High-resolution seismic tomography of Long Beach, CA using machine learning

Seismic bulletins, with trustworthy phase picks, origin times, and source locations are key for regional seismic studies, such as travel-time (TT) tomography, attenuation tomography, and anisotropy studies. To lay the groundwork for such studies in Israel, we revised the seismic bulletin of Israel and the surrounding area and obtained a trustworthy TT data set. From the earthquake and explosion bulletins of the Geophysical Institute of Israel, we compiled a starting data set of about 123,000 earthquakes and explosions that occurred during the past 40 yr. After screening out the poorly recorded events, we were left with a data set of ∼38,000 well-recorded events. We then revised the remaining data set in two consecutive steps. In the first, we reviewed and updated station metadata, including changes in station metadata parameters over time. In the second step, we jointly relocated a list of selected seismic events, using the Bayesian hierarchical location software package (BayesLoc) of Myers et al. (2007) that performs joint relocation of multiple events. We observed striking dissimilarities between the spatial distributions of the newly relocated catalog and the initial locations. Although the depth distribution of the starting catalog is trimodal with peaks at 0, 5, and 10 km, the distribution in this study is unimodal, with a broad peak between 7.5 and 12.5 km. By differencing the observed arrival times and the origin times obtained through relocation with BayesLoc, we obtained a revised TT database that consists of 261,336 Pg, 132,876 Pn, 114,816 Sg, and 60,394 Sn arrivals, from a set of 30,458 jointly relocated seismic sources. We compared prerevision and postrevision TTs as a function of epicentral distance and concluded that the revised data set contains far fewer outliers and inconsistencies than the original data set. The revised TT data set may be used for seismic studies, such as TT tomography, attenuation tomography, and anisotropy studies.

Velocity anisotropy caused by fine layering has a strong effect on the travel paths at large vertical offsets in crosshole seismic data. This effect must be considered in seismic waveform tomography, so that the inversion procedure can match the waveforms in real data. In this paper, we propose to properly take into account this anisotropic effect in two stages. First, we invert for the anisotropy parameter simultaneously with the (horizontal) velocity in travel time tomography. Then we incorporate this estimated anisotropy parameter model in the waveform simulation during waveform tomography, to refine the velocity image. When considering this anisotropic effect, far-offset data are properly used in the inversion, then both travel time and waveform tomography may have a better ray coverage, producing a velocity image with a higher resolution.

Detecting and imaging dense non-aqueous phase liquids (DNAPLs) in the subsurface is a challenging problem that is of interest to the Department of Energy, Department of Defense, and several local and state agencies. Current DNAPL detection techniques, such as Direct Push Probe Technologies (DPT) and In-Situ Tracers (IST) have risks and limitations. Cross-well radar (CWR) is a radar-based geophysical technique with low invasiveness for real time monitoring of DNAPLs. This technique uses electromagnetic waves transmitted and received through antennas in the subsurface. The computerized tomography is applied to a pilot-scale experimental facility constructed by the authors (referred as SoilBED) in this research. Cross-tomography data at multiple depths and locations were collected to simulate a 1/100 scaled contaminated soil problem. The experimental results are converted to the time domain signals. The resulting signal travel times are compared with the simulated results by FDTD. The transformed signals are used to prepare signal intensity and travel time tomograms of the SoilBED cross-sectional slices, for background and scattered field to study the signature of the scatterers. The results are able to visualize the dielectric objects in the saturated soil. Introduction Cross-well radar is a potential technique for DNAPL detection. Most of the research and field works done in Cross-well radar and other radar based geophysical techniques deal with a single frequency. Selecting one single frequency may be effective for one non-dispersive medium, but perhaps not for other non-dispersive or most dispersive media. This research covers a broadband of frequency from 400 MHz to 2.2 GHz. Having broadband frequency data helps to either use the entire frequency range and work in frequency domain or select a single frequency or narrower frequency range that seems to have enough detection potential out of the entire observed frequency response. It is important to consider cost and practicality of the system. Establishing practical and cost effective systems require strong theoretical and laboratory background conditions. Therefore, both theoretical and experimental aspects of the research are studied in parallel, to evaluate the feasibility of DNAPL detection using CWR and to study wave propagation behavior and forward problem. Collecting repeatable and reproducible frequency response data is a very challenging process. This difficultly achieved data can be manipulated first or directly used towards image processing. As any research subject, conducting both experimental and theoretical approaches is the most desirable way to advance fundamental knowledge in the field of interest. Frequency response and arrival time can be converted to each other. Thus, frequency domain results can be transformed to time domain and used towards techniques such as travel time tomography and vice versa. The experimental frequency domain results can be converted to CTF (Channel Transfer Functions) and compared to the simulated frequency domain results by the first order Born approximation (Farid 2004, Farid et al. 2005, and Zhan 2005). The FDTD simulation code (Farid 2002 and 2003) is available as the theoretical tool to perform the comparison in time domain and study travel time. If the experimental results are converted to time domain and transformed to travel time, they can be compared to the FDTD simulated results. Travel Time Computed by FDTD To establish a better knowledge of wave propagation in soil, both dry and saturated backgrounds are studied. The sandy soil in the SoilBED facility with 3.88% moisture content is referred to as dry soil.

