MLMarcel van Laaten

Marcel van Laaten

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Marcel van Laaten has a total of 14 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Exploring the Signal Content of Continuous Seismic Recordings at Station NKC (West-Bohemia/Vogtland) With a Self-Supervised Vision Transformer

    by Marcel van Laaten, Roman Esefelder

    Continuous seismic recordings contain a wide variety of signal types (earthquakes, quarry blasts, induced events, and diverse background and anthropogenic noise) but sorting them into meaningful categories usually relies on manual inspection. Here we take an exploratory approach and ask a deliberately simple question: what kinds of signals are actually present in the continuous record of station NKC, and can we let the data organise themselves without imposing labels beforehand? Station NKC lies in the West-Bohemia/Vogtland earthquake swarm region and records a mixture of earthquakes and non-earthquake transients. We select waveform windows with an STA/LTA trigger, represent each window as a continuous-wavelet-transform spectrogram of all three components, and use a self-supervised vision transformer (DINOv3) to group the windows by similarity. Because the model learns its representation directly from the unlabelled spectrograms, the resulting structure reflects the data rather than a predefined catalogue. We then examine the groups that emerge and compare them, where possible, against earthquake catalogues and against the waveforms themselves. The learned representation separates earthquakes clearly from the background, and unsupervised clustering recovers groups that are strongly enriched in earthquakes and that concentrate in the known swarm episodes. Equally informative is what the clustering reveals about the rest of the record. A large fraction of triggered windows have no catalogue counterpart, and inspection shows that many of these are themselves earthquake-like signals. We present the signal groups identified and discuss their likely physical interpretation.

  • Open Access

    Explainable Seismic Event Discrimination: Improved Explainability With Vision Transformers

    by Valentin Kasburg, Markus Zehner, Marcel van Laaten, Jozef Müller, Nina Kukowski

    The discrimination of seismic events traditionally relies on manual expert analysis or cross‐correlation techniques for pattern recognition. Although flexible algorithms from deep learning, such as convolutional neural networks, have demonstrated high discrimination accuracy for different seismic event origins based on spectral characteristics, their use in automated discrimination remains limited. Two main challenges hinder further deployment: first, these models are typically not transferable across regions with differing event characteristics; second, deep learning models often lack interpretability due to their black‐box nature. In this study, we investigate vision transformers as a novel architecture for seismic event discrimination and compare their accuracy and explainability to a convolutional neural network baseline. Vision transformers leverage attention mechanisms offering a distinct approach to visualizing the decision‐making process of deep learning models. We evaluate both model types on seismic data from two regions in Germany: the Vogtland‐West Bohemian region where earthquakes along the Leipzig‐Regensburg fault zone are discriminated from quarry blasts (QBs), and the Ruhr Area, where mining‐induced events are discriminated from QBs. Our results show that vision transformers achieve high discrimination accuracy on test data from both regions (97%–98%) performing on par with the baseline. For explainability, we apply Attention Rollout and the gradient‐based LeGrad method for vision transformers and SmoothGrad for the baseline. Combined with occlusion techniques, these methods demonstrate that vision transformers analyze spectrograms more comprehensively. Our findings highlight the potential of vision transformers as both a high‐performing and interpretable alternative to convolutional neural networks for automated seismic event discrimination.


  • Open Access

    Intrinsic and Scattering Attenuation of Shear Waves: Depth- and Frequency-Dependent Attenuation Insights using the QEST Code

