We present the results of a 3-D seismic tomography study of the upper crust beneath a quadrant of the peak ring structure of the Chicxulub meteorite impact crater, Mexico. Reflection and refraction traveltimes from a grid of seismic profiles recorded by a 6 km streamer and 48 ocean bottom seismometer stations were inverted to give a well-resolved 3-D velocity model to a maximum depth of 6-8 km. The model comprised the thin water layer, a layer of low seismic velocity post-impact sedimentary infill and the crater basement, which was separated from the fill by the interface representing the top of the crater, defined by normal incidence reflection picks. The crater basement shows a cylinder-shaped feature extending vertically downwards beneath the topographic peak ring to at least 8 km, the depth of resolution of this survey, characterized by slower seismic velocities than in the surrounding rocks at the same depth. This result supports and extends the observations of previous seismic refraction work in the peak ring, and also of scientific drilling, that the material in the peak ring has significantly reduced seismic velocity compared to typical granitic basement lithologies. We used the velocity model to perform a pre-stack depth migration of a key seismic reflection profile. In the best-fitting model presented here the prominent dipping reflector previously identified on seismic reflection profiles, which projects to the outer edge of the peak ring, dips inwards and crosses the low velocity cylinder without an apparent first order contrast in impedance. This result implies that the reflectivity of this dipping reflector is due to a thin, high-contrast layer such as entrapped impact melt or hydrothermal alteration within the overturned structures. We also identify well-imaged slump blocks from the crater rim/inner ring inwards, variations in the height and width of the peak ring, and associated variations in the velocity contrast that characterizes the anomaly beneath the peak ring.

Summary We apply adaptive waveform inversion, reflection waveform inversion and broadband FWI imaging to conventional narrow-azimuth towed-streamer seismic data acquired over a deep-water heterogeneous basaltic section. The resulting high-resolution velocity model reveals the structure and physical properties of the basaltic layers in the sub-surface. Spatial differentiation of the broadband acoustic-impedance model recovered using short-offset 60-Hz FWI applied to the unprocessed field data generates a multiple-free deghosted true-amplitude depth-migrated image of acoustic reflectivity. The image is significantly superior to the existing conventional PSDM generated using conventional processing. Parameter selection and real-time dynamic control of the inversion are automated, driven using measurements on data and model that include local phase differences between predicted and observed data weighted by amplitude envelope.

Summary Surface-offset common image gathers (SOCIGs) are an essential tool in processes of velocity model building and reservoir characterisation. However, wave-equation-propagation based methods of FWI and RTM do not directly produce SOCIGs due to their simultaneous backward-propagation of data from all offsets, which thereby limites the versatality of these methods. To overcome this limitation, we propose to generate SOCIGs via single-trace backward wavefield synthesis. The method runs at a computational cost of a fraction of that of backward-propagating each offset/trace separately using a conventional finite-difference solver, and its cost effectiveness increases for higher dimensionality, higher-order stencils, and more-complete wave equations. We demonstrate the efficacy and efficiency of the proposed method using a 2D towed-streamer synthetic example, and a real 3D land dataset.

To enhance the efficiency of mine planning, mining companies wish to understand the structure and extent of ore-bearing rocks as well as possible. Conventional seismic reflection surveys are not well suited for this purpose as they provide an image containing only the location of reflectors, and do not provide physical property information to discriminate between ore and gangue material. Full-waveform inversion (FWI) is a powerful inversion technique, which is able to recover the physical properties of the subsurface at a far greater spatial resolution and accuracy than conventional seismic methods. In this study, we synthetically examined the feasibility of using FWI to image quartz vein-hosted gold deposits. We utilised the Curraghinalt gold deposit in Northern Ireland to parameterise our models, where mineralisation is bound entirely to thin (1–3 m) and steeply dipping (>45°) quartz sulphide veins. Firstly, we demonstrated that a conventional surface seismic reflection survey geometry alongside FWI is infeasible for imaging quartz vein-hosted gold deposits. Secondly, we explored a cross-hole seismic survey geometry consisting of sources and receivers placed down vertical boreholes. This cross-hole survey geometry is capable of generating synthetic datasets such that FWI can recover the position of the veins in space accurate to within 0.5 m relative to their true positions, and recover their physical properties with an accuracy greater than 90%, beginning from an entirely homogeneous starting model. We conclude it is essential the source and receiver boreholes be positioned such that both transmitted and reflected arrivals are present in the datasets, otherwise FWI will fail to accurately recover the position and physical properties of the veins. This opens a new avenue for FWI to play a major role in the planning stages and development of gold mines around the world.

