Abstract: Uniaxial compression tests with combined acousto-optical monitoring techniques are conducted on flawed granite specimens, with the aim of investigating the fracture-related acoustic emission (AE) event rate characteristics at the unstable cracking phase in flawed rocks. The interevent time (IET) function F(τ) is adopted to interpret the AE time-series from damage stress (σcd) to ultimate failure, and photographic data are used to evaluate unstable cracking behaviours in flawed granite. The results show that a high AE event rate is always registered but intermittently interrupted by macrofracturing at the unstable cracking phase. The reversed U-shaped curve relation between the AE event rate and the loading time is documented in unstable flawed granite for the first time. The acoustic quiescence has a mechanismic linkage and quantitative correlation with stress drop, and this synchronous acousto-mechanical behaviour is a typical result of the initiation, growth and coalescence of macrocracks initiated from the flaw tips. Moreover, the reactivation and intensification of fracture process zones (FPZs) by increasing loads are the dominant mechanism triggering unstable crack growth in flawed granite.

Abstract: N.L. Celli ,1,2 S. Lebedev,1 A.J. Schaeffer,3 M. Ravenna1,4 and C. Gaina5 1Dublin Institute for Advanced Studies, Geophysics Section, School of Cosmic Physics, Dublin, Ireland. E-mail: nlscelli@gmail.com, 2Trinity College Dublin, Geology Department, Dublin, Ireland 3Geological Survey of Canada, Pacific Division, Sidney Subdivision, Natural Resources Canada, Sidney, BC, Canada 4P-LAB Srl, Verona, Italy. 5Centre for Earth Evolution and Dynamics (CEED), University of Oslo, Oslo, Norway

Abstract: Conventional geophysical inversion techniques suffer from several limitations including computational cost, nonlinearity, non-uniqueness and dimensionality of the inverse problem. Successful inversion of geophysical data has been a major challenge for decades. Here, a novel approach based on deep learning (DL) inversion via convolutional neural network (CNN) is proposed to instantaneously estimate subsurface electrical conductivity (σ) layering from electromagnetic induction (EMI) data. In this respect, a fully convolutional network was trained on a large synthetic data set generated based on 1-D EMI forward model. The accuracy of the proposed approach was examined using several synthetic scenarios. Moreover, the trained network was used to find subsurface electromagnetic conductivity images (EMCIs) from EMI data measured along two transects from Chicken Creek catchment (Brandenburg, Germany). Dipole–dipole electrical resistivity tomography data were measured as well to obtain reference subsurface σ distributions down to a 6 m depth. The inversely estimated models were juxtaposed and compared with their counterparts obtained from a spatially constrained deterministic algorithm as a standard code. Theoretical simulations demonstrated a well performance of the algorithm even in the presence of noise in data. Moreover, application of the DL inversion for subsurface imaging from Chicken Creek catchment manifested the accuracy and robustness of the proposed approach for EMI inversion. This approach returns subsurface σ distribution directly from EMI data in a single step without any iterations. The proposed strategy simplifies considerably EMI inversion and allows for rapid and accurate estimation of subsurface EMCI from multiconfiguration EMI data.

Abstract: This study describes a deep convolutional neural network (CNN) based technique for the prediction of intensity measurements (IMs) of ground shaking. The input data to the CNN model consists of multistation 3C broadband and accelerometric waveforms recorded during the 2016 Central Italy earthquake sequence for M $\ge$ 3.0. We find that the CNN is capable of predicting accurately the IMs at stations far from the epicenter and that have not yet recorded the maximum ground shaking when using a 10 s window starting at the earthquake origin time. The CNN IM predictions do not require previous knowledge of the earthquake source (location and magnitude). Comparison between the CNN model predictions and the predictions obtained with Bindi et al. (2011) GMPE (which require location and magnitude) has shown that the CNN model features similar error variance but smaller bias. Although the technique is not strictly designed for earthquake early warning, we found that it can provide useful estimates of ground motions within 15-20 sec after earthquake origin time depending on various setup elements (e.g., times for data transmission, computation, latencies). The technique has been tested on raw data without any initial data pre-selection in order to closely replicate real-time data streaming. When noise examples were included with the earthquake data, the CNN was found to be stable predicting accurately the ground shaking intensity corresponding to the noise amplitude.

