Emerging high-performance computing systems, combined with increasingly detailed 3-D Earth models... more Emerging high-performance computing systems, combined with increasingly detailed 3-D Earth models and physically consistent numerical wave propagation solvers, are opening up new opportunities for urgent seismic computing. This may help, for instance, to guide emergency response teams in the wake of large earthquakes. A key component of urgent seismic computing is the early availability of source mechanism estimates, well before conventional and time-consuming moment tensor inversions are carried out and published. Here, we introduce a methodology that rapidly estimates focal mechanisms (FM) for moderate and large earthquakes (M w > 4.0) by means of statistical and clustering algorithms. The fundamental rationale behind the method is that events of a certain size tend to be similar to other events of similar size in similar locations. In this work, two different strategies are used to provide different FM solutions: the first is based only in spatial considerations including statistical analysis, and the other one is based on a data clustering algorithm. We exemplify our methodology with six different subsets of the openaccess Global Centroid Moment Tensor (GCMT) catalog. Specifically, our study datasets include events from Japan, New Zealand, California, Mexico, Iceland, and Italy, which represent six seismically active regions, with a large FM variability. Our results show a 70-85% agreement between our fast FM estimates and inversion results, depending on the particular tectonic region, dataset size, and magnitude threshold. In addition, our FM estimation strategies only spend few seconds for processing, since they are totally independent of seismic record retrieval and inversion. Albeit not meant to be a substitute for CMT inversions, our methodologies can bridge the time gap between earthquake detection and FM inversion.
Los efectos destructivos de los terremotos pueden incrementarse en las cuencas sedimentarias debi... more Los efectos destructivos de los terremotos pueden incrementarse en las cuencas sedimentarias debido no solo a la presencia de sedimentos, sino tambien a la geometria propia de la cuenca. En este trabajo se realiza el analisis de los resultados del modelado numerico 2D de la propagacion de las ondas sismicas para estimar la respuesta del suelo en las ciudades de Barquisimeto y Cabudare, ubicadas en las cercanias de la falla de Bocono, en el occidente de Venezuela. En esta region se consideraron 9 perfiles utilizando como fuente sismica dos eventos escenarios de Mw 6,0 y 7,0, respectivamente. Para el modelado de las ondas sismicas se utilizo el metodo hibrido que combina la tecnica analitica de suma modal con el metodo numerico de diferencias finitas. A partir de los sismogramas sinteticos, se obtuvieron las razones de los espectros de respuesta de la aceleracion ( response spectra ratio, RSR) en 879 sitios distribuidos sobre los 9 perfiles. Los RSR fueron posteriormente agrupados med...
<p>Traditional Probabilistic Seismic Hazard Analysis (PSHA) estimates the level of earthqua... more <p>Traditional Probabilistic Seismic Hazard Analysis (PSHA) estimates the level of earthquake ground shaking that is expected to be exceeded with a given recurrence time on the basis of&#160; historical earthquake catalogues and empirical and time-independent Ground Motion Prediction Equations (GMPEs). The smooth nature of GMPEs usually disregards some well known drivers of ground motion characteristics associated with fault rupture processes, in particular in the near-fault region, complex source-site propagation of seismic waves, and sedimentary basin response. Modern physics-based earthquake simulations can consider all these effects, but require a large set of input parameters for which constraints may often be scarce. However, with the aid of high-performance computing (HPC) infrastructures&#160; the parameter space may be sampled in an efficient and scalable manner allowing for a large suite of site-specific ground motion simulations that approach the center, body and range of expected ground motions.&#160;</p><p>CyberShake is a HPC platform designed to undertake physics-based PSHA from a large suite of earthquake simulations. These simulations are based on seismic reciprocity, rendering PSHA computationally tractable for hundreds of thousands potential earthquakes. For each site of interest, multiple kinematic rupture scenarios, derived by varying slip distributions and hypocenter location across the pre-defined fault system, are generated from an input Earthquake Forecast Model (EFM). Each event is simulated to determine ground motion intensities, which are synthesized into hazard results. CyberShake has been developed by the Southern California Earthquake Center, and used so far to assess seismic hazard in California. This work focuses on the CyberShake migration to the seismic region of South Iceland (63.5&#176;- 64.5&#176;N, 20&#176;-22&#176;W) where the largely sinistral East-West transform motion across the tectonic margin is taken up by a complex array of near-vertical and parallel North-South oriented dextral transform faults in the South Iceland Seismic Zone (SISZ) and the Reykjanes Peninsula Oblique Rift (RPOR). Here, we describe the main steps of migrating CyberShake to the SISZ and RPOR, starting