A fast marching algorithm for the factored eikonal equation

<p>The Ordos block is located on the west side of the North China Carton, adjacent to the northeastern part of the Tibetan Plateau. Affected by two tectonic movements, Ordos block internal structure remains relatively stable structure, but surrounded by active tectonic belts. With the development of the second and third part of the “China Seismological Science Array”, the distribution of seismic observation stations in Ordos region has been greatly improved. This study will use the new seismic observation data of "Array III", and combined with the phase observation data of "Array II" to form a more complete seismic phase travel time data set. The regional seismic body-wave travel time tomography will figure out a more reliable three-dimensional velocity structure of P waves in Ordos.</p><p>Our study area spans from 32°N to 42°N and 108°E to 114°E , which includes the Ordos block and its adjacent structures . The seismic data we used for inversion were recorded by 1244 stations including: 198 permanent stations and 1043 temporary stations (ChinArray II and III), from November 2013 to August 2017. After manual labeled the seismic phase, we select events with more than ten phase records of individual seismic events. The epicentral distance is less than 200km. Finally, we obtained about 22,500 phase records of 1882 local seismic events.</p><p>The preliminary results are consistent with previous studies and surface structures of a wide range of velocity distributions. However, in the middle-upper crust under the Liupan Mountain west, the low-speed anomaly extending downward is shown, which may be caused by the shallow crustal damage caused with the continuous eastward compression of asthenosphere in the northeastern margin of the Qinghai-Tibet Plateau during the Cenozoic. It is worth noting that there is an EW-trending low-velocity zone under the Dingbian-Suide fault beneath the Ordos Basin, with a depth form lower crust to 50 km in upper mantel. This low-velocity anomaly divides the high-speed disturbance in the Ordos block into two parts，indicate the depth of the fault can reach the upper mantel. In the Taihang Mountains in the west of the study area, low-velocity anomalies extending to the upper layer of the mantle are shown. We initially believe that this anomaly is related to the volcanic thermal motion that once existed on the area.</p>

Travel time is a basic measure based on which intelligent transportation systems such as traveler information systems, traffic management systems, public transportation systems are developed. Although many methodologies have been proposed, they have not yet adequately solved many challenges associated with travel time, in particular, travel time estimation for all links in a large and dynamic urban traffic network is still an open problem that needs addressing. Typically focus is placed on major roads such as motorways and main city arteries but there is an increasing need to know accurate travel times for minor urban roads. Such information is crucial for tackling air quality problems, accommodate the growing number of cars and provide accurate information for routing. This study aims to address the aforementioned challenges by introducing a methodology, namely Similar Model Searching (SMS), to estimate travel times by using historical sparse travel time data. The SMS learns the temporal and spatial relationship between the travel time of adjacent links and utilize labeled data of similar models in order to improve its overall performance. The effectiveness of the proposed method is evaluated on a section of Leicestershire traffic network in the UK. The obtained results show that SMS efficiently estimates travel time of target links using models of adjacent traffic links.