Abstract: We have constructed a 3-D shear-wave velocity (Vs) model for the crust and uppermost mantle beneath the Middle East using Rayleigh wave records obtained from ambient-noise cross-correlations and regional earthquakes. We combined one decade of data collected from 852 permanent and temporary broadband stations in the region to calculate group-velocity dispersion curves. A compilation of >54000 ray paths provides reliable group-velocity measurements for periods between 2 and 150 s. Path-averaged group velocities calculated at different periods were inverted for 2-D group-velocity maps. To overcome the problem of heterogeneous ray coverage, we used an adaptive grid parametrization for the group-velocity tomographic inversion. We then sample the period-dependent group-velocity field at each cell of a predefined grid to generate 1-D group-velocity dispersion curves, which are subsequently inverted for 1-D Vs models beneath each cell and combined to approximate the 3-D Vs structure of the area. The Vs model shows low velocities at shallow depths (5-10 km) beneath the Mesopotamian foredeep, South Caspian Basin, and the Black Sea, possibly indicating a relatively thin crust. In contrast, the Zagros mountain range, the Sanandaj-Sirjan metamorphic zone in western central Iran, the easternmost Anatolian plateau and Lesser Caucasus are characterized by low velocities at these depths. Some of these anomalies may be related to thick crustal roots that support the high topography of these regions. In the upper mantle depth range, high-velocity anomalies are obtained beneath the Arabian Platform, southern Zagros, Persian Gulf and the eastern Mediterranean, in contrast to low velocities beneath the Red Sea, Arabian Shield, Afar depression, eastern Turkey and Lut Block in eastern Iran. Our Vs model may be used as a new reference crustal model for the Middle East in a broad range of future studies.

Abstract: High seismicity rates in eastern Indonesia occur due to the complex interaction of several tectonic plates which resulted in two deadly, destructive earthquake sequences that occurred in Lombok Island and the city of Palu, Sulawesi in 2018. The first sequence began in July with an Mw 6.4 event near Lombok, culminating in an Mw 7.0 event 8 d later. This was then followed by a nearby Mw 6.9 event 12 d later. Approximately 1000 km to the northeast, a separate sequence began several weeks later near Palu where an Mw 7.5 event occurred that triggered a tsunami. In this study, we present hypocentre relocations for both earthquake sequences as well as all other regional earthquakes in eastern Indonesia. The relocations were performed using a teleseismic double-difference relocation method and arrival times for P and S waves from stations at local, regional, and teleseismic distances. The catalogue and phase data were taken from the Agency for Meteorology, Climatology and Geophysics (BMKG) of Indonesia and the International Seismological Centre (ISC) for the period of April 2009 through November 2018. The relocated catalogue provides an improved view of seismicity in eastern Indonesia over the study period, sharpening locations and interpretations of seismogenic features throughout the region. In the Lombok area, the relocated earthquakes clearly show a backarc thrust to the north of the Sunda-Banda Arc transition zone. The relocated aftershocks show that the destructive Mw 7.0 and Mw 6.9 earthquakes of the Lombok sequence ruptured two different regions: The Mw 7.0 earthquake propagated westward, whereas the Mw 6.9 earthquake propagated eastward. The entire sequence of Lombok earthquakes was most likely started by the Mw 6.4 event as the initial event or foreshock, which then triggered backarc thrusts on both sides. Several weeks later and far to the northeast, the Mw 7.5 Palu earthquake occurred along the Palu-Koro Fault, filling a seismic gap that had not ruptured in an Mw 6.0 event or larger since at least 1900. The distribution of aftershocks indicates that the northern part of the Palu-Koro Fault has lower relative seismicity rates than the southern part at shallow depths, and that off fault aftershocks are mostly located to the east of the Palu-Koro Fault.

