Seismic Noise Research Articles

B3AM: A beamforming toolbox for three-component ambient seismic noise analysis

We introduce the MATLAB toolbox B3AM for beamforming of three-component ambient noise array data. We explain the theory behind three-component beamforming and polarisation analysis in particular, provide an overview of the workflow, and discuss the output using a worked example. The strength of the presented code package is the analysis of multiple beam response maps from multiple time windows. Hence, it provides statistical information about the ambient noise wavefield recorded over a period of time, such as the ratio of surface to body waves, average dispersion velocities, or dominant propagation direction. It can be used to validate assumptions made about the ambient noise wavefield in a particular location, helping to interpret results from other techniques, such as the analysis of horizontal-to-vertical spectral ratios or ambient noise interferometry, and enabling more precise monitoring of specific wavefield components. While designed initially with seismic networks in mind, B3AM is applicable over a wide range of frequencies and array sizes and can thus be adapted also for laboratory settings or civil engineering applications.

3-D intrinsic attenuation tomography using ambient seismic noise applied to La Palma Island (Canary Islands)

The potential of the island of La Palma (Canary Islands) to host geothermal resources is very high, mainly due to its high volcanic activity. The primary goal of this study is to get a tridimensional image of the seismic intrinsic attenuation using ambient seismic noise and to identify anomalies that may be linked to active geothermal reservoirs on La Palma island. For this purpose, we developed a new Ambient Noise Attenuation Tomography (ANAT) technique, which uses seismic ambient noise for imaging intrinsic attenuation in 3-D at a local scale down to 5 km depth. Our research identifies two areas with high attenuation in the island’s southern region. One area could be associated with hydrothermal alteration zones beneath the Cumbre Vieja volcanic complex. Another high-attenuation zone was observed in the island’s southern part, which could be associated with extensively fractured rocks that might facilitate the circulation of heated fluids. Furthermore, we discuss the geothermal relevance of such anomalies, making a comparison with previous resistivity, S-wave velocity and density models. This study confirms that intrinsic attenuation retrieved from the coda of Rayleigh waves is more sensitive to the presence of fluids than velocity. Fluids being a key component of active geothermal reservoirs, it is reasonable to expect intrinsic attenuation anomalies in these systems. Therefore, we conclude that ANAT can be useful for geothermal exploration.

Quadratic Unet for seismic random noise attenuation

In the field of seismic data processing, seismic denoising is essential to improve the quality of seismic data. Various deep learning methods have been proposed, showing promising performance for seismic denoising. However, most deep learning methods are based on lin ear neurons, resulting in limited expressive ability for complex seismic signals. To enhance denoising capability using non-linear neurons, we propose a supervised quadratic U-shaped network (QUnet) for seismic random noise attenuation. The quadratic neurons in QUnet are represented by a second-order polynomial of input data and weighted parameters. Nu merical synthetic seismic data experiments show that the proposed QUnet method achieves higher signal-to-noise ratio results and preserves the continuity of reflectors more effectively. We have also tested the effectiveness of QUnet on both prestack field data and post-stack seismic data. The detailed structures and weak signals of seismic data are better preserved by our proposed QUnet method.

Multichannel Analysis of Surface Waves based on Common Virtual Source Gathers of Seismic Ambient Noise Cross-Correlations: A Case Study at an Earth Dam in Brazil

The S-wave velocity (Vs) is a valuable parameter for assessing the mechanical properties of subsurface materials for geotechnical purposes. Seismic surface wave methods have become prominent for estimating near-surface Vs models. Researchers have proposed methods based on passive seismic signals as efficient alternatives to enhance depth of investigation, lateral resolution and reduce field effort. This study presents the Multichannel Analysis of Surface Waves (MASW) utilizing Common Virtual Source Gathers (CVSGs) derived from seismic ambient noise cross-correlations, based on Ambient Noise Seismic Interferometry concepts. The method is applied to passive data acquired with an array of receivers at the Paranoá earth dam in Brasília, Brazil, to construct a pseudo-2D Vs image of the massif for interpretation. Our findings showcase the adopted processing flow and combination of methods as an effective approach for near-surface Vs estimation, demonstrating its usability also for large earth dam embankments.