Data is one of the most important limiting factors of deep machine learning (ML) model in drilling applications. Though a big size of historical data can be available, high-quality cleaned and labeled data is usually limited. In this case study, we show that with limited labeled data, physics-based data augmentation combined with unsupervised learning significantly improves both stability and accuracy in bit wear ML model. It provides a pathway to overcome labeled data shortage and field data quality limitations. Labeled bit wear data is usually limited because only the final bit dull state can be labeled using dull photos for the entire drilling distance. To overcome this difficulty, an encoder-decoder unsupervised ML framework based on Bi-directional LSTM architecture is first applied to the data of entire drilling distance to extract and maximize data learnings. Then, a physics-based wear estimator is implemented in the learned latent space to augment labeled wear dataset, guided by true wear labels and offset-well data. Weighting factor in loss function is applied to optimize augmented and true datasets in final supervised learning step. The proposed real-time bit wear model is built on a combination of physics-based data augmentation and unsupervised data learning method. The model is applied on multiple field bit runs. Results show that by implementing the unsupervised data learnings only, the prediction accuracy is improved by 30% compared to the baseline ML model. By combining physics-based augmentation, the accuracy is further improved by 10%. More importantly, adding physics-based data augmentation significantly reduces prediction variance and unphysical wear outputs, therefore improving prediction stability by more than 30%. It should be noted that the prediction stability of AI/ML model is crucial in real-time application and decision making. The results show physics-based data augmentation not only increases the size of label dataset and prevents model overfitting, but also applies physics-based guidance to the ML model effectively. It is learned weighting factor plays a crucial role in balancing loss contributions from true wear labels and physics-based labels. While low weighting factor of physics-based labels diminishes the augmented data, high weighting factor disrespects the true wear labels leading to a high prediction bias. Overall, the bit wear model study shows physic-based data augmentation combined with unsupervised data learning can effectively improve model accuracy, stability, and overcome labeled data shortage difficulty. The proposed paper shows a case study for the bit wear ML model using a combination of physics-based data augmentation and unsupervised data learning. While labeled data is one of the major challenges of ML models in many drilling applications, this study provides a pathway to improve both accuracy and stability of deep ML models with limited labeled data.

Trunk rehabilitation exercises such as those for remediating core stability can help improve the seated balance of patients with weakness or loss of proprioception caused by diseases such as stroke, and aid the recovery of other functions such as gait. However, there has not yet been any reported method for automatically determining the parameters that define exercise difficulty on a trunk rehabilitation robot (TRR) based on data such as the patient's demographic information, balancing ability, and training sequence, etc. We have proposed a machine learning (ML)-based difficulty adjustment method to determine an appropriate virtual damping gain (Dvirtual) of the controller for the TRR's unstable training mode. Training data for the proposed system is obtained from 37 healthy young adults, and the trained ML model thus obtained is tested through experiments with a separate population of 25 healthy young adults. The leave-one-out cross validation results (37 subjects) from the training group for validation of the designed ML model showed 80.90% average accuracy (R2 score) for using the given information to predict the desired difficulty levels, which are represented by the level of balance performance quantified as Mean Velocity Displacement (MVD) of the center of pressure. Statistical analysis (Repeated measures analysis of variance) of subject performance also showed that ground truth difficulty levels from the training data and predicted difficulty levels did not differ significantly under any of the three exercise modes used in this study (Hard, Medium, and Easy), and the standard deviations were reduced by 16.39, 41.39, and 25.68%, respectively. Moreover, the Planar Deviation (PD) of the center of pressure, which was not the target parameter here, also showed results similar to the MVD, which indicates that the predicted Dvirtual affected the difficulty level of balance performance. Therefore, the proposed ML model-based difficulty adjustment method has potential for use with people who have varied balancing abilities.

Correction of the travel time estimation for ambulances of the red cross Tijuana using machine learning

Accurate real-time bus arrival information is essential for an efficient public transport network, as it significantly impacts passenger experience, system reliability, reduced waiting times, dwell times, and operational efficiency. In Sri Lanka's public transport system, current bus arrival time computations primarily rely on static data, neglecting real-time information and critical factors influencing travel times. This highlights the need to identify the unique variables affecting bus arrival times within the Sri Lankan context and to develop robust prediction models that account for these influences. While traditional methods such as Historical Average Models, Regression, Time Series Analysis, and Kalman Filtering have been used in previous research for short-term travel time predictions, Machine Learning (ML) approaches have proven to deliver superior accuracy. ML models are regarded as the most effective for heterogeneous, lane-less traffic conditions with varying traffic volumes, such as those found in Sri Lanka. ML techniques excel in processing large, high-quality datasets and provide accurate predictions by accounting for all relevant variables influencing travel times. Although research has been conducted on developing various basic ML models for travel time prediction, there is a noticeable gap in studies comparing these models to determine the most suitable one for the Sri Lankan context. A Long- Short Term Memory (LSTM) neural network is a deep learning model that is capable of handling long-term dependencies. In the context of bus travel time prediction, LSTMs can leverage historical traffic and travel data to capture temporal patterns and fluctuations that influence travel times. By evaluating LSTMs against basic machine learning models, this study seeks to explore the advantages of applying deep learning techniques to transportation forecasting, ultimately contributing to more accurate and efficient predictive systems in transit planning. The ML models selected in this study include two basic traditional models K- Nearest Neighbours (KNN) and Support Vector Regression (SVR) and four advanced models that utilize ensemble techniques and advanced optimization as Random Forest Regression (RFR), Ada Boost, XG Boost and Gradient Boosting Machine (GBM). The performance of these models was compared with the LSTM model to identify the gap in their accuracies.

