[1] Intermediate-depth earthquakes occur at depths where temperatures and pressures exceed those at which brittle failure is expected. There are two leading candidates for the physical mechanism behind these earthquakes: dehydration embrittlement and self-localizing thermal shear runaway. A complete energy budget for a range of earthquake sizes can help constrain whether either of these mechanisms might play a role in intermediate-depth earthquake rupture. The combination of high stress drop and low radiation efficiency that we observe for M w 4-5 earthquakes in the Bucaramanga Nest implies a temperature increase of 600-1000°C for a centimeter-scale layer during earthquake failure. This suggests that substantial shear heating, and possibly partial melting, occurs during intermediate-depth earthquake failure. Our observations support thermal shear runaway as the mechanism for intermediate-depth earthquakes, which would help explain differences in their behavior compared to shallow earthquakes. Citation: Prieto, G. A., M. Florez, S. A. Barrett, and G. C. Beroza (2013), Seismic evidence for thermal runaway during intermediate-depth earthquake rupture, Geophys. Res. Lett., 40,[6064][6065][6066][6067][6068]

No abstract

Double seismic zones are ubiquitous features of subduction zones, where seismicity is distributed along two layers separated by a region with significantly less seismic activity. Dehydration embrittlement is thought to be responsible for earthquakes in the subducting crust (upper layer), but the case for it in the lithospheric mantle (lower layer) is less clear. We apply a recently developed relative relocation technique to characterize seismicity in 32 slab segments. The high‐precision hypocentral depths allow us to assign events to either the upper or lower layer and to separately estimate frequency size distributions for each plane. We find consistently larger b values, correlating with slab age, for the upper layer and roughly constant values for the lower. We also show that thermal parameter and plate age are the key controls on double seismic zone geometry. Our results point to a relatively dry lower layer and suggest a fundamentally different mechanism for lithospheric mantle earthquakes.

Precise determination of hypocentral depth remains one of the most relevant problems in earthquake seismology. It is well known that using depth phases allows for significant improvement in event depth determination; however, routinely and systematically picking such phases, for teleseismic or regional arrivals, is problematic due to poor signal‐to‐noise ratios around the pP and sP phases. To overcome this limitation, we have taken advantage of the additional information carried by seismic arrays. We use velocity spectral analysis to precisely measure pP‐P times. The individual estimates obtained at different subarrays, for all pairs of earthquakes, are combined using a double‐difference algorithm, in order to precisely map seismicity in regions where it is tightly clustered. We illustrate this method by relocating intermediate‐depth earthquakes in the Nazca subducting plate, beneath northern Chile, where we confirm the existence of a narrowly spaced double seismic zone, previously imaged using a local dedicated deployment. As a second example we relocate the aftershock sequence of the 2014 Mw 7.9 intermediate depth, Rat Islands earthquake, and provide evidence of a subvertical fault plane for the main shock. Finally, we show that the resulting relative depth errors are typically smaller than 2 km.

Exploration seismic data are heavily manipulated before human interpreters are able to extract meaningful information regarding subsurface structures. This manipulation adds modeling and human biases and is limited by methodological shortcomings. Alternatively, using seismic data directly is becoming possible thanks to deep learning (DL) techniques. A DL-based workflow is introduced that uses analog velocity models and realistic raw seismic waveforms as input and produces subsurface velocity models as output. When insufficient data are used for training, DL algorithms tend to overfit or fail. Gathering large amounts of labeled and standardized seismic data sets is not straightforward. This shortage of quality data is addressed by building a generative adversarial network (GAN) to augment the original training data set, which is then used by DL-driven seismic tomography as input. The DL tomographic operator predicts velocity models with high statistical and structural accuracy after being trained with GAN-generated velocity models. Beyond the field of exploration geophysics, the use of machine learning in earth science is challenged by the lack of labeled data or properly interpreted ground truth, since we seldom know what truly exists beneath the earth's surface. The unsupervised approach (using GANs to generate labeled data)illustrates a way to mitigate this problem and opens geology, geophysics, and planetary sciences to more DL applications.