Abstract: The volcanism of the Eifel volcanic field (EVF), in west-central Germany, is often considered an example of hotspot volcanism given its geochemical signature and the putative mantle plume imaged underneath. EVF’s setting in a stable continental area provides a rare natural laboratory to image surface deformation and test the hypothesis of there being a thermally buoyant plume. Here we use Global Positioning System (GPS) data to robustly image vertical land motion (VLM) and horizontal strain rates over most of intraplate Europe. We find a spatially-coherent positive VLM anomaly over an area much larger than the EVF and with a maximum uplift of ~1 mm yr−1 at the EVF (when corrected for glacial isostatic adjustment). This rate is considerably higher than averaged over the Late-Quaternary. Over the same area that uplifts, we find significant horizontal extension surrounded by a radial pattern of shortening, a superposition that strongly suggests a common dynamic cause. Besides the Eifel, no other area in NW Europe shows significant positive VLM coupled with extensional strain rates, except for the much broader region of glacial isostatic adjustment. We refer to this 3D deformation anomaly as the Eifel Anomaly. We also find an extensional strain rate anomaly near the Massif Central volcanic field surrounded by radial shortening, but we do not detect a significant positive VLM signal there. The fact that the Eifel Anomaly is located above the Eifel plume suggests the plume causes the anomaly. Indeed, we show that buoyancy forces induced by the plume at the bottom of the lithosphere can explain this remarkable surface deformation. Plume-induced deformation can also explain the relatively high rate of regional seismicity, particularly along the Lower Rhine Embayment.

Abstract: Late-time transient electromagnetic (TEM) data contain deep subsurface information and are important for resolving deeper electrical structures. However, due to their relatively small signal amplitudes, TEM responses later in time are often dominated by ambient noises. Therefore, noise removal is critical to the application of TEM data in imaging electrical structures at depth. De-noising techniques for TEM data have been developed rapidly in recent years. Although strong efforts have been made to improving the quality of the TEM responses, it is still a challenge to effectively extract the signals due to unpredictable and irregular noises. In this study, we develop a new type of neural network architecture by combining the long short-term memory (LSTM) network with the autoencoder structure to suppress noise in TEM signals. The resulting LSTM-autoencoders yield excellent performance on synthetic data sets including horizontal components of the electric field and vertical component of the magnetic field generated by different sources such as dipole, loop and grounded line sources. The relative errors between the de-noised data sets and the corresponding noise-free transients are below 1% for most of the sampling points. Notable improvement in the resistivity structure inversion result is achieved using the TEM data de-noised by the LSTM-autoencoder in comparison with several widely-used neural networks, especially for later-arriving signals that are important for constraining deeper structures. We demonstrate the effectiveness and general applicability of the LSTM-autoencoder by de-noising experiments using synthetic 1-D and 3-D TEM signals as well as field data sets. The field data from a fixed loop survey using multiple receivers are greatly improved after de-noising by the LSTM-autoencoder, resulting in more consistent inversion models with significantly increased exploration depth. The LSTM-autoencoder is capable of enhancing the quality of the TEM signals at later times, which enables us to better resolve deeper electrical structures.

Abstract: Seismic full-waveform inversion (FWI) can produce high resolution images of the Earth's subsurface. Since full waveform modelling is significantly nonlinear with respect to velocities, Monte Carlo ...

Abstract: The spatial distribution of temporal variations in seismic wavespeed is key to understanding the sources and physical mechanisms of various geophysical processes. The imaging of wavespeed changes requires accurate measurements of traveltime delays with both high lapse-time and frequency resolutions. However, traditional methods for time-shift estimation suffer from their limited resolutions. In this paper we propose a new approach, the wavelet method, to measure the traveltime changes in the time–frequency domain. This method is based on wavelet cross-spectrum analysis, and can provide optimal time–frequency joint resolution while being computationally efficient. It can deal not only with coda but also dispersive surface waves even in the presence of cycle skipping. Using synthetic coda, we show that the wavelet method can retrieve traveltime shifts more stably and accurately than traditional methods. An application at Salton Sea Geothermal Field indicates that the wavelet method is less affected by spectral smearing and better discriminates dv/v variations at different frequencies. Furthermore, upon investigations on synthetic coda, we illustrate that the bias on dv/v measurements due to changes in source frequency content is likely to be negligible, either with traditional methods or with the new wavelet method. The wavelet method sheds lights on applications of seismic interferometry that aim to locate changes in space.