The anisotropic full-waveform inversion (FWI) is a seis-mic inverse problem for multiple parameters, which aims to simultaneously reconstruct the vertical velocity and the anisotropic parameters of the earth's subsurface. This multi -parameter inverse problem suffers from two issues. First, the objective function of the data fitting is less sensitive to the anisotropic parameters. Second, the crosstalk effect among the different parameters worsens the model update in the iterative inversion. We have developed a method that sta-tistically regularizes the anisotropic FWI using Wasserstein adversarial networks, by penalizing the Wasserstein distance between the distribution of the current model parameters and that of the parameters at the borehole locations. The regu-larizer can mitigate the issues of anisotropic FWI with multi-ple parameters and therefore it also can be applied to other inverse problems with multiple parameters.

Objective: Despite being a low-cost, portable and safe medical imaging technique, transcranial ultrasound imaging is not used widely in adults because of the severe degradation and distortion of signals caused by the skull. Full-waveform inversion (FWI) has recently been found to have potential as an effective method for transcranial ultra-sound tomography to obtain high-quality, subwavelength-resolution acoustic models of the brain using low -frequency ultrasound data. In this study is the first demonstration of this method in recovering a high-resolution 2-D reconstruction of a brain and skull ultrasound imaging phantom using experimentally acquired data.Methods: A 2:5 scale brain phantom encased within a 3-D-printed skull-mimicking layer was created to simulate a clinical transcranial imaging target. To obtain tomographic ultrasound data on the brain and skull phantom, a tomo-graphic ultrasound acquisition system was designed and implemented using commercially available low-frequency cardiac probes. FWI reconstructions of the brain and skull phantom were performed using the acquired tomographic data and were compared with corresponding synthetic reconstructions. This comparison was used to evaluate the feasibility of the proposed imaging system when employing different transducer array configurations.Results: We demonstrate the successful FWI reconstruction of the brain phantom within the skull mimic from experi-mentally acquired tomographic ultrasound data. To mitigate the effects of the skull-mimicking material, a reflection-matching algorithm was applied to model the morphology of the skull layer prior to performing the inversion.Conclusion: The findings of this study provide a promising step toward the clinical use of FWI for transcranial ultrasound imaging in adults.

Many geophysical tasks are hindered in practice by the high costs of generating and processing data. We have developed a potential solution, mitigating the cost of certain data acquisition and generation processes by using data-domain-translation deep neural networks. Generative adversarial networks have demonstrated success in data translation in a wide variety of applications. By providing training data from domain A and domain B, networks can be trained to estimate the distributions of both domains, and hence establish a mapping from one to the other. We apply such data-translation neural networks to 3D geophysical field data examples and determine that they can be used as cost-reduction tools, providing an efficient mapping between different data types of interest in expensive data-processing workflows. Our approach is especially relevant for the translation between acoustic and elastic data sets during full-waveform inversion, which mitigates the elastic effect in the acoustic inversion.

We demonstrate that accurate amplitude-vs-angle param-eters can be extracted from raw unprocessed seismic data using purely acoustic full-waveform inversion. The result-ant parameters incorporate the full elastic response of the observed data, and it is not necessary to use elastic FWI in order to determine AVA. This approach naturally corrects for a multitude of other amplitude effects including reflector geometry and transmission losses, and it deals correctly with near-critical and post-critical reflections. FWI-based work-flows are consequently simpler than conventional work-flows, and AVA parameters can normally be generated by FWI within days of raw field data first becoming available.

