Continuous Seismic Waveforms Research Articles

Using Moment Tensor Inversions for Rapid Seismic Source Detection and Characterization: Application to the North Korean Nuclear Tests

AbstractThe rapid detection and source characterization of any type of seismic events including earthquakes and nuclear explosions is one of the missions of many seismological laboratories. Most often, the techniques used are based on phase picking and amplitude measurements for detecting and locating events, and for estimating magnitudes. From these parameters, especially for anthropogenic sources, event screening is then done empirically, and this may lead to misinterpretations of the source nature. However, it is known that seismic waveform inversion for the determination of the moment tensor has proven to be a reliable source physics-based method for event characterization. Here, we present a technique already used for earthquake monitoring in tectonically active regions, and we test it on very shallow natural and anthropogenic events recorded in the vicinity of the Democratic People’s Republic of Korea (DPRK). From a grid of potential locations and by scanning continuous seismic waveforms, it is possible to implement a rapid detector of seismic events providing the full information of the sources (origin time, location, magnitude, mechanism, and source decomposition). We show its overall performances on all past DPRK nuclear tests and regional earthquakes. From such an approach fast complete event screening is achieved. Source uncertainties can also be estimated. This stand-alone detector and identifier of seismic events may help monitoring seismological agencies to provide a rapid and complete alert for any events within a region of interest before more in-depth discrimination analysis can be run.

A High-Resolution Aftershock Catalog for the 2014 Ms 6.5 Ludian (China) Earthquake Using Deep Learning Methods

A high-resolution catalog for the 2014 Ms 6.5 Ludian aftershocks was constructed based on the deep learning phase-picking model (CERP) and seismic-phase association technology (PALM). A specific training strategy, which combines the advantages of the conventional short–long window average energy ratio algorithm (STA/LTA) and AI algorithms, is employed to retrain the CERP model. The P- and S-wave phases were accurately detected and picked on continuous seismic waveforms by the retained AI model. Hypoinverse and HypoDD were utilized for the precise location of 3286 events. Compared to the previous results, our new catalog exhibits superior performances in terms of location accuracy and the number of aftershock events, thereby enabling a more detailed depiction of the deep-seated tectonic features. According to the distribution of aftershocks, it can be inferred that (1) the seismogenic fault of the Ludian earthquake is the NW-trending Baogunao–Xiaohe Fault, (2) the Ludian aftershocks interconnected with the discontinuous NW-trending Baogunao–Xiaohe Fault, and they also intersected with the Zhaotong–Ludian Fault. (3) This suggests that the NE-trending Zhaotong–Ludian Fault may have been intersected by the NW-trending Baogunao–Xiaohe Fault, indicating that the Baogunao–Xiaohe Fault is likely a relatively young Neogene fault.

Mapping Glacier Basal Sliding Applying Machine Learning

AbstractDuring the RESOLVE project (“High‐resolution imaging in subsurface geophysics: development of a multi‐instrument platform for interdisciplinary research”), continuous surface displacement and seismic array observations were obtained on Glacier d’Argentière in the French Alps for 35 days in May 2018. The data set is used to perform a detailed study of targeted processes within the highly dynamic cryospheric environment. In particular, the physical processes controlling glacial basal motion are poorly understood and remain challenging to observe directly. Especially in the Alpine region for temperate based glaciers where the ice rapidly responds to changing climatic conditions and thus, processes are strongly intermittent in time and heterogeneous in space. Spatially dense seismic and Global Positioning System (GPS) measurements are analyzed applying machine learning to gain insight into the processes controlling glacial motions of Glacier d’Argentière. Using multiple bandpass‐filtered copies of the continuous seismic waveforms, we compute energy‐based features, develop a matched field beamforming catalog and include meteorological observations. Features describing the data are analyzed with a gradient boosting decision tree model to directly estimate the GPS displacements from the seismic noise. We posit that features of the seismic noise provide direct access to the dominant parameters that drive displacement on the highly variable and unsteady surface of the glacier. The machine learning model infers daily fluctuations and longer term trends. The results show on‐ice displacement rates are strongly modulated by activity at the base of the glacier. The techniques presented provide a new approach to study glacial basal sliding and discover its full complexity.