Abstract: The classical procedure of the probabilistic seismic hazard analysis (PSHA) requires a Poissonian distribution of earthquakes. Seismic catalogues follow a Poisson distribution just after the application of a declustering algorithm that leaves only one earthquake for each seismic sequence (usually the stronger, i.e. the main shock). Removing earthquakes from the seismic catalogues leads to underestimation of the annual rates of the events and consequently associate with low seismic hazard as indicated by several studies. In this study, we aim investigating the performance of two declustering methods on the Italian instrumental catalogue and the impact of declustering on estimation of the b-value and on the seismic hazard analysis. To this end, first the spatial variation in the seismicity rate was estimated from the declustered catalogues using the adaptive smoothed seismicity approach, considering small earthquakes (Mw ≥ 3.0). We then corrected the seismicity rates using new approach that allows for counting all events in the complete seismic catalogue by simply changing the magnitude frequency distribution. The impact of declustering on seismic hazard analysis is illustrated using PSHA maps in terms of peak ground acceleration and spectral acceleration in 2 s, with 10 per cent and 2 per cent probability of exceedance in 50 yr, for Italy. We observed that the hazard calculated from the declustered catalogues was always lower than the hazard computed using the complete catalogue. These results are in agreement with previous results obtained in different parts of the world.

Abstract: The time-averaged geomagnetic field is generally purported to be uniformitarian across Earth history—close to a geocentric axial dipole, with average strength within one order of magnitude of that at present. Nevertheless, recent studies have reported that the field was approximately ten times weaker than present in the mid-Palaeozoic (∼410–360 Ma) and late Ediacaran (∼565 Ma). Here we present the first whole-rock palaeointensity determinations of Ediacaran age outside of Laurentia. These were obtained by the Thellier-Coe, Wilson and microwave methods for basaltic rocks of 560–580 Ma age of the Ediacaran traps, southwestern margin of the East European Craton, Ukraine. All four studied sites showed extremely low instantaneous field values of (3–7) μT with corresponding VDMs of (0.4–1) × 1022 Am2. Summarizing all available data, the Ediacaran field appears to be anomalously characterized by ultra-low dipole moment and ultra-high reversal frequency. According to some geodynamo models, this state could indicate a weak dipole field regime prior to the nucleation of the solid inner core. However, given that ultra-low field intensities have also been detected in the Devonian, and that virtually no palaeointensity data exist for the intervening ∼150 Ma, the date of inner core nucleation remains extremely uncertain. Our new evidence of persistent ultra-weak magnetospheric shielding in the Ediacaran may be considered consistent with the recently hypothesized link between enhanced UV-B radiation in this interval and the subsequent Cambrian evolutionary radiation.

Abstract: Central Mongolia is a prominent region of intracontinental surface deformation and intraplate volcanism. To study these processes, which are poorly understood, we collected magnetotelluric data in the Hangai and Gobi-Altai region in central Mongolia and derived the first three-dimensional (3-D) resistivity model of the crustal and upper mantle structure in this region. The geological and tectonic history of this region is complex, resulting in features over a wide range of spatial scales, which that are coupled through a variety of geodynamic processes. Many Earth properties that are critical for the understanding of these processes, such as temperature as well as fluid and melt properties, affect the electrical conductivity in the subsurface. 3-D imaging using magnetotellurics can resolve the distribution of electrical conductivity within the Earth at scales ranging from tens of metres to hundreds of kilometres, thereby providing constraints on possible geodynamic scenarios. We present an approach to survey design, data acquisition, and inversion that aims to bridge various spatial scales while keeping the required field work and computational cost of the subsequent 3-D inversion feasible. Magnetotelluric transfer functions were estimated for a 650 x 400 km^2 grid, which included measurements on an array with regular 50 x 50 km^2 spacing and along several profiles with a denser 5-15 km spacing. The use of telluric-only data loggers on these profiles allowed for an efficient data acquisition with a high spatial resolution. A 3-D finite element forward modelling and inversion code was used to obtain the resistivity model. Locally refined unstructured hexahedral meshes allow for a multi-scale model parametrization and accurate topography representation. The inversion process was carried out over four stages, whereby the result from each stage was used as input for the following stage that included a finer model parametrization and/or additional data (i.e. more stations, wider frequency range). The final model reveals a detailed resistivity structure and fits the observed data well, across all periods and site locations, offering new insights into the subsurface structure of central Mongolia. A prominent feature is a large low-resistivity zone detected in the upper mantle. This feature suggests a non-uniform lithosphere-asthenosphere boundary that contains localized upwellings that shallow to a depth of 70 km, consistent with previous studies. The 3-D model reveals the complex geometry of the feature, which appears rooted below the Eastern Hangai Dome with a second smaller feature slightly south of the Hangai Dome. Within the highly resistive upper crust, several conductive anomalies are observed. These may be explained by late Cenozoic volcanic zones and modern geothermal areas, which appear linked to mantle structures, as well as by major fault systems, which mark terrane boundaries and mineralized zones. Well resolved, heterogeneous low-resistivity zones that permeate the lower crust may be explained by fluid-rich domains.