by setting up a relational input database describing potential causative faults and rupture characteristics, and key sites of interest. To simulate our EFM, we use the open source code SHERIFS, a logic-tree method that converts the slip rates of complex fault systems to the corresponding annual seismicity rate. The fault slip rates are taken from a new 3D physics-based fault model for the SISZ-RPOR transform fault system. To validate model and simulation parameters, two validation steps using key CyberShake modeling tools have been carried out. First, we perform simulations of historical earthquakes and compare the synthetics with recorded ground motions and results from other forward simulations. Second, we adjust the rupture kinematics to make slip distributions more representative of SISZ-type earthquakes by comparing with static slip distributions of past significant earthquakes. Finally, we run CyberShake and compare key parameters of the synthetic ground motions with new GMPEs available for the study region. The successful migration and use of CyberShake in South Iceland is the first step of a full-scale physics-based PSHA in the region, and showcases the implementation of CyberShake in new regions.</p>
Seismic modelling algorithms, with few exceptions, discretize space and time independently. While... more Seismic modelling algorithms, with few exceptions, discretize space and time independently. While spatial discretization is frequently achieved with high-order accuracy to allow for coarser grids, the temporal integration is regarded as less important and hence dealt with using low accuracy operators. We will compare different time integration techniques for elastic wave modelling and evaluate their cost and benefit in terms of the accuracy they can achieve, additional memory required and overall compute cost. Our implementations include a standard twostep Leap-frog scheme, a less popular Leap-frog method of three levels, Runge Kutta of three and four stages and other multistep methods. We will focus on staggered-grid explicit time-domain finite-difference algorithms for the elastic wave equation. Our tests involve both different propagation distances and Courant numbers. We will end up with a recommendation of spatial and temporal orders and integration techniques depending on the ...
Freelance, part-time, work-at-home, and other flexible jobs are changing the concept of workplace... more Freelance, part-time, work-at-home, and other flexible jobs are changing the concept of workplace, and bringing information and content exchange problems to companies. For instance, the emerging case of companies with mobile workers with different types of corporate devices. Geographically spread corporations may use remote distribution of software and data to attend employees' demands, by exploiting emerging delivery technologies. In this context, cost-efficient software distribution is crucial to allow business evolution and make IT infrastructures more agile. On the other hand, container based virtualization technology is shaping the new trends of software deployment and infrastructure design. We envision current and future enterprise IT management trends evolving towards container based software delivery over Hybrid CDNs. This paper presents a novel cost-efficient Quality-of-Service (QoS) aware analytical model and a Hybrid CDN-P2P architecture for enterprise software distribution. First, we test this model for a wide range of companies ranging from multinationals to SMEs, by using the network simulation tool PeerSim. These experiments emulate content distribution carry out by the real commercial delivery networks KeyCDN, MaxCDN, CDN77 and BunnyCDN, operating under various industrial hypothetical scenarios. We conclude that MaxCDN offers the best cost-QoS trade-off. Later, we include our CDN-P2P architecture into this experimental setting, and use CDN-based results as reference. This study aims at assessing if building a proprietary-owned hybrid CDN-P2P is the most cost-effective solution in a long-term. Simulation results show overall economic savings between 5% and 20%, compared to just hiring resources from a commercial CDN, while guaranteeing satisfactory QoS levels in terms of response times and number of served requests.
Summary This work is an exploratory study of coupling high-order time integrations to a finite-di... more Summary This work is an exploratory study of coupling high-order time integrations to a finite-difference (FD) spatial discretization of the 1-D wave equation that combines eigth-order differencing at grid interior, with lateral formulas of order sixth and fourthat boundary neighborhood. This reduction of spatial accuracy at the grid vecinity of free surfaces is a known stability limitation of FD methods, when coupled to the two-step Leap-frog (LF) time stepping, which is widely used on seismic modeling. We first implement LF time integrations with an arbitrary accuracy order, as given from a standard Lax-Wendroff procedure, and compare results from the fourth-, sixth-, and twelfth order schemes, against the popular second-order LF. Our emphirical analyses establish the CFL stability constraints for propagation on an homogeneous medium, as first results, and then consider velocity heterogeneities when assessing dispersion and dissipation anomalies. Finally, we use a rapid expansion ...