Summary We demonstrate that purely acoustic FWI, run at true amplitude, can be used to extract accurate AVO parameters using offset-restricted subsets of raw unprocessed seismic data. Rather than producing conventional AVO parameters directly, acoustic FWI instead produces AVO anomalies showing the departure of the input data from the AVO displayed by a purely acoustic model. It is trivial to transform this into conventional AVO. AVO extraction via acoustic FWI applied to elastic synthetic data show that the AVO recovery using acoustic FWI is near perfect. When combined with full-bandwidth FWI, this approach removes any requirement for conventional data processing, model building, explicit migration or Kirchhoff-based AVO extraction. A final depth-migrated reflectivity volume, and accurate AVO parameters, can both be generated purely by acoustic FWI.

Arc volcanoes are underlain by complex systems of molten-rock reservoirs ranging from melt-poor mush zones to melt-rich magma chambers. Petrological and satellite data indicate that eruptible magma chambers form in the topmost few kilometres of the crust. However, no such a chamber has ever been imaged unambiguously, suggesting that large chambers responsible for caldera-forming eruptions are too short-lived to capture. Here we use a high-resolution imaging method based on finite-length seismic waveforms to detect a small, high-melt-fraction magma chamber embedded in a melt reservoir extending from ~2 to at least 4 km b.s.l. beneath Kolumbo – a submarine volcano near Santorini, Greece. The chamber coincides with the termination point of the recent earthquake swarms, and may be a missing link between a deeper melt reservoir and the high-temperature hydrothermal system venting at the crater floor. Though too small to be detected by standard seismic tomography, the chamber is large enough to threaten the nearby islands with tsunamigenic eruptions. Our results suggest that similar reservoirs (relatively small but high melt-fraction) may have gone undetected, and are yet to be discovered, at other active volcanoes.

Surface-offset common-image gathers (CIGs) are an important data format for seismic velocity analysis. However, the reverse time migration (RTM) method, which is wavefield propagation based, does not directly produce surface-offset CIGs because it propagates waves from all the offsets together. Here, we implement the generation of surface-offset CIGs by synthesizing the backward wavefields using the forward wavefields at locations where source and receiver locations overlap. This method produces surface-offset CIGs with high accuracy and low cost when compared with those generated by backpropagating each trace separately. We adopt nonnegative least-squares filters for sparse linear deconvolution. The computational effectiveness of the proposed method increases for higher dimensions, higher-order stencils, and more complicated wave equations. The proposed method works stably on a realistic towed-streamer acquisition system with moderate geometric positioning errors between the locations of airguns and hydrophones.

Summary The central objective of advanced Full Waveform Inversion is to enable rapid turnaround of accurate velocities directly from raw seismic data. AWI with its convolutional filter-based residual represents a fundamental change in the way FWI is normally run, as an ‘add on’ to time-consuming velocity- model building performed on pre-processed data, where its role is to finesse a tomography starting model. Here we show the combined solution of AWI and RWI known collectively as XWI serves as a predictor for unseen drilling logs. The inversion is run on raw data from NW Australia (6 sailline validation test) demonstrating convergence to essentially the same result from two simple 1D starting models. It is able to predict deviations in a sonic log from a starting position over 1000 m/s away from the measured value. The cloud environment where XWI ingests traces directly from blob storage consists of a dynamic pool of interruptible compute instances. This allows for cost-effective frequency sweeps through the iterations and scalable hyperparameter scans. It also is configured for interfacing XWI with interactive processing software for seamless trace preparation and project start up.

