@article{EsfahaniVogelCottonetal.2021, author = {Esfahani, Reza Dokht Dolatabadi and Vogel, Kristin and Cotton, Fabrice and Ohrnberger, Matthias and Scherbaum, Frank and Kriegerowski, Marius}, title = {Exploring the dimensionality of ground-motion data by applying autoencoder techniques}, series = {Bulletin of the Seismological Society of America : BSSA}, volume = {111}, journal = {Bulletin of the Seismological Society of America : BSSA}, number = {3}, publisher = {Seismological Society of America}, address = {El Cerito, Calif.}, issn = {0037-1106}, doi = {10.1785/0120200285}, pages = {1563 -- 1576}, year = {2021}, abstract = {In this article, we address the question of how observed ground-motion data can most effectively be modeled for engineering seismological purposes. Toward this goal, we use a data-driven method, based on a deep-learning autoencoder with a variable number of nodes in the bottleneck layer, to determine how many parameters are needed to reconstruct synthetic and observed ground-motion data in terms of their median values and scatter. The reconstruction error as a function of the number of nodes in the bottleneck is used as an indicator of the underlying dimensionality of ground-motion data, that is, the minimum number of predictor variables needed in a ground-motion model. Two synthetic and one observed datasets are studied to prove the performance of the proposed method. We find that mapping ground-motion data to a 2D manifold primarily captures magnitude and distance information and is suited for an approximate data reconstruction. The data reconstruction improves with an increasing number of bottleneck nodes of up to three and four, but it saturates if more nodes are added to the bottleneck.}, language = {en} } @article{ZaliOhrnbergerScherbaumetal.2021, author = {Zali, Zahra and Ohrnberger, Matthias and Scherbaum, Frank and Cotton, Fabrice and Eibl, Eva P. S.}, title = {Volcanic tremor extraction and earthquake detection using music information retrieval algorithms}, series = {Seismological research letters}, volume = {92}, journal = {Seismological research letters}, number = {6}, publisher = {Seismological Society of America}, address = {Boulder, Colo.}, issn = {0895-0695}, doi = {10.1785/0220210016}, pages = {3668 -- 3681}, year = {2021}, abstract = {Volcanic tremor signals are usually observed before or during volcanic eruptions and must be monitored to evaluate the volcanic activity. A challenge in studying seismic signals of volcanic origin is the coexistence of transient signal swarms and long-lasting volcanic tremor signals. Separating transient events from volcanic tremors can, therefore, contrib-ute to improving upon our understanding of the underlying physical processes. Exploiting the idea of harmonic-percussive separation in musical signal processing, we develop a method to extract the harmonic volcanic tremor signals and to detect tran-sient events from seismic recordings. Based on the similarity properties of spectrogram frames in the time-frequency domain, we decompose the signal into two separate spec-trograms representing repeating (harmonic) and nonrepeating (transient) patterns, which correspond to volcanic tremor signals and earthquake signals, respectively. We reconstruct the harmonic tremor signal in the time domain from the complex spectrogram of the repeating pattern by only considering the phase components for the frequency range in which the tremor amplitude spectrum is significantly contribut-ing to the energy of the signal. The reconstructed signal is, therefore, clean tremor signal without transient events. Furthermore, we derive a characteristic function suitable for the detection of tran-sient events (e.g., earthquakes) by integrating amplitudes of the nonrepeating spectro-gram over frequency at each time frame. Considering transient events like earthquakes, 78\% of the events are detected for signal-to-noise ratio = 0.1 in our semisynthetic tests. In addition, we compared the number of detected earthquakes using our method for one month of continuous data recorded during the Holuhraun 2014-2015 eruption in Iceland with the bulletin presented in Agustsdottir et al. (2019). Our single station event detection algorithm identified 84\% of the bulletin events. Moreover, we detected a total of 12,619 events, which is more than twice the number of the bulletin events.}, language = {en} } @article{ZaliReinKruegeretal.2023, author = {Zali, Zahra and Rein, Teresa and Kr{\"u}ger, Frank and Ohrnberger, Matthias and Scherbaum, Frank}, title = {Ocean bottom seismometer (OBS) noise reduction from horizontal and vertical components using harmonic-percussive separation algorithms}, series = {Solid earth}, volume = {14}, journal = {Solid earth}, number = {2}, publisher = {Coepernicus Publ.