Machine Learning, Optimization, and Data Science, 2019
Reliable earthquake detection and location algorithms are necessary to properly catalog and analy... more Reliable earthquake detection and location algorithms are necessary to properly catalog and analyze the continuously growing seismic records. This paper reports the results of applying Long Short-Term Memory (LSTM) networks to single-station three-channel waveforms for P-wave earthquake detection in western and north central regions of Venezuela. Precisely, we apply our technique to study the seismicity along the dextral strike-slip Boconó and La Victoria-San Sebastián faults, with complex tectonics driven by the interactions between the South American and Caribbean plates.
The cornerstone of mimetic finite differentiation (FD) is that discrete gradients and divergences... more The cornerstone of mimetic finite differentiation (FD) is that discrete gradients and divergences, in combination with a novel boundary flux operator satisfies an approximation to the Gauss–Divergence theorem. In this paper, we propose a spatially staggered second-order discretization of the acoustic wave equation that fully exploits all these mimetic operators on the implementation of free surface and absorbing boundary conditions. Explicit time integration is carried out by using the standard three-level central FD stencil. Applications of this new method to simple 2-D seismic scenarios are also presented and discussed.
An earthquake is the vibration pattern of the Earth’s crust induced by the sliding of geological ... more An earthquake is the vibration pattern of the Earth’s crust induced by the sliding of geological faults. They are usually recorded for later studies. However, strong earthquakes are rare, small-magnitude events may pass unnoticed and monitoring networks are limited in number and efficiency. Thus, earthquake catalog are incomplete and scarce, and researchers have developed simulators of such catalogs. In this work, we start from synthetic catalogs generated with the TREMOL-3D software. TREMOL-3D is a stochastic-based method to produce earthquake catalogs with different statistical patterns, depending on certain input parameters that mimics physical parameters. When an appropriate set of parameters are used, TREMOL-3D could generate synthetic catalogs with similar statistical properties observed in real catalogs. However, because of the size of the parameter space, a manual searching becomes unbearable. Therefore, aiming at increasing the efficiency of the parameter search, we here im...
Mimetic finite difference (MFD) approximations of continuous gradient and divergente operators sa... more Mimetic finite difference (MFD) approximations of continuous gradient and divergente operators satisfy a discrete version of the Gauss-Divergente theorem on staggered grids. On the mimetic approximation of this integral conservation principle, an unique boundary flux operator is introduced that also intervenes on the discretization of a given boundary value problem (BVP). In this work, we present a second-order MFD scheme for seismic wave propagation on staggered grids that discretized free surface and absorbing boundary conditions (ABC) with same accuracy order. This scheme is time explicit after coupling a central three-level finite difference (FD) stencil for numerical integration. Here, we briefly discuss the convergence properties of this scheme and show its higher accuracy on a challenging test when compared to a traditional FD method. Preliminary applications to 2-D seismic scenarios are also presented and show the potential of the mimetic finite differene method.
Abstract. Seismicity and magnitude distributions are fundamental for seismic hazard analysis. The... more Abstract. Seismicity and magnitude distributions are fundamental for seismic hazard analysis. The Mexican subduction margin along the Pacific Coast is one of the most active seismic zones in the world, which makes it an optimal region for observation and experimentation analyses. Some remarkable seismicity features have been observed on a subvolume of this subduction region, suggesting that the observed simplicity of earthquake sources arises from the rupturing of single asperities. This subregion has been named SUB3 in a recent seismotectonic regionalization of Mexico. In this work, we numerically test this hypothesis using the TREMOL (sThochastic Rupture Earthquake MOdeL) v0.1.0 code. As test cases, we choose four of the most significant recent events (6.5 Mw 7.8) that occurred in the Guerrero–Oaxaca region (SUB3) during the period 1988–2018, and whose associated seismic histories are well recorded in the regional catalogs. Synthetic seismicity results show a reasonable fit to t...