Rayleigh phase velocity and azimuthal anisotropy from ambient noise data in the Sanjiang lateral collision zone in the SE margin of the Tibetan plateau

The Sanjiang lateral collision zone in the SE margin of the Tibetan Plateau is located at the east edge of the junction of the Eurasian and Indian plates. Using the continuous seismic waveforms recorded by 146 temporary and 21 permanent seismic stations in the study area, we obtain Rayleigh wave phase velocity and azimuthal anisotropy for periods 2 s to 40 s from the surface wave direct tomography method. This direct tomography method can obtain finer high-resolution results than the traditional surface wave tomography. Our results show that the low-velocity anomalies are found beneath the Lijiang-Xiaojinhe fault (LXF), Red River fault (RRF), Chuxiong fault and Tengchong volcanoes, the high-velocity anomalies are in the region of Weixi and Panzhihua at periods 5 ~ 8 s. The fast velocity directions mainly align N-S. At periods 10 ~ 15 s, the distributions of low-velocity anomalies are consistent with the strikes of LXF and RRF. At periods 20 ~ 35 s, the high- and low-velocity anomalies are bounded by the RRF, which may imply the fault is divided by the thick crust (indicated by low-velocity anomalies) and the thin crust with shallow mantle (indicated by high-velocity anomalies). The fast velocity directions at the periods 10 ~ 35 s rotate clockwise from north to south of the study area. The intensity of anisotropy in the low-velocity zone is stronger than that in the high-velocity zone, and the intensity in the north of the study area is stronger than that in the south. Results indicate the source of anisotropy may be different in each subzone.

Research on Self-Noise Characteristics of Nine Types of Seismometers Obtained by PDF Representation Using Continuous Seismic Data from the Malingshan Seismic Station, China.

The self-noise level of a seismometer can determine the performance of the seismic instrument and limit the ability to use seismic data to solve geoscience problems. Accurately measuring and simultaneously comparing the self-noise models from different types of seismometers has long been a challenging task due to the constraints of observation conditions. In this paper, the self-noise power spectral density (PSD) values of nine types of seismometers are calculated using four months of continuous seismic waveforms from Malingshan seismic station, China, and nine self-noise models are obtained based on the probability density function (PDF) representation. For the seismometer STS-2.5, the self-noise levels on the horizontal channels (E−W and N−S) are significantly higher than that on the vertical channel (U−D) in the microseism band (0.1 Hz to 1 Hz), which is a computing bias caused by the misalignment between the sensors in the horizontal direction, while the remarkably elevated noise on the horizontal channels at the low frequencies (<0.6 Hz) may originate from the local variation of atmospheric pressure. As for the very broadband seismometers Trillium-Horizon-120 and Trillium-120PA, and the ultra-broadband seismometers Trillium-Horizon-360 and CMG-3T-360, there is a trade-off between the microseism band range and low-frequency range in the PSD curves of the vertical channel. When the level of self-noise in the microseism band is high, the self-noise at low frequencies is relatively low. Although compared with the other very broadband seismometers, the self-noise level of the vertical component of the STS-2.5 is 3 dB to 4 dB lower at frequencies less than 1 Hz, the self-noise level of the STS-2.5 at high frequencies (>2 Hz) is slightly higher than others. From our observations, we conclude that the nine seismometers cannot reach the lowest noise level in all frequency bands within the working range.

Monitoring microseismicity of the Hengill Geothermal Field in Iceland

Induced seismicity is one of the main factors that reduces societal acceptance of deep geothermal energy exploitation activities, and felt earthquakes are the main reason for closure of geothermal projects. Implementing innovative tools for real-time monitoring and forecasting of induced seismicity was one of the aims of the recently completed COSEISMIQ project. Within this project, a temporary seismic network was deployed in the Hengill geothermal region in Iceland, the location of the nation’s two largest geothermal power plants. In this paper, we release raw continuous seismic waveforms and seismicity catalogues collected and prepared during this project. This dataset is particularly valuable since a very dense network was deployed in a seismically active region where thousand of earthquakes occur every year. For this reason, the collected dataset can be used across a broad range of research topics in seismology ranging from the development and testing of new data analysis methods to induced seismicity and seismotectonics studies.

