When humans started started exploiting the abundant underground natural resources the Earth has to offer such as hydrocarbons, minerals and heat, we started to experience earthquakes that are related to this exploitation, so called induced earthquakes. Under certain conditions those can damage local infrastructure. However, most events are weak and only sensed by seismic sensors. Microseismic monitoring plays a vital role to optimize and insure the safety of these underground activities and new technologies such as carbon capture and storage. One key task besides the detection of microseimsic events is to determine the source location of these events using data recorded at the surface. In this thesis we investigate a method to localize weak microseismic events, using a deterministic approach, assuming a dense network of sensors. In simple words this method takes the seismic signals recorded at the Earth’s surface and sends them back into the Earth, where the signals start to focus at the point they originated from. This focusing method uses one-way wavefield extrapolation with an estimate of the background velocity model. The advantage of this method is that the weak signals recorded by the different sensors at the surface are amplified as they approach the location of the event that emitted the signal due to constructive interference. However, this is not enough to reliably recover the source location because typically earthquakes do not radiate seismic waves evenly; complex radiation patterns are typically observed depending on the mechanical properties of the rupture. To obtain a strong focused signal at the optimal source location we therefore perform a grid search over possible source mechanisms and increase the strength of the signal by deconvolution. Without taking the source mechanism into account we are not able to obtain accurate source locations, especially at low signal-to-noise ratios. However, by taking the source mechanism into account we are able to retrieve accurate source locations while also retrieving information about the source mechanism. Good results were obtained for 2D synthetic data for both a simple subsurface model as well as the realistic Annerveen salt model even when realistic noise was added... @en

Retrieving accurate microseismic source locations induced by hydraulic-fracturing operations is an important step to gain insights into the hydraulically stimulated reservoir volume. Recently, deep neural networks have been proposed that directly recover source locations from the seismic waveforms. The optimal performance of the proposed deep neural networks usually requires large training sets. The need for a large training set can be circumvented if a previously trained deep neural network can be used to start the training process with its weights instead of randomly initialized weights. These weights can then be fine-tuned using a smaller training set, which is also known as transfer learning. In this work, we implement a transfer learning workflow to update the weights of a deep neural network that was initially trained on a large synthetic dataset to localize microseismic events. We present two methods of processing, namely one post-monitoring mode and one continuous mode where the processing takes place during the monitoring period. We apply the methods to field data from a hydraulic fracturing site in Texas, USA. In the first scenario, a subset of the field data from the entire monitoring period is used to update the weights of the deep neural network, which is then applied to the remaining data resulting in mean and median distances of 227 and 182 m, respectively, compared to the results of a good localization method. In the second scenario, the deep neural network is updated daily with previously detected and located events and applied to the events detected the following day. Since the observed data used for training generally do not cover a wide range of source locations, we enrich the training set with synthetic data. The addition of synthetics for transfer learning ensures that the updated deep neural network provides accurate source locations for events with locations far from locations used during transfer learning. Transfer learning combining synthetic and real data performs significantly better (more consistent) locations than transfer learning without synthetics.

@en

A main challenge in microseismic monitoring is that the seismic signals recorded at the Earth's surface are weak and thus localization of those microseismic earthquakes becomes challenging. Diffraction stacking is a traditional method used to localize weak earthquakes, which involves stacking the waveforms along precomputed travel-time curves from different locations, where the maximum is used to determine the source location. In this work we aim to recover the source location of weak microseismic earthquakes using a deep neural network (DNN) that resembles the U-Net but uses fewer skip connections. However, the size of the field data is too small to train the DNN from scratch. Thus, we propose to pretrain a DNN using synthetic data that resembles the field data and that learns to map the source location in terms of a 3D Gaussian distribution directly from the seismic signals. This pretrained DNN is capable of localizing the higher magnitude earthquakes in the field data, but fails for the weaker earthquakes. To be able to localize the weaker magnitude earthquakes we therefore, fine tune the pretrained DNN using the higher magnitude field-data earthquakes. We observe that the updated model is able to extrapolate the information learned during the fine tuning step from higher magnitude earthquake data to lower magnitude earthquake data.@en