Abstract: Knowledge of crustal stress field is essential for understanding tectonics and earthquake generation. One way to estimate the crustal stress field is based on focal mechanisms of earthquakes. This ...

Abstract: The Groningen gas reservoir, situated in the northeast of the Netherlands, is western Europe’s largest producing gas field and has been in production since 1963. The gas production has induced both subsidence and seismicity. Seismicity is detected and located using the Koninklijk Nederlands Meteorologisch Instituut shallow-borehole array for the period 2015–2017, incorporating the back projection techniques of QuakeMigrate and the nonlinear location procedure to constrain earthquake locations and depths. The uncertainties on the estimated depths are estimated taking into account velocity model, changes in station array geometry and uncertainties in the measurement of arrival times of the P and S waves. We show that the depth distribution of seismicity is consistent with nucleation within the reservoir (28 per cent) or in the overburden (60 per cent) within ∼500 m from the top of the reservoir. Earthquakes with hypocentres in the overburden likely originate from overlying Zechstein anhydrite caprock. Based on their depth distribution, it seems like the earthquakes are primarily driven by the elastic strain in the reservoir and overburden, induced by the reservoir compaction. We estimate the probability of earthquakes nucleating beneath the reservoir in the underlying Carboniferous limestone and basement, to be no more than 12 per cent.

Abstract: S U M M A R Y We present an accelerated full-waveform inversion based on dynamic mini-batch optimization, which naturally exploits redundancies in observed data from different sources. The method rests on the selection of quasi-random subsets (mini-batches) of sources, used to approximate the misfit and the gradient of the complete data set. The size of the mini-batch is dynamically controlled by the desired quality of the gradient approximation. Within each mini-batch, redundancy is minimized by selecting sources with the largest angular differences between their respective gradients, and spatial coverage is maximized by selecting candidate events with Mitchell’s best-candidate algorithm. Information from sources not included in a specific minibatch is incorporated into each gradient calculation through a quasi-Newton approximation of the Hessian, and a consistent misfit measure is achieved through the inclusion of a control group of sources. By design, the dynamic mini-batch approach has several main advantages: (1) The use of mini-batches with adaptive size ensures that an optimally small number of sources is used in each iteration, thus potentially leading to significant computational savings; (2) curvature information is accumulated and exploited during the inversion, using a randomized quasiNewton method; (3) new data can be incorporated without the need to re-invert the complete data set, thereby enabling an evolutionary mode of full-waveform inversion. We illustrate our method using synthetic and real-data inversions for upper-mantle structure beneath the African Plate. In these specific examples, the dynamic mini-batch approach requires around 20 per cent of the computational resources in order to achieve data and model misfits that are comparable to those achieved by a standard full-waveform inversion where all sources are used in each iteration.

Abstract: P.N. Shebalin ,1 C. Narteau 2 and S.V. Baranov 1,3 1Institute of Earthquake Prediction Theory and Mathematical Geophysics, 84/32 Profsouznaya, Moscow 117997, Russia 2Université de Paris, Institut de physique du globe de Paris, CNRS, F-75005 Paris, France. E-mail: narteau@ipgp.fr 3Kola Branch, Federal Research Center Geophysical Survey, Russian Academy of Sciences, 14 Fersmana, Apatity 184209, Russia