DeepQuake — An application of CNN for seismo-acoustic event classification in The Netherlands

Recent developments in instrumentation and methodology are major driving forces in the advances of solid Earth sciences. The deployment of large and dense sensor networks enables data centres to acquire data of increased volume and quality. The analysis of such data provides scientists with a better understanding about natural phenomena in the subsurface. At the same time new challenges arise to exploit the growing information potential. Innovative methods based on Artificial Intelligence offer concrete opportunities to tackle those challenges. In this paper we investigate Convolutional Neural Networks (CNN) for seismo-acoustic event classification in The Netherlands. A supervised deep learning approach is applied to continuous seismic waveforms in order to automatically detect seismo-acoustic events and discriminate their sources. Our region of interest is characterised by high seismic noise conditions and intense anthropogenic activity which deeply influence the recorded signals. Site conditions and our aim for a tool that could support operational earthquake analysis motivated us to devise a novel specialised CNN architecture.Building on existing approaches, we designed, trained and evaluated two CNN models: arch-time and arch-spect. Our results suggest that as CNN inputs, spectrograms are more suitable than continuous waveforms. Therefore, we propose a model (arch-spect) that achieves a performance among the highest in modern source detection systems.Furthermore, we focus on explainability aspects and propose a method to gain insights in the behaviour of a ‘black-box’ deep learning classification system. By generating visualisations of activations we aim to empower domain experts with a useable cross-checking tool and pave the way for a broader operational uptake of Artificial Intelligence based methods.

Imaging the Ice Sheet and Uppermost Crustal Structures with a Dense Linear Seismic Array in the Larsemann Hills, Prydz Bay, East Antarctica

Abstract Comprehensive geophysical surveys including magnetotelluric, seismic, and aerial gravity–magnetic surveys are essential for understanding the history of Antarctic tectonics. The ice sheet and uppermost structure derived from those geophysical methods are relatively low resolution. Although ice-penetrating radar can provide high-resolution reflectivity images of the ice sheet, it cannot provide constraints on subice physical properties, which are important for geological understanding of the Antarctic continent. To obtain high-resolution images of the ice sheet and uppermost crustal structure beneath the Larsemann Hills, Prydz Bay, East Antarctica, we conduct an ambient noise seismic experiment with 100 short-period seismometers spaced at 0.2 km intervals. Continuous seismic waveforms are recorded for one month at a 2 ms sampling rate. Empirical Green’s functions are extracted by cross correlating the seismic waveform of one station with that of another station, and dispersion curves are extracted using a new phase-shift method. A high-resolution shear-velocity model is derived by inverting the dispersion curves. Furthermore, body waves are enhanced using a set of processing techniques commonly used in seismic exploration. The stacked body-wave image clearly shows a geological structure similar to that revealed by the shear-wave velocity model. This study, which is the first of its kind in Antarctica, possibly reveals a near-vertical intrusive rock covered by an ice sheet with a horizontal extent of 4 km. Our results help to improve the understanding of the subice environment and geological evolution in the Larsemann Hills, Prydz Bay, East Antarctica.

Learning the Low Frequency Earthquake Activity on the Central San Andreas Fault

AbstractLow frequency earthquakes (LFEs) originating below the central San Andreas Fault are associated with slow‐slip beneath the seismogenic zone within the more ductile portion of the crust. Monitoring efforts over 15 years detected &gt;1 million LFEs. We train a gradient boosted tree model using statistical features describing the seismic waveforms to estimate the hourly LFE event count. The burst‐like LFE behavior is reproduced, while lower amplitudes are predicted during the most active periods. The hourly event counts are up to 18% greater than the catalog. The ability to continuously monitor LFE activity provides insight to when geodetic measurements of slow slip are possible, without the need for developing a computational‐intensive template‐matching catalog. Similar waveform statistical features are found between detecting LFEs and tremors, which provides additional evidence tremors are composed of LFEs. The approach extracts information contained in continuous seismic waveforms that might benefit detecting precursory signals.