Hydraulic fracturing plays an important role when it comes to the extraction of resources in unconventional reservoirs. The microseismic activity arising during hydraulic fracturing operations needs to be monitored to both improve productivity and to make decisions about mitigation measures. Recently, deep learning methods have been investigated to localize earthquakes given field-data waveforms as input. For optimal results, these methods require large field data sets that cover the entire region of interest. In practice, such data sets are often scarce. To overcome this shortcoming, we propose initially to use a (large) synthetic data set with full waveforms to train a U-Net that reconstructs the source location as a 3D Gaussian distribution. As field data set for our study we use data recorded during hydraulic fracturing operations in Texas. Synthetic waveforms were modelled using a velocity model from the site that was also used for a conventional diffraction-stacking (DS) approach. To increase the U-Nets ability to localize seismic events, we augmented the synthetic data with different techniques, including the addition of field noise. We select the best performing U-Net using 22 events that have previously been identified to be confidently localized by DS and apply that U-Net to all 1245 events. We compare our predicted locations to DS and the DS locations refined by a relative location (DSRL) method. The U-Net based locations are better constrained in depth compared to DS and the mean hypocenter difference with respect to DSRL locations is 163 meters. This shows potential for the use of synthetic data to complement or replace field data for training. Furthermore, after training, the method returns the source locations in near real-time given the full waveforms, alleviating the need to pick arrival times.@en

Continuous seismic monitoring systems aid us to act on preventing strong earthquakes, such as induced by oil and gas extraction, deep geothermal systems and carbon sequestration. These systems provide the data to detect and locate such events and to determine their source magnitude and mechanism. Estimating the location of the source is one of the first parameters to be determined. Most source localization methods in the field of induced seismicity combine migration-based operators with grid-search algorithms. These are computationally intensive and therefore not instantaneous. To improve the time to locate events via grid-search based methods, we investigate the use of Convolutional Neural Networks (ConvNets) to reduce the search space. This is achieved by feeding the ConvNets with the waveforms and outputting a possible area/volume for the source location. Once the ConvNet is trained, it can produce the output almost in real-time. Therefore, it can be used by grid-search type approaches to focus the grid search over the smaller volume provided by the ConvNet. In this study we train ConvNets with synthetic data and apply them to field data. To our knowledge this is the first attempt of training ConvNets on synthetic data for the task of earthquake localization of field data.

@en

Waveform inversion is a promising method for ultrasound computed tomography able to produce high-resolution images of human breast tissue. However, the computational complexity of waveform inversion remains a considerable challenge, and the costs per iteration are proportional to the number of emitting transducers. We propose a twofold strategy to accelerate the time-to-solution by identifying the optimal number and location of emitters using sequential optimal experimental design (SOED). SOED is a powerful tool to iteratively add the most informative transducer or remove redundant measurements, respectively. This approach simultaneously provides optimized transducer configurations and a cost-benefit curve that quantifies the information gain versus the computational cost. First, we propose a method to identify the emitters that provide reconstructions with minimal expected uncertainties. Using a Bayesian approach, model uncertainties and resolution can be quantified with the trace of the posterior covariance. By linearizing the wave equation, we can compute the posterior covariance using the inverse of the Gauss-Newton approximation of the Hessian. Furthermore, this posterior is independent of the breast model and the experimental data, thus enabling pre-acquisition experimental optimization. Then, for the post-acquisition inversion, we present an approach to select a subsample of sources that accurately approximates the full gradient direction in each iteration. We control the convergence of the angular differences between consecutive gradient directions by randomly adding new emitters into the subsample. We present synthetic studies in 2D and 3D that consider a ring-shaped and a semi-ellipsoidal scanning device, respectively. Numerical results suggest that the provided methods have the potential to identify redundancies from the corresponding cost-benefit curves. Furthermore, the gradient direction rapidly converges to the direction of the full gradient, which appears to be independent of the model and the emitter locations.

@en

Waveform inversion for ultrasound computed tomography (USCT) is a promising imaging technique for breast cancer screening. However, the improved spatial resolution and the ability to constrain multiple parameters simultaneously demand substantial computational resources for the recurring simulations of the wave equation. Hence, it is crucial to use fast and accurate methods for numerical wave propagation, on the one hand, and to keep the number of required simulations as small as possible, on the other hand. We present an efficient strategy for acoustic waveform inversion that combines (i) a spectral-element continuous Galerkin method for solving the wave equation, (ii) conforming hexahedral mesh generation to discretize the scanning device, (iii) a randomized descent method based on mini-batches to reduce the computational cost for misfit and gradient computations, and (iv) a trust-region method using a quasi-Newton approximation of the Hessian to iteratively solve the inverse problem. This approach combines ideas and state-of-the-art methods from global-scale seismology, large-scale nonlinear optimization, and machine learning. Numerical examples for a synthetic phantom demonstrate the efficiency of the discretization, the effectiveness of the mini-batch approximation and the robustness of the trust-region method to reconstruct the acoustic properties of breast tissue with partial information.

@en