    by Marcel van Laaten, Ulrich Wegler

    In local earthquake seismology depth-dependent elastic models of P- and S-wave velocities are indispensable, e. g. to locate earthquakes. If not only travel times but also amplitudes of seismic waves are important, elastic Earth models are insufficient and visco-elastic models are required to include intrinsic absorption of seismic waves. This applies e. g. to the estimation of moment magnitudes and to physics-based ground motion modeling in seismic hazard analysis. The estimation of seismic attenuation parameters of the Earth is significantly more difficult than the estimation of velocities for two reasons: (1) Scattering attenuation as well as intrinsic absorption contribute to the attenuation of seismic waves and it is essential to separate these two effects. (2) Seismic attenuation parameters are inherently frequency dependent, whereas the frequency dependence of seismic P- and S-wave velocities can be neglected in almost all cases. Due to the lack of information, attenuation is often completely neglected in seismic wave simulation, standard values for seismic Q are used, or frequency dependence and depth dependence are ignored. To solve this issue, we develop a computer code, 'QEST - Q estimation'. The code is based on a forward modeling using radiative transfer theory in depth-dependent velocity and attenuation models and a global inversion scheme based on a genetic algorithm. Besides frequency and depth dependent intrinsic as well as scattering attenuation parameters, earthquake source spectra and frequency dependent site amplifications are also a result of the inversion with QEST. We applied the technique to seismograms of earthquakes in three regions: the Upper Rhine Graben (Germany), the Leipzig-Regensburg fault zone (Germany) and the Alaska Subduction Zone (Alaska). These regions were selected to represent exemplary areas with thick sedimentary layers, without thick sedimentary layers and the lithosphere and asthenosphere, respectively. Results show a clear depth and frequency dependence of both, scattering attenuation as well as intrinsic absorption, within the thick sediments of the graben. In contrast, the Leipzig-Regensburg fault zone exhibits a clear frequency dependence of the attenuation parameters, albeit only a smaller depth dependence, while the results of the Alaska Subduction Zone show a depth and frequency dependence that is particularly evident in the scattering attenuation.

  • Open Access

    Automating Seismic Event Discrimination: A Comparative Study of Convolutional Neural Networks and Vision Transformers

    by Valentin Kasburg, Marcel van Laaten, Markus Zehner, Jozef Müller, Nina Kukowski

    To date, the discrimination of seismic events recorded in seismic networks is often performed manually by experts, classifying events into categories such as earthquakes, quarry blasts, or mining events. While recent studies have shown that deep learning algorithms, particularly Convolutional Neural Networks (CNNs), can efficiently and accurately distinguish between different types of seismic events, their application for automated seismic event discrimination remains limited. This limitation arises from several factors, including the absence of globally applicable models that maintain high precision for local seismic networks, the scarcity of data required for fine-tuning Deep Learning (DL) models, and the lack of interpretability in the decision-making processes of these black-box models.In this contribution, we explore the use of Vision Transformers (ViTs) as a novel approach for automating seismic event discrimination. To assess their potential for accuracy and explainability, we applied CNNs and ViTs to classify seismic events such as earthquakes, quarry blasts, and mining events. For this purpose, we pretrained the models on openly available seismic event data from Utah and Northern California and then fine-tuned and tested them on data from the Seismic Network of the Ruhr-University Bochum (RuhrNet) and the Thuringian Seismic Network (TSN).Our findings reveal that ViTs can analyze the entire spectrogram of a seismic event in a coherent manner, offering superior generalizability in pattern recognition compared to CNNs. In addition to achieving high discrimination accuracy, the attention weights of ViTs provide insights into the models’ decision-making process, offering a transparent and interpretable explanation of the underlying mechanisms driving its classifications.

  • Open Access

    Monitoring seismic velocity changes at Campi Flegrei (Naples) using seismic noise interferometry - Do we see precursors of the future volcanic activity?

    by Marcel van Laaten, Jozef Müller, Ulrich Wegler

    Campi Flegrei is a volcanic field located in the immediate vicinity of the densely populated area of Naples, Italy. Since 2005, the area has been experiencing a new bradyseismic crisis, i.e., a slow uplift of the subsurface caused by rising fluids in the subsurface. The uplift is accompanied by earthquake activity that has been steadily increasing for years, culminating in the strongest earthquake (ML 4.2) in the last 40 years on September 27, 2023. Such uplift and earthquakes can cause changes in seismic velocity and are often succeeded by a volcanic eruption. In this study, we utilize seismic noise to calculate velocity changes at different levels/frequencies over a 7-year period using passive image interferometry. The observed long-term velocity decrease of 1.39 % near the surface over the period from 2016 to 2023 can be explained by a volume increase of the hydrothermal system at the depth of 3 km. In 2023, the Campi Flegrei underwent several phases of velocity change. After a period of minor velocity changes, there was a gradual 0.7 % increase in velocity starting in May. Following the onset of the earthquake swarm in August, the velocity slowly decreased once again.