Large air pressure changes triggered by P-SV ground motion in a cave in northern Taiwan

Acoustic-gravity waves are generally considered to be one of the major factors that drive changes of the total electron content in the ionosphere. However, causal mechanisms of couplings between sources in the lithosphere and responses in the atmosphere and the ionosphere are not fully understood, yet. A barometer in the cave of the SBCB station records an unusual phenomenon of larger amplitudes in air pressure changes inside than those at the Xinwu station (outside). Accordingly, the comparison between the recorded data at the SBCB and Xinwu station can drive investigations of potential sources of the unusual phenomenon. Analytical results of phase angle differences reveal that the air pressure outside the cave at the Xinwu station often leads air pressure changes inside at the SBCB station at relatively low frequency bands. In contrast, the larger pressure changes at frequencies > ~ 5 × 10–4 Hz inside the cave at the SBCB station lead smaller changes outside at the Xinwu station. To expose causal mechanisms of the unusual phenomenon, continuous seismic waveforms are further conducted for examination. When the horizontal and vertical ground velocities of ground vibrations yield a difference in the phase angle close to 90°, coherence values between the air pressure changes and ground vibrations become large. This suggests that the pressure-shear vertical ground vibrations can drive air pressure changes. Meanwhile, the results shed light on investigating the existence of acoustic waves near the Earth’s surface using a partially confined space underground due to that the assumptions of the waves can propagate upward into the atmosphere driving changes in the ionosphere.

Real‐Time Earthquake Early Warning With Deep Learning: Application to the 2016 M 6.0 Central Apennines, Italy Earthquake

AbstractEarthquake early warning (EEW) systems are required to report earthquake locations and magnitudes as quickly as possible before the damaging S wave arrival to mitigate seismic hazards. Deep learning techniques provide potential for extracting earthquake source information from full seismic waveforms instead of seismic phase picks. We developed a novel deep learning EEW system that utilizes fully convolutional networks to simultaneously detect earthquakes and estimate their source parameters from continuous seismic waveform streams. The system determines earthquake location and magnitude as soon as very few stations receive earthquake signals and evolutionarily improves the solutions by receiving continuous data. We apply the system to the 2016 M 6.0 Central Apennines, Italy Earthquake and its first‐week aftershocks. Earthquake locations and magnitudes can be reliably determined as early as 4 s after the earliest P phase, with mean error ranges of 8.5–4.7 km and 0.33–0.27, respectively.

Spatiotemporal changes of seismicity rate during earthquakes

Abstract. Scientists demystify stress changes within tens of days before a mainshock and often utilize its foreshocks as an indicator. Typically, foreshocks are detected near fault zones, which may be due to the distribution of seismometers. This study investigates changes in seismicity far from mainshocks by examining tens of thousands of M≥2 quakes that were monitored by dense seismic arrays for more than 10 years in Taiwan and Japan. The quakes occurred within epicentral distances ranging from 0 to 400 km during a period of 60 d before and after the mainshocks that are utilized to exhibit common behaviors of seismicity in the spatiotemporal domain. The superimposition results show that wide areas exhibit increased seismicity associated with mainshocks occurring more than several times to areas of the fault rupture. The seismicity increase initially concentrates in the fault zones and gradually expands outward to over 50 km away from the epicenters approximately 40 d before the mainshocks. The seismicity increases more rapidly around the fault zones approximately 20 d before the mainshocks. The stressed crust triggers ground vibrations at frequencies varying from ∼5×10-4 to ∼10-3 Hz (i.e., variable frequency) along with earthquake-related stress that migrates from exterior areas to approach the fault zones. The variable frequency is determined by the observation of continuous seismic waveforms through the superimposition processes and is further supported by the resonant frequency model. These results suggest that the variable frequency of ground vibrations is a function of areas with increased seismicity leading to earthquakes.

