictional-viscous” characteristics are parameterized into a constitutive relation, or “friction law”, where fault strength is partitioned into a frictional and a viscous component. For simplicity, the frictional strength is set to be slip weakening, while the viscous strength increases linearly as the bulk shear rate (slip rate) increases. We explore the rupture behaviors of the above model both analytically and numerically. We find that: 1. Final slip is proportional to the static stress drop and slipping area length, as in fast earthquakes. Peak slip rate increases with dynamic frictional stress drop, while a high viscous coefficient can significantly reduce slip rate, leading to slow slip behaviors. 2. Rupture propagation speed is mainly controlled by the radiation damping factor and viscous coefficient and can be significantly reduced compared to typical fast earthquakes when the viscous coefficient is high. 3. The slip rate decay time increases with the viscous coefficient and slipping area length, which eventually predicts M~T^3 scaling. When frictional stress drop is ~1MPa and the viscous coefficient is smaller than the radiation damping factor μ/(2β), the above models predict the fast slip behavior of regular fast earthquakes. Our model predicts slow slip behaviors in a wide range of parameter space when stress drop is low and viscous coefficient is high. In particular, a frictional strength drop of ~10 kPa and viscous coefficient of 10^4-10^5 μ/(2β) can simultaneously explain many independent characteristic rupture parameters of SSEs. Our model can be further tested with future geophysical, geological, and experimental data.