Summary FWI run to full bandwidth can remove multiples and ghosts from raw field data. Differentiation of the resultant high-resolution velocity model then allows the generation of a full-bandwidth PSDM reflectivity image without conventional processing or migration. We have run FWI to 100 Hz on raw field data from a marine towed-streamer dataset. We show that the results are similar to, and are broader bandwidth than, conventionally processed PSDM images. When run on the public cloud, this approach allows the production of final full-bandwidth PSDM volumes within in a few days of completion of data acquisition.

Full‐waveform inversion (FWI) can resolve subsurface physical properties to high resolutions, yet high‐performance computing resources have only recently made it practical to invert for high frequencies. A benefit of high‐frequency FWI is that recovered velocity models can be differentiated in space to produce high‐quality depth images (FWI images) of a comparable resolution to conventional reflection images.

Magnetic resonance imaging and X-ray computed tomography provide the two principal methods available for imaging the brain at high spatial resolution, but these methods are not easily portable and cannot be applied safely to all patients. Ultrasound imaging is portable and universally safe, but existing modalities cannot image usefully inside the adult human skull. We use in silico simulations to demonstrate that full-waveform inversion, a computational technique originally developed in geophysics, is able to generate accurate three-dimensional images of the brain with sub-millimetre resolution. This approach overcomes the familiar problems of conventional ultrasound neuroimaging by using the following: transcranial ultrasound that is not obscured by strong reflections from the skull, low frequencies that are readily transmitted with good signal-to-noise ratio, an accurate wave equation that properly accounts for the physics of wave propagation, and adaptive waveform inversion that is able to create an accurate model of the skull that then compensates properly for wavefront distortion. Laboratory ultrasound data, using ex vivo human skulls and in vivo transcranial signals, demonstrate that our computational experiments mimic the penetration and signal-to-noise ratios expected in clinical applications. This form of non-invasive neuroimaging has the potential for the rapid diagnosis of stroke and head trauma, and for the provision of routine monitoring of a wide range of neurological conditions.

Elastic FWI depends upon an accurate estimate of a constraining Vp to Vs ratio. Such a relation can be obtained empirically from rock-physics relations or from the analysis of the seismic data. The first is case-dependent and the second requires intense human intervention. Herein, we report a new method for a semi-automatic estimation of Vp to Vs ratios from seismic data requiring only a waveform inversion algorithm and minimal data intervention. We show synthetic examples and a real-data case study. Presentation Date: Wednesday, September 18, 2019 Session Start Time: 8:30 AM Presentation Time: 10:10 AM Location: 302B Presentation Type: Oral

Full-waveform inversion (FWI) is a promising technique for recovering the earth models for exploration geophysics and global seismology. FWI is generally formulated as the minimization of an objective function, defined as the L2-norm of the data residuals. The nonconvex nature of this objective function is one of the main obstacles for the successful application of FWI. A key manifestation of this nonconvexity is cycle skipping, which happens if the predicted data are more than half a cycle away from the recorded data. We have developed the concept of intermediate data for tackling cycle skipping. This intermediate data set is created to sit between predicted and recorded data, and it is less than half a cycle away from the predicted data. Inverting the intermediate data rather than the cycle-skipped recorded data can then circumvent cycle skipping. We applied this concept to invert cycle-skipped first arrivals. First, we picked up the first breaks of the predicted data and the recorded data. Second, we linearly scaled down the time difference between the two first breaks of each shot into a series of time shifts, the maximum of which was less than half a cycle, for each trace in this shot. Third, we moved the predicted data with the corresponding time shifts to create the intermediate data. Finally, we inverted the intermediate data rather than the recorded data. Because the intermediate data are not cycle-skipped and contain the traveltime information of the recorded data, FWI with intermediate data updates the background velocity model in the correct direction. Thus, it produces a background velocity model accurate enough for carrying out conventional FWI to rebuild the intermediate- and short-wavelength components of the velocity model. Our numerical examples using synthetic data validate the intermediate-data concept for tackling cycle skipping and demonstrate its effectiveness for the application to first arrivals.