It has become common practice to validate ground-motion simulations based on a variety of time and frequency metrics scaled to quantify the level of agreement between synthetics and data or other reference solutions. There is, however, no agreement about the importance or weight that it ought to be given to each metric. This leads to their selection often being subjective, either based on intended applications or personal preferences. As a consequence, it is difficult for simulators to identify what modeling improvements are needed, which would be easier if they could focus on a reduced number of metrics. We present an analysis that looks into 11 ground-motion validation metrics using semisupervised and supervised machine learning techniques. These techniques help label and classify goodness-of-fit results with the objective of prioritizing and narrowing the choice of these metrics. In particular, we use a validation dataset of a series of physics-based ground-motion simulations done for the 2008 Mw 5.4 Chino Hills, California, earthquake.We study the relationships that exist between 11 metrics and carry out a process where these metrics are understood as part of a multidimensional space. We use a constrained k-means method and conduct a subspace clustering analysis to address the implicit high-dimensional effects. This allows us to label the data in our dataset into four validation categories (poor, fair, good, and excellent) following previous studies.We then develop a family of decision trees using the C5.0 algorithm, from which we select a few trees that help narrow the number of metrics leading to a validation prediction into the four referenced categories. These decision trees can be understood as rapid predictors of the quality of a simulation, or as data-informed classifiers that can help prioritize validation metrics. Our analysis, although limited to the particular dataset used here, indicates that among the 11 metrics considered, the acceleration response spectra and total energy of velocity are the most dominant ones, followed by the peak ground response in terms of acceleration and velocity.

The accurate solution of wave propagation problems requires the appropriate representation of energy losses due to internal friction. These losses are important because their mischaracterization can lead to over- or under-estimation of amplitudes and duration of ground motions. Recent studies show that synthetics from physics-based simulations tend to attenuate at different rates than observations, suggesting that current modeling approaches need to be revised. In physic-based simulation, attenuation is commonly introduced by means of viscoelastic models. Internally, the properties of these models are set based on the material's quality factor, Q. The value of Q for shear waves, Qs, is usually defined according to empirical rules that depend on the shear wave velocity, Vs. Typical Qs-Vs relationships are (piecewise) linear or polynomial functions. Several relationships have been tried in the past, but there is no consensus about what is the most appropriate one. Identifying the parameters that define these relationships through the solution of inverse problems is non-trivial, requires considerable computational resources, and may not lead to a unique solution. In this study we investigate the effectiveness of an approach that combines the use of an artificial neural network (ANN) and a genetic algorithm (GA) to predict ground motion amplitudes and identify the optimal parameters in Qs-Vs relationships. We use direct S-waves as a proxy to evaluate attenuation over distance. First, we train the ANN to predict peak S-wave amplitudes based on a series of simulations with an ample selection of parameters. Then, we use the ANN as a fitness function of the GA, and use the latter to find the optimal parameters based on comparisons between predictions and data. As a proof-of-concept, we test the performance of the proposed approach for the case of the 2008 Mw 5.4 Chino Hills, California, earthquake. Using this approach we find the parameters that lead to the best fit with data. The results include the mean and standard deviation of Qs for each value of Vs. Our initial results for a limited number of stations scattered throughout the simulation domain and for numerical models with a maximum frequency of 0.5 Hz show good promise. We recognize, however, that Q parameters may depend on additional factors such as frequency, depth, path and site effects, and the nature of the traveling waves. Future work will test the method for multiple events and higher frequencies.

