The recent evolution of induced seismicity in Central United States calls for exhaustive catalogs to improve seismic hazard assessment. Over the last decades, the volume of seismic data has increased exponentially, creating a need for efficient algorithms to reliably detect and locate earthquakes. Today's most elaborate methods scan through the plethora of continuous seismic records, searching for repeating seismic signals. In this work, we leverage the recent advances in artificial intelligence and present ConvNetQuake, a highly scalable convolutional neural network for earthquake detection and location from a single waveform. We apply our technique to study the induced seismicity in Oklahoma (USA). We detect 20 times more earthquakes than previously cataloged by the Oklahoma Geological Survey. Our algorithm is orders of magnitude faster than established methods.

Over the last decades, the volume of seismic data has increased exponentially, creating a need for efficient algorithms to reliably detect and locate earthquakes. Today's most elaborate methods scan through the plethora of continuous seismic records, searching for repeating seismic signals. In this work, we leverage the recent advances in artificial intelligence and present ConvNetQuake, a highly scalable convolutional neural network for probabilistic earthquake detection and location from single stations. We apply our technique to study two years of induced seismicity in Oklahoma (USA). We detect 20 times more earthquakes than previously cataloged by the Oklahoma Geological Survey. Our algorithm detection performances are at least one order of magnitude faster than other established methods.

The mass loss from the West Antarctic Ice Sheet is dominated by numerous rapidly flowing ice streams, which are separated from stagnant ice in the adjacent ridges by zones of concentrated deformation known as shear margins. Because the discharge from a single ice stream depends sensitively on the ice stream width, determining the physical processes that control shear margin location is crucial to a full understanding of ice stream dynamics. Previous work has shown that the transition from a deforming to an undeforming bed within a shear margin concentrates large stresses on the undeforming bed beneath the ridge. In this paper we investigate how the presence of a drainage channel collocated with the transition from a deforming to an undeforming bed perturbs the stress field within the shear margin. We show that the channel limits the maximum shear stress on the undeforming bed and alters the yield strength of the till by changing the normal stress on the ice-till interface. By comparing the maximum stress with the till strength, we show that the transition from a deforming to an undeforming bed can occur across a channel whenever the water flux in the channel exceeds a critical value. This critical flux is sensitive to the rheology and loading of the shear margin, but we conclude that there are some scenarios where the transition from a deforming to an undeforming bed can be collocated with a drainage channel, though this configuration is probably not typical.

We develop a micromechanics-based permeability evolution model for brittle materials at high strain rates (\(\ge 100\) s\(^{-1}\)). Extending for undrained deformation the mechanical constitutive description of brittle solids, whose constitutive response is governed by micro-cracks, we now relate the damage-induced strains to micro-crack aperture. We then use an existing permeability model to evaluate the permeability evolution. This model predicts both the percolative and connected regime of permeability evolution of Westerly Granite during triaxial loading at high strain rate. This model can simulate pore pressure history during earthquake coseismic dynamic ruptures under undrained conditions.

Fast-flowing ice streams in West Antarctica are separated from the nearly stagnant ice in the adjacent ridge by zones of highly localized deformation known as shear margins. It is presently uncertain what mechanisms control the location of shear margins and possibly allow them to migrate. In this paper we show how subglacial hydrological processes can select the shear margin location, leading to a smooth transition from a slipping to a locked bed at the base of an ice stream. Our study uses a two-dimensional thermomechanical model in a cross section perpendicular to the direction of flow. We confirm that the intense straining at the shear margins can generate large temperate regions within the deforming ice. Assuming that the melt generated in the temperate ice collects in a drainage channel at the base of the margin, we show that a channel locally decreases the pore pressure in the subglacial till. Therefore, the basal shear strength just outside the channel, assuming a Coulomb-plastic rheology, can be substantially higher than that inferred under the majority of the stream. Results show that the additional basal resistance produced by the channel lowers the stress concentrated on the locked portion of the bed. Matching the model to surface velocity data, we find that shear margins are stable when the slipping-to-locked bed transition occurs less than 500m away from a channel operating at an effective pressure of 200kPa and for a hydraulic transmissivity equivalent to a basal water film of order 0.2mm thickness.

Ice streams are fast flowing bands of ice separated from stagnant ridges by shear margins. The mechanisms controlling the location of the margins remain unclear. We use published ice deformation data and a simple one‐dimensional thermal model to show that West Antarctic ice stream margins have temperate ice over a substantial fraction of their thickness, a condition that may control their width. The model predicts a triple‐valued relation between the thickness‐averaged lateral shear stress and the lateral shear strain rate. Observed strain rates at the margins imply that they support slightly less lateral shear stress than adjacent ice within the stream. This requires enhanced basal resistance near the margin. We suggest, in agreement with the limited observations, the presence of a channelized drainage system at the margin that reduces the pore fluid pressure at the ice‐till interface, thus increasing the shear stress acting on the yielding Coulomb plastic bed.

Flow of glacial ice in the West Antarctic Ice Sheet localizes in narrow bands of fast-flowing ice streams bordered by ridges of nearly stagnant ice, but our understanding of the physical processes that generate this morphology is incomplete. Here we study the thermal and mechanical properties of ice-stream margins, where flow transitions from rapid to stagnant over a few kilometers. Our goal is to explore under which conditions the intense shear deformation in the margin may lead to deformation-induced melting. We propose a 2-D model that represents a cross section through the ice stream margin perpendicular to the downstream flow direction. We limit temperature to the melting point to estimate melt rates based on latent heat. Using rheology parameters as constrained by laboratory data and observations, we conclude that a zone of temperate ice is likely to form in active shear margins.

We develop a micromechanics-based permeability evolution model for brittle materials at high strain rates ( $$\ge 100$$ s $$^{-1}$$ ). Extending for undrained deformation the mechanical constitutive description of brittle solids, whose constitutive response is governed by micro-cracks, we now relate the damage-induced strains to micro-crack aperture. We then use an existing permeability model to evaluate the permeability evolution. This model predicts both the percolative and connected regime of permeability evolution of Westerly Granite during triaxial loading at high strain rate. This model can simulate pore pressure history during earthquake coseismic dynamic ruptures under undrained conditions.