}, address = {G{\"o}ttingen}, issn = {1869-9529}, doi = {10.5194/se-14-181-2023}, pages = {181 -- 195}, year = {2023}, abstract = {Records from ocean bottom seismometers (OBSs) are highly contaminated by noise, which is much stronger compared to data from most land stations, especially on the horizontal components. As a consequence, the high energy of the oceanic noise at frequencies below 1 Hz considerably complicates the analysis of the teleseismic earthquake signals recorded by OBSs. Previous studies suggested different approaches to remove low-frequency noises from OBS recordings but mainly focused on the vertical component. The records of horizontal components, which are crucial for the application of many methods in passive seismological analysis of body and surface waves, could not be much improved in the teleseismic frequency band. Here we introduce a noise reduction method, which is derived from the harmonic-percussive separation algorithms used in Zali et al. (2021), in order to separate long-lasting narrowband signals from broadband transients in the OBS signal. This leads to significant noise reduction of OBS records on both the vertical and horizontal components and increases the earthquake signal-to-noise ratio (SNR) without distortion of the broadband earthquake waveforms. This is demonstrated through tests with synthetic data. Both SNR and cross-correlation coefficients showed significant improvements for different realistic noise realizations. The application of denoised signals in surface wave analysis and receiver functions is discussed through tests with synthetic and real data.}, language = {en} } @misc{ZaliReinKruegeretal.2023, author = {Zali, Zahra and Rein, Teresa and Kr{\"u}ger, Frank and Ohrnberger, Matthias and Scherbaum, Frank}, title = {Ocean bottom seismometer (OBS) noise reduction from horizontal and vertical components using harmonic-percussive separation algorithms}, series = {Zweitver{\"o}ffentlichungen der Universit{\"a}t Potsdam : Mathematisch-Naturwissenschaftliche Reihe}, journal = {Zweitver{\"o}ffentlichungen der Universit{\"a}t Potsdam : Mathematisch-Naturwissenschaftliche Reihe}, number = {1320}, issn = {1866-8372}, doi = {10.25932/publishup-58882}, url = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus4-588828}, pages = {15}, year = {2023}, abstract = {Records from ocean bottom seismometers (OBSs) are highly contaminated by noise, which is much stronger compared to data from most land stations, especially on the horizontal components. As a consequence, the high energy of the oceanic noise at frequencies below 1 Hz considerably complicates the analysis of the teleseismic earthquake signals recorded by OBSs. Previous studies suggested different approaches to remove low-frequency noises from OBS recordings but mainly focused on the vertical component. The records of horizontal components, which are crucial for the application of many methods in passive seismological analysis of body and surface waves, could not be much improved in the teleseismic frequency band. Here we introduce a noise reduction method, which is derived from the harmonic-percussive separation algorithms used in Zali et al. (2021), in order to separate long-lasting narrowband signals from broadband transients in the OBS signal. This leads to significant noise reduction of OBS records on both the vertical and horizontal components and increases the earthquake signal-to-noise ratio (SNR) without distortion of the broadband earthquake waveforms. This is demonstrated through tests with synthetic data. Both SNR and cross-correlation coefficients showed significant improvements for different realistic noise realizations. The application of denoised signals in surface wave analysis and receiver functions is discussed through tests with synthetic and real data.}, language = {en} } @article{BoraCottonScherbaum2019, author = {Bora, Sanjay Singh and Cotton, Fabrice and Scherbaum, Frank}, title = {NGA-West2 Empirical Fourier and Duration Models to Generate Adjustable Response Spectra}, series = {Earthquake spectra : the professional journal of the Earthquake Engineering Research Institute}, volume = {35}, journal = {Earthquake spectra : the professional journal of the Earthquake Engineering Research Institute}, number = {1}, publisher = {Sage Publ.}, address = {Thousand Oaks}, issn = {8755-2930}, doi = {10.1193/110317EQS228M}, pages = {61 -- 93}, year = {2019}, abstract = {Adjustment of median ground motion prediction equations (GMPEs) from one region to another region is one of the major challenges within the current practice of seismic hazard analysis. In our approach of generating response spectra, we derive two separate empirical models for a) Fourier amplitude spectrum (FAS) and b) duration of ground motion. To calculate response spectra, the two models are combined within the random vibration theory (RVT) framework. The models are calibrated on recordings obtained from shallow crustal earthquakes in active tectonic regions. We use a subset of NGA-West2 database with M3.2-7.9 earthquakes at distances 0-300 km. The NGA-West2 database expanded over a wide magnitude and distance range facilitates a better constraint over derived models. A frequency-dependent duration model is derived to obtain adjustable response spectral ordinates. Excellent comparison of our approach with other NGA-West2 models implies that it can also be used as a stand-alone model.