Seismic random noise suppression via mining multi-scale local and global information

Suppressing random noise is critical for revealing real subsurface structures. Convolutional neural networks (CNNs), the leading seismic data denoising methods, excel at extracting local features but struggle to capture global representations. Unet can extract and reuse multi-scale features, aiding in the precise detection of details and semantic information; however, being based on convolutional operations, it struggles to capture global information. To capture global representations, researchers normally employ Transformers in high-level visual tasks, owing to their self-attention mechanisms. This paper introduces a method for mining multi-scale local and global information based on hybrid-gated Unet (HGUnet), which integrates Transformer, CNN, and Unet architectures to enhance the feature representation capability for seismic random noise suppression tasks. HGUnet comprises hybrid-gated blocks (HGB) embedded within a U-shaped architecture, employing a concurrent structure of Octave convolution and lightweight multi-head self-attention mechanism to efficiently extract multi-scale local and global features simultaneously. Moreover, at the conclusion of the HGB, to precisely leverage information and reduce computing costs, a gated feedforward network is designed to retain valuable information and prune redundancies for feature fusion. Synthetic and field experimental results demonstrate that HGUnet improves denoising quality over traditional and CNN methods without adding significant computing costs.

A Self-Supervised One-Shot Learning Approach for Seismic Noise Reduction

A Self-Supervised One-Shot Learning Approach for Seismic Noise Reduction

Seismic site conditions of RESNOM network

The Northwest Seismic Network of Mexico (RESNOM) is operated by personnel from the Center for Scientific Research and Higher Education of Ensenada, Baja California (CICESE), which supervises station installation, improvement, and maintenance. We employed seismic noise and the Horizontal to Vertical Spectral Ratio (HVSR) method to determine, for each station, the following site condition parameters: the depth of the rock layer (Heng_bed), and the geotechnical parameter VS30, obtained from 1D shear wave velocity models. Other parameters as the fundamental frequency (f0) and the average amplitude at the fundamental frequency (A0) were also estimated. Our results show clear differences between the values obtained for the Mexicali Valley and the Peninsular ranges regions. The VS30 obtained for stations of the Mexicali Valley region falls in the range from 173 m/s to 535 m/s, while for the Peninsular Ranges region is between 213 m/s and 958 m/s. Regarding the Heng_bed parameter, the values are similar between both regions, from 23 m to 850 m for the Peninsular and from 42 m to 926 m for the Mexicali Valley. Additionally, from the VS30 values, we propose the site classification according to the U.S. National Earthquake Hazards Reduction Program (NEHRP).

Rayleigh wave ellipticity from ambient noise: A practical method for monitoring seismic velocity variations in the near-surface

This study explores the feasibility and limitations of using Rayleigh wave (Rg) ellipticity for noise-based seismic monitoring at near-surface depths (4–70 m). We use the degree of polarization (DOP) method to extract the Rayleigh wave ellipticity from seismic noise recordings, employing normalized cross-correlation and cross-covariance coefficients to quantify ellipticity variations over time. Synthetic models and field data from three distinct case studies—Garner Valley, California; Riotinto mine, Spain; and the 2011 submarine volcanic eruption on El Hierro Island, Canary Islands—validate our approach. In these field applications, our method effectively tracks the seasonality of the shallow groundwater levels in Garner Valley, monitors pore pressure variations at the tailings dam of Riotinto mine, and detects volcanic induced changes on El Hierro, demonstrating robust performance even with variable noise sources. Our results indicate that Rayleigh wave ellipticity is a versatile tool for subsurface monitoring, capable of detecting velocity changes across a broad depth range. Our methodology represents a new independent and non-interferometric approach that enhances the detection of subsurface changes while improving resolution and exploration depth in seismic monitoring techniques.