An Automated Method for Developing a Catalog of Small Earthquakes Using Data of a Dense Seismic Array and Nearby Stations

AbstractWe propose a new automated procedure for using continuous seismic waveforms recorded by a dense array and its nearby regional stations for P-wave arrival identification, location, and magnitude estimation of small earthquakes. The method is illustrated with a one-day waveform dataset recorded by a dense array with 99 sensors near Anza, California, and 24 surrounding regional stations within 50 km of the dense array. We search a wide range of epicentral locations and apparent horizontal slowness values (0–15 s/km) in the 15–25 Hz range and time shift the dense array waveforms accordingly. For each location–slowness combination, the average neighboring station waveform similarity (avgCC) of station pairs &amp;lt;150 m apart is calculated for each nonoverlapping 0.5 s time window. Applying the local maximum detection algorithm gives 966 detections. Each detection has a best-fitting location–slowness combination with the largest avgCC. Of 331 detections with slowness &amp;lt;0.4 s/km, 324 (about six times the catalog events and 98% accuracy) are found to be earthquake P-wave arrivals. By associating the dense array P-wave arrivals and the P- and S-wave arrivals from the surrounding stations using a 1D velocity model, 197 detections (∼4 times of the catalog events) have well-estimated locations and magnitudes. Combining the small spacing of the array and the large aperture of the regional stations, the method achieves automated earthquake detection and location with high sensitivity in time and high resolution in space. Because no preknowledge of seismic-waveform features or local velocity model is required for the dense array, this automated algorithm can be robustly implemented in other locations.

Characteristics of Ground Motion Generated by Wind Interaction With Trees, Structures, and Other Surface Obstacles

AbstractAnalysis of continuous seismic waveforms from a temporary deployment at Sage Brush Flats on the San Jacinto fault reveals earthquake‐ and tremor‐like signals generated by the interaction of wind with obstacles above the surface. Tremor‐like waveforms are present at the site during wind velocities above 2 m/s, which occur for 70% of the deployment duration. The response to the wind has significant spatial variability with highest ground motions near large surface objects. The wind‐related signals show ground velocities that exceed the average ground motions of M1.0–1.5 earthquakes for 6–31% of the day. Waveform spectra indicate a modulation of amplitude that correlates with wind velocity and distance from local structures. Earthquake‐like signals are found to originate from local structures and vegetation, and are modified on length scales of tens of meters. Transient signals originating beyond the study area are also observed with amplitudes greater than some microseismic events. The wind‐related ground motions contribute to local high‐frequency seismic noise. Some of these signals may be associated with small failures of the subsurface material. During elevated wind conditions a borehole seismometer at a depth of 148 m shows increased energy in the 1–8‐Hz band that is commonly used for earthquake and tremor detection. The wind‐related earthquake‐ and tremor‐like signals should be accounted for in earthquake detection algorithms due to the similar features in both time and frequency domains. Proper recognition of wind‐related ground motions can contribute to understanding the composition of continuous seismic waveforms and characterize mechanical properties of the shallow crust.

Detection of random noise and anatomy of continuous seismic waveforms in dense array data near Anza California

