The knowledge of the local soil structure is important for the assessment of seismic hazards. A widespread, but time-consuming technique to retrieve the parameters of the local underground is the drilling of boreholes. Another way to obtain the shear wave velocity profile at a given location is the inversion of surface wave dispersion curves. To ensure a good resolution for both superficial and deeper layers, the used dispersion curves need to cover a wide frequency range. This wide frequency range can be obtained using several arrays of seismic sensors or a single array comprising a large number of sensors. Consequently, these measurements are time-consuming. A simpler alternative is provided by the use of the ellipticity of Rayleigh waves. The frequency dependence of the ellipticity is tightly linked to the shear wave velocity profile. Furthermore, it can be measured using a single seismic sensor. As soil structures obtained by scaling of a given model exhibit the same ellipticity curve, any inversion of the ellipticity curve alone will be ambiguous. Therefore, additional measurements which fix the absolute value of the shear wave velocity profile at some points have to be included in the inversion process. Small-scale spatial autocorrelation measurements or MASW measurements can provide the needed data. Using a theoretical soil structure, we show which parts of the ellipticity curve have to be included in the inversion process to get a reliable result and which parts can be omitted. Furthermore, the use of autocorrelation or high-frequency dispersion curves will be highlighted. The resulting guidelines for inversions including ellipticity data are then applied to real data measurements collected at 14 different sites during the European NERIES project. It is found that the results are in good agreement with dispersion curve measurements. Furthermore, the method can help in identifying the mode of Rayleigh waves in dispersion curve measurements.

Earthquake monitoring and many seismological studies depend on earthquake locations from phase arrival‐times. We present an extended, arrival‐time earthquake location procedure (NLL‐SSST‐coherence) which approaches the precision of differential‐timing based, relative location methods. NLL‐SSST‐coherence is based on the probabilistic, global‐search NonLinLoc (NLL) location algorithm which defines a probability density function (PDF) in 3D space for hypocenter location and is highly robust to outlier data. NLL‐SSST‐coherence location first reduces velocity model error through iteratively generated, smooth, source‐specific, station travel‐time corrections (SSST). Next, arrival‐time error is reduced by consolidating location information across events based on inter‐event waveform coherency. If the waveforms at a station for multiple events are very similar (have high coherency) up to a given frequency, then the distance separating these “multiplet” events is small relative to the seismic wavelength at that frequency. NLL‐coherence relocation for a target event is a stack over 3D space of the NLL‐SSST location PDF for the event and the PDF's for other multiplet events, each weighted by its waveform coherency with the target. NLL‐coherence relocation requires waveforms from only one or a few seismic stations, enabling precise relocation with sparse networks, for foreshocks and early aftershocks of significant events before installation of temporary stations, and for older data sets with few waveform data. We show the behavior and performance of NLL‐SSST‐coherence with synthetic and ground‐truth tests, and through application and comparison to relative locations for California earthquake sequences with dense and sparse station coverage.

We propose to implement a compact convolutional transformer (CCT) for picking the earthquake phase arrivals (EQCCT). The proposed method consists of two branches, with each of them responsible for picking the arrival times of the P- or S-wave phases. We use the STEAD dataset to train and validate the proposed EQCCT algorithm. We split the STEAD dataset into 85% for training, 5% for validation, and 10% for testing To facilitate the training process, we implement several data augmentation strategies to the training set by adding Gaussian noise, randomly shifting the waveforms, adding a second earthquake to the input window, and dropping one or two channels from the seismogram in the STEAD dataset. As a result, the EQCCT model outperforms both EQTransformer and PhaseNet, the two most popular deep-learning-based phase-picking methods. Considering the true positive criterion as the picked phases arriving within 0.5 s of the reference times, the EQCCT achieves the lowest mean absolute error (MAE) compared to the EQTransformer and PhaseNet methods for the STEAD, Japanese, Instance and Texas datasets. Our EQCCT network also demonstrates superior performance in other metrics such as precision, recall, and F1 score. We apply the pre-trained model to three independent datasets (not included in the training set), i.e., the Japanese, Texas, and Instance datasets, and achieve higher picking accuracy than the EQTransformer and the PhaseNet in terms of various statistical metrics, demonstrating a stronger robustness and generalization ability of the EQCCT. The real-time application of EQCCT in the Texas Seismological Network (TexNet) further demonstrates its production-ready performance in terms of detection and phase-picking accuracy.

The detection of earthquake signals is a fundamental yet challenging task in observational seismology. A robust automatic earthquake detection algorithm is strongly demanded in view of the ever‐growing global seismic dataset. Here, we develop an automatic earthquake detection framework based on a deep learning approach (SCALODEEP). It extracts high‐order features embedded in three‐component seismograms by encoding a time‐frequency representation of the data (scalogram) into a deep network with skip connections. The SCALODEEP is trained and validated on an open‐source dataset from North California, and then employed to seismicity detection in four areas, including Arkansas, Japan, Texas, and Egypt. Despite vastly varying characteristics of regional earthquakes (e.g., focal mechanism, duration, and noise level), SCALODEEP successfully detects seismic signals over a broad range of local magnitudes (as low as −1.3) and outperforms conventional algorithms such as STA/LTA, FAST, and template matching. Compared to recently proposed deep learning based frameworks (e.g., CRED and Earthquake transformer), SCALODEEP achieves a superior generalization ability via a sophisticated network architecture. In summary, our study offers a promising new tool to improve existing earthquake detection systems and, as importantly, sheds light on designing an effective deep learning network for generalized earthquake detection.