Emerging high-performance computing systems, combined with increasingly detailed 3-D Earth models... more Emerging high-performance computing systems, combined with increasingly detailed 3-D Earth models and physically consistent numerical wave propagation solvers, are opening up new opportunities for urgent seismic computing. This may help, for instance, to guide emergency response teams in the wake of large earthquakes. A key component of urgent seismic computing is the early availability of source mechanism estimates, well before conventional and time-consuming moment tensor inversions are carried out and published. Here, we introduce a methodology that rapidly estimates focal mechanisms (FM) for moderate and large earthquakes (M w > 4.0) by means of statistical and clustering algorithms. The fundamental rationale behind the method is that events of a certain size tend to be similar to other events of similar size in similar locations. In this work, two different strategies are used to provide different FM solutions: the first is based only in spatial considerations including statistical analysis, and the other one is based on a data clustering algorithm. We exemplify our methodology with six different subsets of the openaccess Global Centroid Moment Tensor (GCMT) catalog. Specifically, our study datasets include events from Japan, New Zealand, California, Mexico, Iceland, and Italy, which represent six seismically active regions, with a large FM variability. Our results show a 70-85% agreement between our fast FM estimates and inversion results, depending on the particular tectonic region, dataset size, and magnitude threshold. In addition, our FM estimation strategies only spend few seconds for processing, since they are totally independent of seismic record retrieval and inversion. Albeit not meant to be a substitute for CMT inversions, our methodologies can bridge the time gap between earthquake detection and FM inversion.
Los efectos destructivos de los terremotos pueden incrementarse en las cuencas sedimentarias debi... more Los efectos destructivos de los terremotos pueden incrementarse en las cuencas sedimentarias debido no solo a la presencia de sedimentos, sino tambien a la geometria propia de la cuenca. En este trabajo se realiza el analisis de los resultados del modelado numerico 2D de la propagacion de las ondas sismicas para estimar la respuesta del suelo en las ciudades de Barquisimeto y Cabudare, ubicadas en las cercanias de la falla de Bocono, en el occidente de Venezuela. En esta region se consideraron 9 perfiles utilizando como fuente sismica dos eventos escenarios de Mw 6,0 y 7,0, respectivamente. Para el modelado de las ondas sismicas se utilizo el metodo hibrido que combina la tecnica analitica de suma modal con el metodo numerico de diferencias finitas. A partir de los sismogramas sinteticos, se obtuvieron las razones de los espectros de respuesta de la aceleracion ( response spectra ratio, RSR) en 879 sitios distribuidos sobre los 9 perfiles. Los RSR fueron posteriormente agrupados med...
<p>Traditional Probabilistic Seismic Hazard Analysis (PSHA) estimates the level of earthqua... more <p>Traditional Probabilistic Seismic Hazard Analysis (PSHA) estimates the level of earthquake ground shaking that is expected to be exceeded with a given recurrence time on the basis of&#160; historical earthquake catalogues and empirical and time-independent Ground Motion Prediction Equations (GMPEs). The smooth nature of GMPEs usually disregards some well known drivers of ground motion characteristics associated with fault rupture processes, in particular in the near-fault region, complex source-site propagation of seismic waves, and sedimentary basin response. Modern physics-based earthquake simulations can consider all these effects, but require a large set of input parameters for which constraints may often be scarce. However, with the aid of high-performance computing (HPC) infrastructures&#160; the parameter space may be sampled in an efficient and scalable manner allowing for a large suite of site-specific ground motion simulations that approach the center, body and range of expected ground motions.&#160;</p><p>CyberShake is a HPC platform designed to undertake physics-based PSHA from a large suite of earthquake simulations. These simulations are based on seismic reciprocity, rendering PSHA computationally tractable for hundreds of thousands potential earthquakes. For each site of interest, multiple kinematic rupture scenarios, derived by varying slip distributions and hypocenter location across the pre-defined fault system, are generated from an input Earthquake Forecast Model (EFM). Each event is simulated to determine ground motion intensities, which are synthesized into hazard results. CyberShake has been developed by the Southern California Earthquake Center, and used so far to assess seismic hazard in California. This work focuses on the CyberShake migration to the seismic region of South Iceland (63.5&#176;- 64.5&#176;N, 20&#176;-22&#176;W) where the largely sinistral East-West transform motion across the tectonic margin is taken up by a complex array of near-vertical and parallel North-South oriented dextral transform faults in the South Iceland Seismic Zone (SISZ) and the Reykjanes Peninsula Oblique Rift (RPOR). Here, we describe the main steps of migrating CyberShake to the SISZ and RPOR, starting