SUMMARY We develop a methodology to separate continuous seismic waveforms into random noise (RN), not random noise (NRN) produced by earthquakes, wind, traffic and other sources of ground motions, and an undetermined mixture of signals. The analysis is applied to continuous records from a dense seismic array on the San Jacinto fault zone. To detect RN signals, we cut hourly waveforms into non-overlapping 1 s time windows and apply cross-correlations to separate RN candidates from outliers. The cross-correlation coefficients between different RN candidates fall into a tight range (i.e. 0.09–0.35), while cross-correlation coefficients of RN candidates with NRN signals (e.g. seismic or air-traffic events) are lower. The amplitude spectra of RN candidates have a well-defined level, while the amplitude spectra of other signals deviate from that level. Using these properties, we examine the amplitude spectra of moving time windows and cross-correlation coefficients with RN templates in each hour. The hourly RN is quasi-stationary and the results cluster tightly in the parameter space of cross-correlation coefficients and L2 norm deviations from the mean spectra of RN candidates. Time windows with parameters in this tight cluster are identified as RN, windows that deviate significantly from the RN cluster are identified as NRN and windows with values in between are identified as mixed signals. Several iterations on each hourly data are used to update and stabilize the selection of RN templates and mean noise spectra. For the days examined, the relative fractions of RN, NRN and mixed signals in local day (night) times are about 26 (42 per cent), 40 (33 per cent) and 34 per cent (25 per cent), respectively.

Analysis of surface and seismic sources in dense array data with match field processing and Markov chain Monte Carlo sampling

SUMMARY We introduce a methodology based on array processing to detect and locate weak seismic events in a complex fault zone environment. The method is illustrated using data recorded by a dense array of 1108 vertical component geophones in a 600 m × 600 m area on the Clark branch of the San Jacinto Fault. Because surface and atmospheric sources affect weak ground motion, it is necessary to discriminate them from weak seismic sources at depth. Source epicentral positions and associated apparent velocities are extracted from continuous seismic waveforms using Match Field Processing (MFP). We implement MFP at specific frequencies targeting surface and subsurface sources, using for computational efficiency a forward model of acoustic source in a homogenous medium and Markov Chain Monte Carlo sampling. Surface sources such as Betsy gun shots and a moving vehicle are successfully located. Weak seismic events are also detected outside of the array, and their backazimuth angle is retrieved and found to be consistent with the fault geometry. We also show that the homogeneous acoustic model does not yield satisfying results when extracting microseismic event depth, because of the ambiguity between depth and the apparent velocity based on surface data.

Characteristics of Airplanes and Helicopters Recorded by a Dense Seismic Array Near Anza California

AbstractWe observe frequent air traffic events in continuous seismic waveforms recorded for about 30 days by 1,108 vertical geophones in a tight array on the San Jacinto fault zone. The waveforms of the air traffic events resemble tremor or collections of small earthquakes. However, time‐frequency analysis reveals clear Doppler effects that can be modeled with basic equations and fitted well with parameters corresponding to airplanes and helicopters. The flying traces can be inverted by fitting the parameters at each station across the entire array. About 31 air traffic events are detected per day in the relatively remote study area, with peak activity from about 8:00 to 18:00 and significantly fewer events between 23:00 and 4:00 local time. The average event duration is about 200 s, so they cover together &gt;7% of the day. To estimate the total time covered by earthquakes, we derive a scaling relation logτ(M) = 0.41M + 0.89 between earthquake duration and magnitudes using data of 266 earthquakes in the magnitude range 0 ≤ M ≤ 3. The results indicate that in most places, the duration of air traffic events is likely to exceed considerably the total time covered by earthquakes.

Seismicity Structure of the Northern Chile Forearc From &gt;100,000 Double‐Difference Relocated Hypocenters

AbstractIn this study, we present high‐resolution seismicity images of the northern Chile subduction zone forearc. We used 8 years of continuous seismic waveform data from the Integrated Plate Boundary Observatory Chile network and auxiliary stations to produce an extensive earthquake catalog containing 101,601 double‐difference relocated earthquake hypocenters using automatic event detection and phase picking routines. The minimum magnitude of retrieved events is &lt;2, and the catalog is estimated to be complete at magnitudes above ∼2.8. Intraslab seismicity makes up the majority of detected earthquakes. Where the seismogenic zone of the megathrust is active, a clear separation of seismicity into three distinct planes can be observed. The uppermost plane corresponds to the plate interface, which is observed to terminate downdip at a depth of 50–55 km. The other two planes, located ∼7 and ∼26 km below the slab surface, dip at a constant angle of about 20° until they are absorbed by a 25 km thick highly active cluster of intermediate‐depth seismicity at depths of 80–120 km. Downdip of this cluster, the slab steepens and lower plate seismicity is considerably sparser, even absent in the northern part of the study area. Upper plate seismicity is also considerable, with a segment between 21 and 21.6°S standing out for featuring pervasive activity occurring all the way down to the plate interface. Here the seismicity resembles a wedge in west‐east profile view and occurs where the upper plate crust is coldest based on thermal models.