Project summary

Project summary

Project summary

Project summary

Significant effort has been devoted over the last two decades to the development of various seismic velocity models for the region of southern California, United States. These models are mostly used in forward wave propagation simulation studies, but also as base models for tomographic and source inversions. Two of these models, the community velocity models CVM-S and CVM-H, are among the most commonly used for this region. This includes two alternative variations to the original models, the recently released CVM-S4.26 which incorporates results from a sequence of tomographic inversions into CVM-S, and the user-controlled option of CVM-H to replace the near-surface profiles with a VS30-based geotechnical model. Although either one of these models is regarded as acceptable by the modeling community, it is known that they have differences in their representation of the crustal structure and sedimentary deposits in the region, and thus can lead to different results in forward and inverse problems. In this paper, we evaluate the accuracy of these models when used to predict the ground motion in the greater Los Angeles region by means of an assessment of a collection of simulations of recent events. In total, we consider 30 moderate-magnitude earthquakes (3.5 < Mw < 5.5) between 1998 and 2014, and compare synthetics with data recorded by seismic networks during these events. The simulations are done using a finite-element parallel code, with numerical models that satisfy a maximum frequency of 1 Hz and a minimum shear wave velocity of 200 m/s. The comparisons between data and synthetics are ranked quantitatively by means of a goodness-of-fit (GOF) criteria. We analyse the regional distribution of the GOF results for all events and all models, and draw conclusions from the results and how these correlate to the models. We find that, in light of our comparisons, the model CVM-S4.26 consistently yields better results.

Project summary

Project summary

This article presents a seismic hazard assessment for northern Iran, where a smoothed seismicity approach has been used in combination with an updated seismic catalog and a ground motion prediction equation recently found to yield good fit with data. We evaluate the hazard over a geographical area including the seismic zones of Azerbaijan, the Alborz Mountain Range, and Kopeh-Dagh, as well as parts of other neighboring seismic zones that fall within our region of interest. In the chosen approach, seismic events are not assigned to specific faults but assumed to be potential seismogenic sources distributed within regular grid cells. After performing the corresponding magnitude conversions, we decluster both historical and instrumental seismicity catalogs to obtain earthquake rates based on the number of events within each cell, and smooth the results to account for the uncertainty in the spatial distribution of future earthquakes. Seismicity parameters are computed for each seismic zone separately, and for the entire region of interest as a single uniform seismotectonic region. In the analysis, we consider uncertainties in the ground motion prediction equation, the seismicity parameters, and combine the resulting models using a logic tree. The results are presented in terms of expected peak ground acceleration (PGA) maps and hazard curves at selected locations, considering exceedance probabilities of 2 and 10% in 50 years for rock site conditions. According to our results, the highest levels of hazard are observed west of the North Tabriz and east of the North Alborz faults, where expected PGA values are between about 0.5 and 1 g for 10 and 2% probability of exceedance in 50 years, respectively. We analyze our results in light of similar estimates available in the literature and offer our perspective on the differences observed. We find our results to be helpful in understanding seismic hazard for northern Iran, but recognize that additional efforts are necessary to obtain more robust estimates at specific areas of interest and different site conditions.

Tehran is one of the densely populated metropolises located in earthquake-prone regions. Tehran, the population of which surpasses 8 million people, is the most populated area in Iran. There are historical evidences confirming that catastrophic earthquakes have destroyed the city in past years. In the present paper, our study covers all parts of Tehran because there is the potential of significant earthquake damage and loss for the entire city. In other words, the development of high-rise building construction in the northern part, the high density of population in the southern area including old masonry buildings, and the existence of important structures in central regions, prevent us from omitting any particular part of the city from damage assessment process. We have used two sets of last available formal data published in 1996 and 2006. To consider the influence of soil conditions, Tehran has been divided into 1246 sub regions; however, in our study the results have been presented using municipality regions and in cumulative manner. Since there is no acceptable statistical data involving estimation of non-structural damage, only structural damages have been assessed. The open source software SELENA is applied to perform probabilistic loss estimates. Due to the lack of studies providing required information from structural point of view in our country, and the existence of similarity between structural codes of Iran and that of United States, HAZUS-MH (Hazard Us Ð Multi Hazard Loss) structures coefficients are used. According to the results, from 1996 to 2006, the mean damage ratio and number of casualties have been reduced, while the economic loss has been increased.