}, language = {en} } @book{ScherbaumMzhavanadzeArometal.2020, author = {Scherbaum, Frank and Mzhavanadze, Nana and Arom, Simha and Rosenzweig, Sebastian and M{\"u}ller, Meinard}, title = {Tonal Organization of the Erkomaishvili Dataset: Pitches, Scales, Melodies and Harmonies}, series = {Computational Analysis Of Traditional Georgian Vocal Music}, journal = {Computational Analysis Of Traditional Georgian Vocal Music}, number = {1}, editor = {Scherbaum, Frank}, publisher = {Universit{\"a}tsverlag Potsdam}, address = {Potsdam}, issn = {2702-2641}, doi = {10.25932/publishup-47614}, url = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus4-476141}, publisher = {Universit{\"a}t Potsdam}, pages = {64}, year = {2020}, abstract = {In this study we examine the tonal organization of a series of recordings of liturgical chants, sung in 1966 by the Georgian master singer Artem Erkomaishvili. This dataset is the oldest corpus of Georgian chants from which the time synchronous F0-trajectories for all three voices have been reliably determined (M{\"u}ller et al. 2017). It is therefore of outstanding importance for the understanding of the tuning principles of traditional Georgian vocal music. The aim of the present study is to use various computational methods to analyze what these recordings can contribute to the ongoing scientific dispute about traditional Georgian tuning systems. Starting point for the present analysis is the re-release of the original audio data together with estimated fundamental frequency (F0) trajectories for each of the three voices, beat annotations, and digital scores (Rosenzweig et al. 2020). We present synoptic models for the pitch and the harmonic interval distributions, which are the first of such models for which the complete Erkomaishvili dataset was used. We show that these distributions can be very compactly be expressed as Gaussian mixture models, anchored on discrete sets of pitch or interval values for the pitch and interval distributions, respectively. As part of our study we demonstrate that these pitch values, which we refer to as scale pitches, and which are determined as the mean values of the Gaussian mixture elements, define the scale degrees of the melodic sound scales which build the skeleton of Artem Erkomaishvili's intonation. The observation of consistent pitch bending of notes in melodic phrases, which appear in identical form in a group of chants, as well as the observation of harmonically driven intonation adjustments, which are clearly documented for all pure harmonic intervals, demonstrate that Artem Erkomaishvili intentionally deviates from the scale pitch skeleton quite freely. As a central result of our study, we proof that this melodic freedom is always constrained by the attracting influence of the scale pitches. Deviations of the F0-values of individual note events from the scale pitches at one instance of time are compensated for in the subsequent melodic steps. This suggests a deviation-compensation mechanism at the core of Artem Erkomaishvili's melody generation, which clearly honors the scales but still allows for a large degree of melodic flexibility. This model, which summarizes all partial aspects of our analysis, is consistent with the melodic scale models derived from the observed pitch distributions, as well as with the melodic and harmonic interval distributions. In addition to the tangible results of our work, we believe that our work has general implications for the determination of tuning models from audio data, in particular for non-tempered music.}, language = {en} } @article{DialloKuleshHolschneideretal.2006, author = {Diallo, Mamadou Sanou and Kulesh, Michail and Holschneider, Matthias and Kurennaya, Kristina and Scherbaum, Frank}, title = {Instantaneous polarization attributes based on an adaptive approximate covariance method}, series = {Geophysics}, volume = {71}, journal = {Geophysics}, number = {5}, publisher = {SEG}, address = {Tulsa}, issn = {0016-8033}, doi = {10.1190/1.2227522}, pages = {V99 -- V104}, year = {2006}, abstract = {We introduce a method for computing instantaneous-polarization attributes from multicomponent signals. This is an improvement on the standard covariance method (SCM) because it does not depend on the window size used to compute the standard covariance matrix. We overcome the window-size problem by deriving an approximate analytical formula for the cross-energy matrix in which we automatically and adaptively determine the time window. The proposed method uses polarization analysis as applied to multicomponent seismic by waveform separation and filtering.}, language = {en} } @article{MolkenthinScherbaumGriewanketal.2017, author = {Molkenthin, Christian and Scherbaum, Frank and Griewank, Andreas and Leovey, Hernan and Kucherenko, Sergei and Cotton, Fabrice}, title = {Derivative-Based Global Sensitivity Analysis: Upper Bounding of Sensitivities in Seismic-Hazard Assessment Using Automatic Differentiation}, series = {Bulletin of the Seismological Society of America}, volume = {107}, journal = {Bulletin of the Seismological Society of America}, publisher = {Seismological Society of America}, address = {Albany}, issn = {0037-1106}, doi = {10.1785/0120160185}, pages = {984 -- 1004}, year = {2017}, abstract = {Seismic-hazard assessment is of great importance within the field of engineering seismology. Nowadays, it is common practice to define future seismic demands using probabilistic seismic-hazard analysis (PSHA). Often it is neither obvious nor transparent how PSHA responds to changes in its inputs. In addition, PSHA relies on many uncertain inputs. Sensitivity analysis (SA) is concerned with the assessment and quantification of how changes in the model inputs affect the model response and how input uncertainties influence the distribution of the model response. Sensitivity studies are challenging primarily for computational reasons; hence, the development of efficient methods is of major importance. Powerful local (deterministic) methods widely used in other fields can make SA feasible, even for complex models with a large number of inputs; for example, automatic/algorithmic differentiation (AD)-based adjoint methods. Recently developed derivative-based global sensitivity measures can combine the advantages of such