by setting up a relational input database describing potential causative faults and rupture characteristics, and key sites of interest. To simulate our EFM, we use the open source code SHERIFS, a logic-tree method that converts the slip rates of complex fault systems to the corresponding annual seismicity rate. The fault slip rates are taken from a new 3D physics-based fault model for the SISZ-RPOR transform fault system. To validate model and simulation parameters, two validation steps using key CyberShake modeling tools have been carried out. First, we perform simulations of historical earthquakes and compare the synthetics with recorded ground motions and results from other forward simulations. Second, we adjust the rupture kinematics to make slip distributions more representative of SISZ-type earthquakes by comparing with static slip distributions of past significant earthquakes. Finally, we run CyberShake and compare key parameters of the synthetic ground motions with new GMPEs available for the study region. The successful migration and use of CyberShake in South Iceland is the first step of a full-scale physics-based PSHA in the region, and showcases the implementation of CyberShake in new regions.</p>
Seismic modelling algorithms, with few exceptions, discretize space and time independently. While... more Seismic modelling algorithms, with few exceptions, discretize space and time independently. While spatial discretization is frequently achieved with high-order accuracy to allow for coarser grids, the temporal integration is regarded as less important and hence dealt with using low accuracy operators. We will compare different time integration techniques for elastic wave modelling and evaluate their cost and benefit in terms of the accuracy they can achieve, additional memory required and overall compute cost. Our implementations include a standard twostep Leap-frog scheme, a less popular Leap-frog method of three levels, Runge Kutta of three and four stages and other multistep methods. We will focus on staggered-grid explicit time-domain finite-difference algorithms for the elastic wave equation. Our tests involve both different propagation distances and Courant numbers. We will end up with a recommendation of spatial and temporal orders and integration techniques depending on the ...
Freelance, part-time, work-at-home, and other flexible jobs are changing the concept of workplace... more Freelance, part-time, work-at-home, and other flexible jobs are changing the concept of workplace, and bringing information and content exchange problems to companies. For instance, the emerging case of companies with mobile workers with different types of corporate devices. Geographically spread corporations may use remote distribution of software and data to attend employees' demands, by exploiting emerging delivery technologies. In this context, cost-efficient software distribution is crucial to allow business evolution and make IT infrastructures more agile. On the other hand, container based virtualization technology is shaping the new trends of software deployment and infrastructure design. We envision current and future enterprise IT management trends evolving towards container based software delivery over Hybrid CDNs. This paper presents a novel cost-efficient Quality-of-Service (QoS) aware analytical model and a Hybrid CDN-P2P architecture for enterprise software distribution. First, we test this model for a wide range of companies ranging from multinationals to SMEs, by using the network simulation tool PeerSim. These experiments emulate content distribution carry out by the real commercial delivery networks KeyCDN, MaxCDN, CDN77 and BunnyCDN, operating under various industrial hypothetical scenarios. We conclude that MaxCDN offers the best cost-QoS trade-off. Later, we include our CDN-P2P architecture into this experimental setting, and use CDN-based results as reference. This study aims at assessing if building a proprietary-owned hybrid CDN-P2P is the most cost-effective solution in a long-term. Simulation results show overall economic savings between 5% and 20%, compared to just hiring resources from a commercial CDN, while guaranteeing satisfactory QoS levels in terms of response times and number of served requests.
Summary This work is an exploratory study of coupling high-order time integrations to a finite-di... more Summary This work is an exploratory study of coupling high-order time integrations to a finite-difference (FD) spatial discretization of the 1-D wave equation that combines eigth-order differencing at grid interior, with lateral formulas of order sixth and fourthat boundary neighborhood. This reduction of spatial accuracy at the grid vecinity of free surfaces is a known stability limitation of FD methods, when coupled to the two-step Leap-frog (LF) time stepping, which is widely used on seismic modeling. We first implement LF time integrations with an arbitrary accuracy order, as given from a standard Lax-Wendroff procedure, and compare results from the fourth-, sixth-, and twelfth order schemes, against the popular second-order LF. Our emphirical analyses establish the CFL stability constraints for propagation on an homogeneous medium, as first results, and then consider velocity heterogeneities when assessing dispersion and dissipation anomalies. Finally, we use a rapid expansion ...