Abstract: The Greater Geneva Basin is one of the key targets for geothermal exploration in Switzerland. Until recently, information about the subsurface structure of this region was mostly composed of well-logs, seismic reflection lines, and gravity measurements. As part of the current effort to further reduce subsurface uncertainty, and to test passive seismic methods for exploration purposes, we performed an ambient-noise tomography of the Greater Geneva Basin. We used ∼1.5 years of continuous data collected on a temporary seismic network composed of 28 broadband stations deployed within and around the basin. From the vertical component of the continuous noise recordings, we computed cross-correlation functions and retrieved Rayleigh-wave group-velocity dispersion curves. We then inverted the dispersion curves to obtain 2D group-velocity maps and proceeded to a subsequent inversion step to retrieve a large-scale 3D shear-wave velocity model of the basin. We discuss the retrieved features of the basin in the light of local geology, previously acquired geophysical datasets, and ongoing geothermal exploration. The Greater Geneva Basin is an ideal natural laboratory to test innovative geothermal exploration methods because of the substantial geophysical datasets available for comparison. While we point out the limits of ambient-noise exploration with sparse networks and current methodology, we also discuss possible ways to develop ambient-noise tomography as an affordable and efficient subsurface exploration method.

Abstract: Aurélien Mordret, Roméo Courbis2−3, Florent Brenguier, Małgorzata Chmiel2−3, Stéphane Garambois, Shujuan Mao, Pierre Boué, Xander Campman, Thomas Lecocq, Wim Van der Veen and Dan Hollis 1Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology (MIT), Cambridge, Massachusetts, USA 2Université Grenoble Alpes, Univ. Savoie Mont Blanc, CNRS, IRD, IFSTTAR, ISTerre, UMR 5275, 38000 Grenoble, France 3Sisprobe, Meylan, 38240, France 4Shell Global Solutions International BV, Amsterdam, the Netherlands 5Royal Observatory of Belgium Seismology, Avenue Circulaire, 3, BE-1180 Brussels, Belgium 6Nederlandse Aardolie Maatschappij. 3

Abstract: We present an improved crustal Vs model and Moho depth map using ambient noise wave-equation tomography. The so-called 'ambient noise wave-equation tomography' is a method to invert seismic ambient noise phase dispersion data based on elastic waveform simulation, which accounts for 3-D and finite frequency effects. We use cross-correlations of up to four years of continuous vertical-component ambient seismic noise recordings from 304 high-quality broadband stations in the Alpine region. We use model LSP_Eucrust1.0 obtained from traditional ambient noise tomography as initial model, and we iteratively improve the initial model by minimizing frequency-dependent phase traveltime differences between the observed and synthetic waveforms of Rayleigh waves in the period range 10-50 s. We obtain the final model after 15 iterations with ∼65% total misfit reduction compared to the initial model. At crustal depth, the final model significantly enhances the amplitudes and adjusts the shapes of velocity anomalies. At Moho and upper mantle depth, the final model corrects an obvious systematic velocity shift of the initial model. The resulting iso-velocity Moho map confirms a Moho step along the external side of the external crystalline massifs of the Northwestern Alps and reveals underplated gabbroic plutons in the lower most crust of the Central and Eastern Alps. Ambient noise wave-equation tomography turns out to be a useful tool to refine shear-wave velocity models obtained by traditional ambient noise tomography based on ray theory.

Abstract: Spectral methods have been applied for more than 40 yr to calculate the depth to magnetic sources and the Curie point depth or Moho depth. Unfortunately, conceptual, theoretical and methodological errors, as well as, subjective approaches and lack of rigor in the application of the method are common. This has affected reproducibility, credibility and confidence in this method. We carried out a detailed mathematical and methodological revision of the spectral methods (centroid, spectral peak, forward and fractal methods) applied to magnetic data for the determination of the depth to magnetic sources. Moreover, a systematic analysis of more than 110 articles was done in order to compile and compare the different approaches and values of key parameters used by different authors in their calculations. We discuss difficulties, conditions and methodological procedures, as well as several misapplications and misinterpretation of the different methods.

Abstract: Station noise levels play a fundamental limitation in our ability to detect seismic signals. These noise levels are frequency-dependent and arise from a number of physically different drivers. At periods greater than 100 s, station noise levels are often limited by the self-noise of the instrument as well as the sensitivity of the instrument to non-seismic noise sources. Recently, station operators in the Global Seismographic Network (GSN) have deployed several Streckeisen STS-6A very broad-band borehole seismometers. These sensors provide a potential replacement for the no-longer-produced Streckeisen STS-1 seismometer and the GeoTech KS-54 000 borehole seismometer. Along with showing some of the initial observational improvements from installing modern very broad-band seismometers at depth, we look at current limitations in the seismic resolution from earth tide periods 100 000 s (0.01 mHz) to Nyquist at most GSN sites (0.02 s or 50 Hz). Finally, we show the potential for improved observations of continuously excited horizontal Earth hum as well as the splitting of very long-period torsional modes. Both of these observations make use of the low horizontal noise levels which are obtained by installing very broad-band borehole seismometers at depth.