Sources of seismic noise in an open-pit mining environment

We present the characterization of the sources of vibration in the open-pit Riotinto mine (southern Spain), based on the data recorded by a dense seismic network of 30 stations located along a 1-km long segment of a tailings dam. We describe the most common transient signals detected, including local and distant earthquakes, blasting and vehicles. The time variations in the amplitude of the 10–40 Hz frequency band are then used to define three phases of activity during the recording period. The phase with the highest seismic amplitudes is observed during a time interval of five weeks, coinciding with the civil works carried out for the regrowth of the dam, necessary to correctly capture the continuously increasing amount of tailings. In the last three months of operation of our network, the seismic noise is dominated by the deposition of tailings into the pond, enabling the use of seismic data to monitor into detail the evolution of the deposition process.

Fast and full characterization of large earthquakes from prompt elastogravity signals

Prompt ElastoGravity Signals are light-speed gravity-induced signals recorded before the arrival of seismic waves. They have raised interest for early warning applications but their weak amplitudes close to the background seismic noise have questioned their actual potential for operational use. A deep-learning model has recently demonstrated its ability to mitigate this noise limitation and to provide in near real-time the earthquake magnitude (Mw). However, this approach was efficient only for large earthquakes (Mw ≥ 8.3) of known focal mechanism. Here we show unprecedented performance in full earthquake characterization using the dense broadband seismic network deployed in Alaska and Western Canada. Our deep-learning model provides accurate magnitude and focal mechanism estimates of Mw ≥ 7.8 earthquakes, 2 minutes after origin time (hence the tsunamigenic potential). Our results represent a major step towards the routine use of prompt elastogravity signals in operational warning systems, and demonstrate its potential for tsunami warning in densely-instrumented areas.

Monitoring seismic velocity changes at Campi Flegrei (Italy) using seismic noise interferometry

Campi Flegrei is a volcanic field located west of Naples (Italy) in a densely populated area. Since 2005, its ground has been rising steadily due to the accumulation of fluids at shallow depths. The inflation of volcanic edifices is a possible precursor of an impending eruption. The uplift is accompanied by increasing seismic activity. This raises concerns about the possibility that the volcano may be on the verge of an eruption. To track the fluid movement, it is possible to monitor subtle changes of velocities of seismic waves by exploring ambient seismic noise. By examining different frequency bands, we can observe velocity changes at different depths. We interpret these changes as a monitoring of depth-dependent deformation in addition to the standard monitoring of surface deformation. We observe a velocity decrease in the long-term trend, presumably due to the extension of the hydrothermal system at shallow depths. To explain the long-term changes, we model a spherical pressure source to simulate volumetric strain changes induced by recent fluid activity. The model explains both, surface and subsurface deformation which leads to the opening of microcracks and pores, resulting in the observed velocity decrease. The short-term velocity changes are mainly driven by temperature or groundwater level changes. Once velocity changes are corrected for seasonal effects, remaining short term velocity changes can be associated with volcanic activity and earthquake swarms.

Finite element analysis for the measurement error of electrostatic accelerometer due to the electrode misalignment