Machine Learning, Optimization, and Data Science, 2019
Reliable earthquake detection and location algorithms are necessary to properly catalog and analy... more Reliable earthquake detection and location algorithms are necessary to properly catalog and analyze the continuously growing seismic records. This paper reports the results of applying Long Short-Term Memory (LSTM) networks to single-station three-channel waveforms for P-wave earthquake detection in western and north central regions of Venezuela. Precisely, we apply our technique to study the seismicity along the dextral strike-slip Boconó and La Victoria-San Sebastián faults, with complex tectonics driven by the interactions between the South American and Caribbean plates.
The cornerstone of mimetic finite differentiation (FD) is that discrete gradients and divergences... more The cornerstone of mimetic finite differentiation (FD) is that discrete gradients and divergences, in combination with a novel boundary flux operator satisfies an approximation to the Gauss–Divergence theorem. In this paper, we propose a spatially staggered second-order discretization of the acoustic wave equation that fully exploits all these mimetic operators on the implementation of free surface and absorbing boundary conditions. Explicit time integration is carried out by using the standard three-level central FD stencil. Applications of this new method to simple 2-D seismic scenarios are also presented and discussed.
An earthquake is the vibration pattern of the Earth’s crust induced by the sliding of geological ... more An earthquake is the vibration pattern of the Earth’s crust induced by the sliding of geological faults. They are usually recorded for later studies. However, strong earthquakes are rare, small-magnitude events may pass unnoticed and monitoring networks are limited in number and efficiency. Thus, earthquake catalog are incomplete and scarce, and researchers have developed simulators of such catalogs. In this work, we start from synthetic catalogs generated with the TREMOL-3D software. TREMOL-3D is a stochastic-based method to produce earthquake catalogs with different statistical patterns, depending on certain input parameters that mimics physical parameters. When an appropriate set of parameters are used, TREMOL-3D could generate synthetic catalogs with similar statistical properties observed in real catalogs. However, because of the size of the parameter space, a manual searching becomes unbearable. Therefore, aiming at increasing the efficiency of the parameter search, we here im...
Mimetic finite difference (MFD) approximations of continuous gradient and divergente operators sa... more Mimetic finite difference (MFD) approximations of continuous gradient and divergente operators satisfy a discrete version of the Gauss-Divergente theorem on staggered grids. On the mimetic approximation of this integral conservation principle, an unique boundary flux operator is introduced that also intervenes on the discretization of a given boundary value problem (BVP). In this work, we present a second-order MFD scheme for seismic wave propagation on staggered grids that discretized free surface and absorbing boundary conditions (ABC) with same accuracy order. This scheme is time explicit after coupling a central three-level finite difference (FD) stencil for numerical integration. Here, we briefly discuss the convergence properties of this scheme and show its higher accuracy on a challenging test when compared to a traditional FD method. Preliminary applications to 2-D seismic scenarios are also presented and show the potential of the mimetic finite differene method.
Abstract. Seismicity and magnitude distributions are fundamental for seismic hazard analysis. The... more Abstract. Seismicity and magnitude distributions are fundamental for seismic hazard analysis. The Mexican subduction margin along the Pacific Coast is one of the most active seismic zones in the world, which makes it an optimal region for observation and experimentation analyses. Some remarkable seismicity features have been observed on a subvolume of this subduction region, suggesting that the observed simplicity of earthquake sources arises from the rupturing of single asperities. This subregion has been named SUB3 in a recent seismotectonic regionalization of Mexico. In this work, we numerically test this hypothesis using the TREMOL (sThochastic Rupture Earthquake MOdeL) v0.1.0 code. As test cases, we choose four of the most significant recent events (6.5 Mw 7.8) that occurred in the Guerrero–Oaxaca region (SUB3) during the period 1988–2018, and whose associated seismic histories are well recorded in the regional catalogs. Synthetic seismicity results show a reasonable fit to t...
Uploads
Papers by Otilio Rojas