Traditional alloy design typically relies on trial and error and experience. Machine learning can significantly accelerate the discovery and design process of new materials. However, as the number of elements in the alloy and target performance metrics increase, alloy optimization becomes more challenging. To address this, this paper proposes a machine learningbased multi-objective optimization method for magnesium alloy fracturing balls. The machine learning model trained on the magnesium alloy corrosion and ultimate compressive strength database achieves an accuracy of 0.98 on the training set and 0.93 on the test set. By using a multi-objective genetic algorithm to optimize the element ratios of the magnesium alloy, Mg-6.4Al-3.4Zn-4.6Cu was obtained, with a corrosion rate of 538 mm/year and an ultimate compressive strength of 369 MPa. This provides a new method for the efficient, rapid, and precise preparation of novel degradable magnesium alloys.

Analyses of groundwater level in a data-scarce region based on assessed precipitation products and machine learning

In seismic refraction exploration, travel time tomography is the most widely used method in engineering and environmental geophysical exploration. In this paper, we mainly optimize the travel time tomography of refraction. First, with respect to the forward algorithm, we introduce a new travel time calculation method to improve the accuracy and efficiency of forward calculation. Based on the fast marching method (FMM), we introduce an improved forward calculation method called the multi-stencil fast marching method (MSFM). In the process of inversion, we propose a dynamic prior model composite constraint (DPMCC) method based on the T0 difference method from the idea of multi-scale inversion. Meanwhile, we use the prior information to improve the accuracy of inversion. Furthermore, we use the dynamic regularization factor selection method to make the inversion solution more stable and reliable. Finally, we test and analyze the synthetic data and the measured data to verify the effectiveness of the optimized travel time tomography algorithm.

A number of different geodynamic models have been proposed to explain the extension that occurred during the Miocene in the Alboran Sea region of the western Mediterranean despite the continued convergence and shortening of northern Africa and southern Iberia. In an effort to provide additional geophysical constraints on these models, we performed a local, regional, and teleseismic tomographic travel time inversion for the lithospheric and upper mantle velocity structure and earthquake locations beneath the Alboran region in an area of 800×800 km2. We picked P and S arrival times from digital and analog seismograms recorded by 96 seismic stations in Morocco and Spain between 1989 and 1996 and combined them with arrivals carefully selected from local and global catalogs (1964–1998) to generate a starting data set containing over 100,000 arrival times. Our results indicate that a N‐S line of intermediate‐depth earthquakes extending from crustal depths significantly inland from the southern Iberian coast to depths of over 100 km beneath the center of the Alboran Sea coincides with a W to E transition from high to low velocities imaged in the uppermost mantle. A high‐velocity body, striking approximately NE‐SW, is imaged to dip southeastwards from lithospheric depths beneath the low‐velocity region to depths of ∼350 km. Between 350 and 500 km the imaged velocity anomalies become more diffuse. However, pronounced high‐velocity anomalies are again imaged at 600 km near an isolated cluster of deep earthquakes. In addition to standard tomographic methods of error assessment, the effects of systematic and random errors were assessed using block shifting and bootstrap resampling techniques, respectively. We interpret the upper mantle high‐velocity anomalies as regions of colder mantle that originate from lithospheric depths. These observations, when combined with results from other studies, suggest that delamination of a continental lithosphere played an important role in the Neogene and Quaternary evolution of the region.

Using waveforms and travel times from deep earthquakes, we constructed 16 seismic profiles, each of which constrains the radial variation in Vp over a small area beneath the northern Philippine Sea. Taken together, the azimuthal coverage of these profiles also places tight bounds on the lateral extent of a region of anomalously high Vp (up to 3% faster than average Earth models) originally suggested by travel time tomography. Unlike travel time tomography, which relies heavily on arrival times of the direct P phase, we utilize the waveforms and move‐out of later arrivals that mainly sample the mantle transition zone of interest. Our results identify three important characteristics of the northern Philippine Sea anomaly that are distinct from previous results. First, being approximately a subhorizontal, laterally uniform feature, the anomaly is localized beneath the northwestern corner of the Philippine Sea, within a region of approximately 500×500 km2 immediately east of the Ryukyu arc. Second, the anomaly is well constrained to occur in the lower portion of the transition zone, extending all the way down to the 660‐km discontinuity. Third, the presence of such a distinct anomaly reduces the contrast in Vp across the 660‐km discontinuity from approximately 6% to 3%. Such a configuration is consistent with the interpretation that the anomaly is caused by a remnant of subducted slab, as negative buoyancy should rest the slab just above the 660‐km discontinuity where resistance to subduction is expected from a negative Clapeyron slope during the spinel—Mg‐Fe‐perovskite transition.