Abstract The electrostatic accelerometer is the key payload of many space missions, such as satellite gravity measurements, space gravitational experiments, with a typical precision requirement from nano-g to femto-g. The measurement error model analysis is an important issue for the electrostatic accelerometer with numerous error sources. The traditional method by theoretical analysis is complicated to evaluate the complete measurement error models quantitatively especially when the machining and assembly errors of the sensor head are considered. Limited by the earth gravity and seismic noise which is much higher than the intrinsic noise, the complete error model of electrostatic accelerometers can also hardly be tested on ground. In this paper, the method by using finite element analysis of the sensor head is studied. The simulation model for an electrostatic accelerometer sensor head is built, and the multi-conductor capacitance matrix data is simulated. The accuracy of the capacitance simulation is evaluated in three ways including the convergence check of the capacitance with the mesh size, the symmetry verification, and the sensitivity comparison with the theoretical model. Finally, the accelerometer measurement model is analyzed based on the simulation data, using the case of rotating installation misalignment of the upper electrode as an example. The measurement model and error items of one horizontal axis and one rotating axis are quantitatively evaluated. The method proposed in this paper provide a new effective approach to the measurement model analysis of the electrostatic accelerometers in complex working scenarios both in orbit and on ground, which could be helpful to enhance the performance and the efficiency of application.

Insights into the crustal and the magmatic feeding structure at the Payunia Volcanic Province highlighted by geophysical methods, in the retroarc of the Southern Central Andes

The Payunia Volcanic Province is a Quaternary volcanic plateau emplaced in the retroarc area in the northern Neuquén Mesozoic Basin, associated with a hydrocarbon system. At deeper levels, this basin is linked to different intrusive systems that developed in the retroarc region at different times, during the Jurassic, Cretaceous, Eocene, Miocene, and even the Quaternary which influenced the hydrocarbon system maturity. We analyzed the Moho structure through this retroarc region, as well as the crustal structure affected by different stages of regional extension. From continuous seismic noise data, we calculated the autocorrelograms to obtain the reflection response below each seismological station. This allowed imaging the surface of primary crustal reflectors and in a few stations the top of an asthenospheric anomaly (SWAP) found by magnetotelluric survey and in concordance with satellite magnetic data. The crustal reflectors were identified in all stations at a mean two-way travel time of about ∼8.5 s and ∼12.5 s using frequency bands of about 1.0–2.4 Hz. Therefore, this is the first geophysical research that estimates the depth of the magmatic system, hosted at the top of the lower crust and the Moho discontinuity. The deepest reflector, only recognized in 4 stations, was observed with a two-way travel time of 17.2 s to 19.6 s. We used a mean one-dimensional Vp model to obtain the corresponding reflector depths which constrain the two-dimensional forward gravity model that fits with the observed regional anomaly for the region. We finally established a relationship between the shallowest sublithospheric electrical conductivity anomalies determined in previous researches and the strong deep reflections observed in some of the seismological stations. This information may help to constrain geochemical and petrological models and re-evaluate the hydrocarbon system maturity of the northern Neuquén basin.

Matrix imaging as a tool for high-resolution monitoring of deep volcanic plumbing systems with seismic noise

Volcanic eruptions necessitate precise monitoring of magma pressure and inflation for improved forecasting. Understanding deep magma storage is crucial for hazard assessment, yet imaging these systems is challenging due to complex heterogeneities that disrupt standard seismic migration techniques. Here we map the magmatic and hydrothermal system of the La Soufrière volcano in Guadeloupe by analyzing seismic noise data from a sparse geophone array under a matrix formalism. Seismic noise interferometry provides a reflection matrix containing the signature of echoes from deep heterogeneities. Using wave correlations resistant to disorder, matrix imaging successfully unscrambles wave distortions, revealing La Soufrière’s internal structure down to 10 km with 100 m resolution. This method surpasses the diffraction limit imposed by geophone array aperture, providing crucial data for modeling and high-resolution monitoring. We see matrix imaging as a revolutionary tool for understanding volcanic systems and enhancing observatories’ abilities to monitor dynamics and forecast eruptions.

Research on the Application of CEEMD and CEEMDAN in Seismic Random Noise Suppression

Empirical Mode Decomposition (EMD) is an adaptive method for processing nonlinear and non-stationary data. Complementary Ensemble Empirical Mode Decomposition (CEEMD) and Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) are noise-assisted signal processing methods developed from EMD. During seismic data acquisition, various types of noise, including random noise, are inevitably captured, and current methods for dealing with such noise are limited. In this study, these two signal processing methods were applied in simulated experiments and real seismic data to suppress random noise. By analyzing the processing effects and runtime, it was found that CEEMDAN offers better processing performance and speed, making it a valuable tool for practical applications.