Abstract: The horizontal-to-vertical spectral ratio (HVSR) of ambient noise measurement is commonly used to estimate a site's resonance frequency (${f_0}$). For sites with a strong impedance contrast, the HVSR peak frequency (${f_{0,\\mathrm{ HVSR}}}$) has been shown to be a good estimate of ${f_0}$. However, the random nature of ambient noise (both in time and space), in conjunction with variable environmental conditions and sensor coupling issues, can lead to uncertainty in ${f_{0,\\mathrm{ HVSR}}}$ estimates. Hence, it is important to report ${f_{0,\\mathrm{ HVSR}}}$ in a statistical manner (e.g. as a mean or median value with standard deviation). In this paper, we first discuss widely accepted procedures to process HVSR data and estimate the variance in ${f_{0,\\mathrm{ HVSR}}}$. Then, we propose modifications to improve these procedures in two specific ways. First, we propose using a lognormal distribution to describe ${f_{0,\\mathrm{ HVSR}}}$ rather than the more commonly used normal distribution. The use of a lognormal distribution for ${f_{0,\\mathrm{ HVSR}}}$ has several advantages, including consistency with earthquake ground motion processing and allowing for a seamless transition between HVSR statistics in terms of both frequency and its reciprocal, period. Second, we introduce a new frequency-domain window-rejection algorithm to decrease variance and enhance data quality. Finally, we use examples of 114 high-variance HVSR measurements and 77 low-variance HVSR measurements collected at two case study sites to demonstrate the effectiveness of the new rejection algorithm and the proposed statistical approach. To encourage their adoption, and promote standardization, the rejection algorithm and lognormal statistics presented in this paper have been incorporated into hvsrpy, an open-source Python package for HVSR processing.

Abstract: Linear arrays are usually deployed for passive surface-wave investigations because of their high efficiency and convenience. In populated urban areas, it is almost impossible to set up a 2-D array in terms of the restriction from the existing infrastructures. The limited azimuthal coverage, however, lacks the ability to attenuate velocity overestimation caused by directional noise sources. We came up with a novel idea to compensate the azimuthal coverage by adding two more offline receivers to a conventional linear array, which is called pseudo-linear-array analysis of passive surface waves (PLAS). We used a beamforming algorithm to capture noise sources distribution and extract accurate dispersion curves. We used array response function to explain the superiority of the pseudo-linear array over the linear array and present the basic workflow of PLAS. Synthetic tests and field examples demonstrated the feasibility of PLAS to measure unbiased dispersion image. Comparison with mostly used passive surface wave methods (refraction microtremor, multichannel analysis of passive surface waves, spatial autocorrelation method, frequency–wavenumber analysis) suggested that PLAS can serve as an alternative passive surface wave method, especially in urban areas with restricted land accessibility and short-time acquisition demands.

Abstract: Teleseismic waveforms contain information on fault slip evolution during an earthquake, as well as on the fault geometry. A linear finite-fault inversion method is a tool for solving the slip-rate function distribution under an assumption of fault geometry as a single or multiple-fault-plane model. An inappropriate assumption of fault geometry would tend to distort the solution due to Greens function modelling errors. We developed a new inversion method to extract information on fault geometry along with the slip-rate function from observed teleseismic waveforms. In this method, as in most previous studies, we assumed a flat fault plane, but we allowed arbitrary directions of slip not necessarily parallel to the assumed fault plane. More precisely, the method represents fault slip on the assumed fault by the superposition of five basis components of potency-density tensor, which can express arbitrary fault slip that occurs underground. We tested the developed method by applying it to real teleseismic P waveforms of the Mw 7.7 2013 Balochistan, Pakistan, earthquake, which is thought to have occurred along a curved fault system. The obtained spatiotemporal distribution of potency-density tensors showed that the focal mechanism at each source knot was dominated by a strike-slip component with successive strike angle rotation from 205° to 240° as the rupture propagated unilaterally towards the south-west from the epicentre. This result is consistent with Earths surface deformation observed in optical satellite images. The success of the developed method is attributable to the fact that teleseismic body waves are not very sensitive to the spatial location of fault slip, whereas they are very sensitive to the direction of fault slip. The method may be a powerful tool to extract information on fault geometry along with the slip-rate function without requiring detailed assumptions about fault geometry.