local SA methods with efficient sampling strategies facilitating quantitative global sensitivity analysis (GSA) for complex models. In our study, we propose and implement exactly this combination. It allows an upper bounding of the sensitivities involved in PSHA globally and, therefore, an identification of the noninfluential and the most important uncertain inputs. To the best of our knowledge, it is the first time that derivative-based GSA measures are combined with AD in practice. In addition, we show that first-order uncertainty propagation using the delta method can give satisfactory approximations of global sensitivity measures and allow a rough characterization of the model output distribution in the case of PSHA. An illustrative example is shown for the suggested derivative-based GSA of a PSHA that uses stochastic ground-motion simulations.}, language = {en} } @article{BoraCottonScherbaumetal.2017, author = {Bora, Sanjay Singh and Cotton, Fabrice and Scherbaum, Frank and Edwards, Benjamin and Traversa, Paola}, title = {Stochastic source, path and site attenuation parameters and associated variabilities for shallow crustal European earthquakes}, series = {Bulletin of earthquake engineering : official publication of the European Association for Earthquake Engineering}, volume = {15}, journal = {Bulletin of earthquake engineering : official publication of the European Association for Earthquake Engineering}, publisher = {Springer}, address = {Dordrecht}, issn = {1570-761X}, doi = {10.1007/s10518-017-0167-x}, pages = {4531 -- 4561}, year = {2017}, abstract = {We have analyzed the recently developed pan-European strong motion database, RESORCE-2012: spectral parameters, such as stress drop (stress parameter, Delta sigma), anelastic attenuation (Q), near surface attenuation (kappa(0)) and site amplification have been estimated from observed strong motion recordings. The selected dataset exhibits a bilinear distance-dependent Q model with average kappa(0) value 0.0308 s. Strong regional variations in inelastic attenuation were also observed: frequency-independent Q(0) of 1462 and 601 were estimated for Turkish and Italian data respectively. Due to the strong coupling between Q and kappa(0), the regional variations in Q have strong impact on the estimation of near surface attenuation kappa(0). kappa(0) was estimated as 0.0457 and 0.0261 s for Turkey and Italy respectively. Furthermore, a detailed analysis of the variability in estimated kappa(0) revealed significant within-station variability. The linear site amplification factors were constrained from residual analysis at each station and site-class type. Using the regional Q(0) model and a site-class specific kappa(0), seismic moments (M-0) and source corner frequencies f (c) were estimated from the site corrected empirical Fourier spectra. Delta sigma did not exhibit magnitude dependence. The median Delta sigma value was obtained as 5.75 and 5.65 MPa from inverted and database magnitudes respectively. A comparison of response spectra from the stochastic model (derived herein) with that from (regional) ground motion prediction equations (GMPEs) suggests that the presented seismological parameters can be used to represent the corresponding seismological attributes of the regional GMPEs in a host-to-target adjustment framework. The analysis presented herein can be considered as an update of that undertaken for the previous Euro-Mediterranean strong motion database presented by Edwards and Fah (Geophys J Int 194(2):1190-1202, 2013a).}, language = {en} } @article{BeauvalHainzlScherbaum2006, author = {Beauval, Celine and Hainzl, Sebastian and Scherbaum, Frank}, title = {The impact of the spatial uniform distribution of seismicity on probabilistic seismic-hazard estimation}, series = {Bulletin of the Seismological Society of America}, volume = {96}, journal = {Bulletin of the Seismological Society of America}, number = {6}, publisher = {GeoScienceWorld}, address = {Alexandria, Va.}, issn = {0037-1106}, doi = {10.1785/0120060073}, pages = {2465 -- 2471}, year = {2006}, abstract = {The first step in the estimation of probabilistic seismic hazard in a region commonly consists of the definition and characterization of the relevant seismic sources. Because in low-seismicity regions seismicity is often rather diffuse and faults are difficult to identify, large areal source zones are mostly used. The corresponding hypothesis is that seismicity is uniformly distributed inside each areal seismic source zone. In this study, the impact of this hypothesis on the probabilistic hazard estimation is quantified through the generation of synthetic spatial seismicity distributions. Fractal seismicity distributions are generated inside a given source zone and probabilistic hazard is computed for a set of sites located inside this zone. In our study, the impact of the spatial seismicity distribution is defined as the deviation from the hazard value obtained for a spatially uniform seismicity distribution. From the generation of a large number of synthetic distributions, the correlation between the fractal dimension D and the impact is derived. The results show that the assumption of spatially uniform seismicity tends to bias the hazard to higher values. The correlation can be used to determine the systematic biases and uncertainties for hazard estimations in real cases, where the fractal dimension has been determined. We apply the technique in Germany (Cologne area) and in France (Alps).}, language = {en} }