Seismic shear wave noise suppression and application to well tie

Abstract Seismic shear waves have been employed in oil and gas exploration for decades. A 2D seismic shear wave inline was acquired in Sanhu area, located in the Qaidam Basin in western China. Although the acquired shear wave data exhibits high resolution and comparable bandwidth to the compressional wave, it is contaminated by various types of noise, including linear noise, single-frequency noise, and especially internal multiples. Internal multiples seriously compromise the primary reflections at both near-offset and far-offset and are difficult to be suppressed. This paper presents a case study of noise attenuation for the seismic shear wave. Firstly, single-frequency noise and linear noise are attenuated through filtering methods. Then, two methods (the Radon transform and the frequency-wavenumber (F-K) filtering) are evaluated for their effectiveness in multiple suppression and amplitude preservation. The results indicate that both methods successfully reduce long-period multiples at far-offset, enhancing the signal-to-noise ratio. We show that F-K filtering retains the characteristic ‘strong-weak-strong’ amplitude variation in the SV-SV wave data gather, making it preferable for subsequent AVO (amplitude variation with offset) analysis and inversion. Finally, a wave-equation-based MSI (multiple suppression inversion) method is used to suppress near-offset internal multiples. This involves iteratively predicting internal multiples and adaptively subtracting them from the original data. Stacked sections of different offsets are compared to demonstrate de-multiple result, and the result is also validated by the improvement of well calibration with the seismic shear wave data.

Forecasting ocean wave-induced seismic noise

Ocean waves induce the power peak in the seismic ground motion seen everywhere in the world between 0.03 and ~ 1 Hz, defining the seismic noise baseline. The precise generation mechanisms are well understood, and the dependence of seismic noise on sea weather has been precisely quantified using long-term time series. However, this knowledge has never been exploited to forecast the seismic noise background. Here we report the prediction of the seismic noise spectrum around 1 Hz at the Low-Noise Underground Laboratory (LSBB) in Rustrel, for up to 16 days in advance, limited by the time span of sea weather forecasts. We first characterize the dependence of the seismic noise at the LSBB on the Mediterranean Sea and Atlantic Ocean weather, using buoy data for 2020–2021. We exploit significant correlation in the 0.15Hz<f<2.5Hz\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$0.15~\ extrm{Hz}< f < 2.5~\ extrm{Hz}$$\\end{document} frequency band to make predictions, converting sea weather forecasts into seismic noise forecasts. The expected seismic background noise can be used to optimize the performance and running costs of scientific and industrial activities, by scheduling them during quiet intervals or adopting adaptive data analysis techniques to identify target signals in the predicted noise.

Wave velocities and anisotropy of rocks: Implication for origin of low velocity zone of the Qinling Orogenic Belt, China

Structure and composition of Earth is of fundamental importance in exploring the dynamic evolution of the crust and mantle. The Qinling Orogenic Belt (QOB) is located between the North China plate and the South China Plate, and is one of the main orogenic belts in China. To explore the composition and origin of anisotropy and the low wave velocity zone of the QOB, ten rock samples (gneiss and schist) were collected from the five sites of the QOB and the P- and S-wave velocities of these samples were measured under 0.6 to 2.0 GPa and 100 to 550 °C. The wave velocities increase with increasing pressure and decreasing temperature. The VP and VS of the schist and gneiss match the velocity of the middle and lower crust of the QOB, indicating that schist and gneiss are important component of the QOB. All the schist and gneiss samples exhibit obvious seismic anisotropy with 1.64%–17.42% for VS and 2.93%–14.78% for VP under conditions of crust and upper mantle. The CPO/LPO and layering distribution of mica in rock samples are the main reasons for this anisotropy. The VS structures below the five sampled sites from seismic ambient noise tomography were built to explore the effect of schist and gneiss on the composition and structure of the QOB. The results indicate that orientation-arranged gneiss and schist driven by the tectonic stresses might be a new origin of the character of VP/VS, seismic anisotropy, and the low velocity zone in the QOB.