Rayleigh wave phase velocities of South China block and its adjacent areas

Using records of continuous seismic waveforms from 609 broadband seismic stations in the South China Block and its adjacent areas in 2010–2012, empirical Green’s functions of surface waves were obtained from cross-correlation functions of ambient noise data between these stations. High quality phase velocity dispersion curves of Rayleigh waves were obtained using time-frequency analysis. These interstation dispersion curves were then inverted to build Rayleigh wave phase velocity maps at periods of 6–50 s. The results of phase velocity maps indicate that phase velocities at 6–10 s periods are correlated with the geological features in the upper crust. Major basins and small-scale grabens and basins display slow velocity anomalies; while most of the orogenic belts and the fold belts display high velocity anomalies. With the gravity gradient zone along Taihang Mountain to Wuling Mountain as the boundary for the phase velocity maps at period of 20–30 s, the western area mainly displays low velocity anomalies, while the eastern side shows high velocity anomalies. Phase velocities in the eastern South China Block south to the Qinling-Dabie orogenic belt is higher than that in the eastern North China Block to the north, which is possibly due to the differences of tectonic mechanisms between the North China Craton and the South China Block. The phase velocities at periods of 40–50 s are possibly related to the lateral variations of the velocity structure in the lower crust and upper mantle: The low-velocity anomalies in the eastern part of the Tibetan Plateau are caused by the thick crust; while the Sichuan Basin and the southern part of the Ordos Basin display distinct high-velocity anomalies, reflecting the stable features of the lithosphere in these blocks. The lateral variation pattern of phase velocities in the southern part of the South China Block is not consistent with the surface trace of the block boundary in the eastern Yunnan Province and its vicinities. The phase velocities in the Sichuan Basin are overall slow at short periods and gradually increase with period from the central part to the edge of the basin, indicating the features of shallower basement in the center and overall stable lithospheric mantle of the basin. The middle and upper crust of the southern Ordos Basin in the North China Block is heterogeneous, while in lower crust and the uppermost mantle the phase velocities mainly exhibit high anomalies. High-velocity anomalies are widespread at the middle of the Qinling-Dabie orogenic belt, as well as the areas in southeastern Guangxi with Caledonian granite explosion, but its detailed mechanism is still unclear.

Automatic detection of alpine rockslides in continuous seismic data using hidden Markov models

AbstractData from continuously recording permanent seismic networks can contain information about rockslide occurrence and timing complementary to eyewitness observations and thus aid in construction of robust event catalogs. However, detecting infrequent rockslide signals within large volumes of continuous seismic waveform data remains challenging and often requires demanding manual intervention. We adapted an automatic classification method using hidden Markov models to detect rockslide signals in seismic data from two stations in central Switzerland. We first processed 21 known rockslides, with event volumes spanning 3 orders of magnitude and station event distances varying by 1 order of magnitude, which resulted in 13 and 19 successfully classified events at the two stations. Retraining the models to incorporate seismic noise from the day of the event improved the respective results to 16 and 19 successful classifications. The missed events generally had low signal‐to‐noise ratio and small to medium volumes. We then processed nearly 14 years of continuous seismic data from the same two stations to detect previously unknown events. After postprocessing, we classified 30 new events as rockslides, of which we could verify three through independent observation. In particular, the largest new event, with estimated volume of 500,000 m3, was not generally known within the Swiss landslide community, highlighting the importance of regional seismic data analysis even in densely populated mountainous regions. Our method can be easily implemented as part of existing earthquake monitoring systems, and with an average event detection rate of about two per month, manual verification would not significantly increase operational workload.