  • Open Access

    Nonlinear Inversion for a Multilayer Seismic S‐Wave Attenuation Model Using Radiative Transfer Theory

    by Marcel van Laaten, Ulrich Wegler

    We numerically solve the acoustic radiative transfer equation for seismic S‐waves via Monte Carlo simulation. By assuming a von Kármán‐type random medium with anisotropic scattering, we are able to simulate a realistic medium and determine its attenuation properties. In this study, we present an improved method, called QEST, to determine the frequency‐dependent intrinsic and scattering attenuation by nonlinear envelope inversion for a 1‐D multilayer model. Additionally, the spectral source energy of earthquakes and the energy site amplification of stations are determined. The code was applied to real data from the northern and southern Leipzig‐Regensburg fault zone (LRZ), Germany, as well as fluid‐induced earthquakes at the Insheim geothermal reservoir, Germany. The attenuation was analyzed in several frequency bands between 4.2 and 33.9 Hz and between 6.0 and 67.9 Hz, respectively. The inversion results reveal that the crystalline crustal subsurface along the LRZ shows little to no depth dependence, but there are differences in attenuation between the north and south. At Insheim, the near‐surface sedimentary basin exhibits significantly greater absorption and scattering than the crystalline basement. The inversion also shows that isotropic scattering can be an oversimplification and thus underestimate attenuation.


  • Open Access

    On the trail of fluids in the northernmost intracontinental earthquake swarm areas of the Leipzig-Regensburg fault zone, Germany

    by Marcel van Laaten, Ulrich Wegler, Tom Eulenfeld

    The Leipzig-Regensburg fault zone is documented as a band of seismic activity extending northwards from the earthquake swarm region NW-Bohemia/Vogtland at the Czech-German border area and is intersected by several Hercynian fault zones. Along the fault zone, there are several earthquake swarm areas, the northernmost of which are Schöneck and Werdau. In this study, we investigate the presumably fluid-induced earthquake swarm activity of the Schöneck and Werdau area. For this purpose, we apply two methods: local earthquake tomography and receiver functions to identify the structural composition of the crust, the areas affected by fluids and the origin of the fluids. We detected potential fluid paths characterised by high Vp/Vs ratios and granite intrusions nearby the swarms characterised by low Vp/Vs anomalies. Receiver function analysis yields the Moho at 25 to 33 km depth and two seismic discontinuities at 55 km and 68 km depth.


  • Coseismic drop of seismic velocity caused by the 2023 Turkey–Syria earthquakes

    by J Müller, M van Laaten, T Eulenfeld, U Wegler

    SUMMARY: The Mw 7.8 earthquake in Turkey on 6 February 2023 was extraordinary for various reasons. It originated in depth of only 10 km, ruptured along a fault plane around 300 km long and the surface was covered by an extensive network of high-quality seismic instruments. The strong motions resulted in a vast number of tragic casualties and huge material losses in Turkey and Syria. However, abundant and proximate seismic observations of this event and numerous aftershocks give an opportunity to deepen the understanding of earthquake processes. In this study, we carried out an assessment of coseismic changes of seismic velocity using passive image interferometry. We used data from one strong-motion and twenty-four broad-band sensors. We observed coseismic drops of seismic velocity, which reached up to −1.79 per cent at a location directly at the ruptured East Anatolian Fault Zone. Along the Mw 7.8 earthquake fault, we observe frequency dependence of the velocity changes. At frequencies above 0.5 Hz, the velocity drops seem to be higher at locations close to the ruptured faults than in the more distant areas.


  • Comparison of Multiple Lapse Time Window Analysis and Qopen to determine intrinsic and scattering attenuation

    by M van Laaten, T Eulenfeld, U Wegler

    This study compares the results of Multiple Lapse Time Windows Analysis (MLTWA) and full envelope inversion (Qopen) to determine intrinsic and scattering attenuation of the crust using the region around the central part of the Leipzig–Regensburg fault zone in Germany as an example. We use 18 of the region’s strongest earthquakes from 2008 to 2019 with a magnitude between 1.4 and 3.0 in the frequency band range between 3 and 34 Hz. The determined attenuation values of both methods are similar within their error bars. The inverse quality factors of the shear wave are relatively low compared to other regions, with values of 3.2 × 10−4 to 8.7 × 10−4 for |$Q^{-1}_{i}$| and 1.4 × 10−4 to 2.8 × 10−4 for |$Q^{-1}_{sc}$|⁠, respectively. As a by-product of Qopen, we also obtain the energy site amplification of the stations used in the inversion as well as source displacement spectra and moment magnitudes of the inverted earthquakes. Several combinations of inversion parameters were tested for MLTWA, with |$Q^{-1}_{i}$| and |$Q^{-1}_{sc}$| providing the lowest trade-off. Likewise, we investigated the influence of window length on the results of Qopen. We found a dependency of the results on the length, if the windows are shorter than 30 s. For longer time windows, the dependence disappears, and the result becomes independent of window length..