A new automated procedure to obtain reliable moment tensor solutions of small to moderate earthquakes (3.0 ≤ M ≤ 5.5) in the Bayesian framework

SUMMARY The complete catalogue of moment tensor (MT) solutions is essential for a wide range of research in solid earth science. However, the number of reliable MT solutions for small to moderate earthquakes (3.0 ≤ M ≤ 5.5) is limited due to uncertainties arising from data and theoretical errors. In this study, we develop a new procedure to enhance the resolvability of MT solutions and provide more reliable uncertainty estimates for these smaller to moderate earthquakes. This procedure is fully automatic and efficiently accounts for both data and theoretical errors through two sets of hybrid linear–non-linear Bayesian inversions. In the inversion process, the covariance matrix is estimated using an empirical approach: the data covariance matrix is derived from the pre-event noise and the theoretical covariance matrix is derived from the residuals of the initial solution. We conducted tests using synthetic data generated from the 3-D velocity model and interference from background seismic noise. The tests found that using a combination of the non-Toeplitz data covariance matrix and the Toeplitz theoretical covariance matrix improves the solution and its uncertainties. Test results also suggest that including a theoretical covariance matrix when analysing MT in complex tectonic regions is essential, even if we have the best 1D velocity model. The application to earthquakes in the northern region of the Banda Arc resulted in the first published Regional Moment Tensor (RMT) catalogue, containing more than three times the number of trusted solutions compared to the Global Centroid Moment Tensor (GCMT) and the Indonesian Agency for Meteorology Climatology and Geophysics Moment Tensor (BMKG-MT) catalogue. The comparison shows that the trusted solutions align well with the focal mechanism of the GCMT and BMKG-MT, as well as with the maximum horizontal stress of the World Stress Map, and tectonic conditions in the study area. The newly obtained focal mechanisms provide several key findings: (i) they confirm that the deformation in the northern and eastern parts of Seram Island is influenced by oblique intraplate convergence rather than by the subduction process; (ii) they validate the newly identified Amahai Fault with a greater number of focal mechanisms and (iii) they reveal an earthquake Mw 4.7 with the same location and source mechanism 6 yr before the 2019 Ambon-Kairatu earthquake (Mw 6.5) which occurred on a previously unidentified fault.

Site-Specific Spectra for the City of Mexicali, Mexico, Obtained from April 2010 Earthquake Records

The April 2010 earthquake (Mw = 7.2), which occurred about 40 km to the southeast of the city of Mexicali, Mexico, caused significant damage to buildings. To improve knowledge of the seismic response of the soil due to the occurrence of earthquakes, a response spectrum at 5% damping was calculated. A comparison between the spectral ordinates obtained in this study and the spectra proposed by the regulations of the Federal Electricity Commission (CFE for its acronym in Spanish) in its seismic design for civil works manual, which is currently used as the design standard throughout the country, was made. We calculated response spectra using records from the April 2010 earthquake and a stratigraphic profile of the city to calculate a transfer function. We first corrected the records for site effect due to stations being over sedimentary soil, and then used them as Green functions to perform a numerical simulation of propagation through the stratigraphic profile to obtain a simulated surface record from which response spectra were calculated. Additionally, ambient seismic noise was measured at the same site to get the dominant period (To). We observed that the transfer function was similar to the spectral quotient up to 5 Hz and that To calculated in both ways gave similar values. The comparison suggests that the design spectrum of the CFE regulation can be considered as a representative spectrum for Mexicali for periods greater than 1.3 s, but not for the zone of short periods.