Abstract: We develop a theoretical framework for framing and solving probabilistic linear(ized) inverse problems in function spaces. This is built on the statistical theory of Gaussian Processes, and allows results to be obtained independent of any basis, avoiding any difficulties associated with the fidelity of representation that can be achieved. We show that the results of Backus–Gilbert theory can be fully understood within our framework, although there is not an exact equivalence due to fundamental differences of philosophy between the two approaches. Nevertheless, our work can be seen to unify several strands of linear inverse theory, and connects it to a large body of work in machine learning. We illustrate the application of our theory using a simple example, involving determination of Earth’s radial density structure.

Abstract: A 2-D orthogonal distributed acoustic sensing (DAS) array designed for seismic experiments was buried horizontally beneath the Kafadar Commons Geophysical Laboratory on the Colorado School of Mines campus at Golden, Colorado. The DAS system using straight fibre-optic cables is a cost-efficient technology that enables dense seismic array deployment for long-term seismic monitoring, favouring both earthquake-based and ambient-noise-based surface wave analysis for subsurface characterization. In our study, the horizontally orthogonal DAS array records ambient noise data for a period of about two months from November 2018 to January 2019. During this time, the array also detected seismic signals from an ML3.6 earthquake at Glenwood Springs, Colorado, which exhibit opposite signal polarities in the orthogonal DAS section recordings. We derive the transformation matrix for DAS strain measurements in horizontally orthogonal cables to retrieve both Rayleigh and Love wave dispersion information from the single-component DAS signals using the 2-D multichannel analysis of surface waves method. In addition, ambient noise interferometry is applied to long-term DAS noise recordings. Our theoretical derivation demonstrates that Rayleigh and Love wave Green's functions are coupled in the noise cross-correlation functions (NCFs) of DAS receiver pairs. Stacking NCFs over the horizontally orthogonal DAS array can constructively recover the radial Rayleigh wave component but destructively suppress the Love wave component. The multimodal Monte Carlo inversion of the earthquake-based Rayleigh wave and Love wave dispersion measurements and the noise-based Rayleigh wave measurement reveals a 1-D layered structure that agrees qualitatively with geological surveys of the site. Our study demonstrates that although straight fibre-optic cables lack broadside sensitivity, using appropriate DAS array configuration and seismic array methods can extend the seismic acquisition ability of DAS and enable its application to a broad range of scenarios.

Abstract: In a previous paper, we introduced a framework for carrying out petrophysically and geologically guided geophysical inversions. In that framework, petrophysical and geological information is modelled with a Gaussian Mixture Model (GMM). In the inversion, the GMM serves as a prior for the geophysical model. The formulation was confined to problems in which a single physical property model was sought, with a single geophysical dataset. In this paper, we extend that framework to jointly invert multiple geophysical datasets that depend on multiple physical properties. The petrophysical and geological information is used to couple geophysical surveys that, otherwise, rely on independent physics. This requires advancements in two areas. First, an extension from a univariate to a multivariate analysis of the petrophysical data, and their inclusion within the inverse problem, is necessary. Second, we address the practical issues of simultaneously inverting data from multiple surveys and finding a solution that acceptably reproduces each one, along with the petrophysical and geological information. To illustrate the efficacy of our approach and the advantages of carrying out multi-physics inversions, we invert synthetic gravity and magnetic data associated with a kimberlite deposit. The kimberlite pipe contains two distinct facies embedded in a host rock. Inverting the datasets individually leads to a binary geological model: background or kimberlite. A multi-physics inversion, with petrophysical information, differentiates between the two main kimberlite facies of the pipe. Through this example, we also highlight the capabilities of our framework to work with interpretive geologic assumptions when minimal quantitative information is available. In those cases, the dynamic updates of the Gaussian Mixture Model allow us to perform multi-physics inversions by learning a petrophysical model.