Figure 4 - uploaded by Luis Ceferino
Content may be subject to copyright.
Source publication
This paper introduces an innovative probabilistic model for assessing earthquake rupture occurrence. This model is able to account for the complexity of time and space interactions of ruptures. The rupture occurrence is modeled as a Multivariate Bernoulli that is updated as a function of the time since the last rupture at different locations of the...
Context in source publication
Context 1
... algorithm uses the calibrated parameters found in the previous subsections. Figure 4a and b show one realization of the rupture process for the next 10,000 years using a γ values equal to 96 and 289 km, respectively. The graphs indicate that as the value of γ increases, the number of events decreases and rupture lengths increase. ...
Citations
... The model presented here is the 2-D extension of the 1-D probabilistic rupture model presented by Ceferino et al. (2017b) in Appendix A. The 1-D probabilistic model was built to assess the spatial and temporal interactions of earthquake mainshock occurrences. The model is based on the fundamental premise of the elastic rebound theory, which states that earthquakes are the result of cyclic processes characterized by accumulation of strain and stress over time in a tectonic fault that are released through earthquake ruptures. ...
... Similar to the 1-D version of the model by Ceferino et al. (2017b) in Appendix A, X t is the rupture vector at year t, where X t ∈ {0, 1} N . N is the total number of fault sections and t is the time index, which is defined to have time steps of one year. ...
Worldwide, earthquakes have caused tens or even hundreds of thousands of injuries in matter of minutes in countries such as China in 2008 and Turkey in 1999.
Because of their suddenness and their long and uncertain recurrence times, large earthquakes pose more critical challenges for emergency response than other natural disasters, for instance, hurricanes or wildfires.
This dissertation develops novel models that capture key processes that govern the emergency response of complex urban centers to large earthquakes, and it also demonstrates through case studies how these methods can aid emergency managers and policymakers to create better risk mitigation measures.
The models and case studies in this dissertation are built using an interdisciplinary two-part approach.
The first part combines engineering with seismology to improve existing methods for seismic hazard assessment.
I develop a time-dependent model that characterizes complex space and time tectonic interactions of large earthquakes.
By applying this model to the subduction fault offshore the coast of Peru, I show that these developments successfully capture seismic gap effects, leading to variations in seismic hazard predictions by a factor of four.
Then, I construct a Bayesian parameter estimation technique that leverages synthetic earthquake catalogs to supplement the existing limited historical catalogs for large earthquakes.
As part of the dissertation, I built the synthetic catalogs for the subduction fault in Peru using physics-based simulations based on the rate-and-state friction law and high-performance computing.
My results show that these synthetic catalgs reduce the uncertainty in the parameter estimates by a factor of two and also improve parameter median estimates, triggering additional variations in the time-dependent hazard estimates by up to 40\%.
The second part of this dissertation merges engineering with emergency medicine to model emergency response and design effective plans to treat and transfer patients more effectively after an earthquake.
First, I extend the performance-based earthquake engineering formulation from focusing on single buildings to multiple buildings. Then, I exploit the variables' interdependence structure to model multiseverity earthquake casualties with computational efficiency.
Applying the model to Lima demonstrates that this model can outperform widely used empirical fatality formulas for countries where earthquake fatality data is sparse and not recent.
Next, I develop a network model that captures critical features of an earthquake emergency response by combining the multiseverity casualty model with predictions of post-earthquake hospital functionality based on an extensive hospital infrastructure dataset in Lima.
The model identifies the neighborhoods that will most likely be underserved by healthcare services after an earthquake.
Finally, I formulate an optimization procedure that designs strategies for patient transfers, ambulance usage and deployment of field hospitals to make treatment more effective after an earthquake.
A case study demonstrates that high-coordination emergency plans in large urban centers can reduce patient waiting times by a factor of three.
I envision that policymakers and emergency planners who leverage the methods and findings in this dissertation will be able to develop more robust risk reduction programs, protect their most vulnerable residents, and potentially save more lives after an earthquake.
... The model presented here is the 2-D extension of the 1-D probabilistic rupture model presented by Ceferino et al. (2017). The 1-D probabilistic model was built to assess the spatial and temporal interactions of earthquake mainshock occurrences. ...
... Figure 1 shows the sections with ruptures occurring at time t as shaded areas. Similar to the 1-D version of the model (Ceferino et al. (2017)), X t is the rupture vector at year t, where X t ∈ {0, 1} N . N is the total number of fault sections and t is the time index, which is defined to have time steps of one year. ...
This paper presents a probabilistic formulation for modeling earthquake rupture processes of mainshocks. A correlated multivariate Bernoulli distribution is used to model rupture occurrence. The model captures time interaction through the use of Brownian passage-time (BPT) distributions to assess rupture interarrival in multiple sections of the fault, and it also considers spatial interaction through the use of spatial correlograms. The correlograms represents the effect of rupture nucleation and propagation. This model is proposed as an attractive alternative to existing probabilistic models because it (1) incorporates time and space interactions of mainshocks, (2) preserves the marginal distributions of interarrival times after including spatial rupture interactions (i.e., model consistency), and (3) has an implicit physical interpretation aligned with recent rupture behavior observations. The proposed model is applied to assess the occurrence of large interface earthquakes in the subduction zone along the Coast of Lima, Peru. The model matches both the annual magnitude exceedance rates and the average seismic moment release in the tectonic region. Time-dependent seismic hazard in the region is also calculated, and the results demonstrate that by accounting for recent earthquake occurrences, the inclusion of time-dependent effects can reduce the 30-year seismic hazard by a factor of four. 2
... Finally, the results of applying the parameter estimation technique are shown and discussed. Ceferino et al. (2018bCeferino et al. ( , 2017 presented a probabilistic formulation for modeling time and space interactions between earthquake mainshocks. The model uses Brownian Passage Time (BPT) distributions to assess rupture interevent times in multiple fault segments to represent time interactions, and it uses spatial correlograms to represent spatial interactions of rupture occurrences. ...
This paper presents a robust parameter estimation technique for a probabilistic earthquake hazard model that captures time and space interactions between earthquake mainshocks. The approach addresses the existing limitations of parameter estimation techniques by developing a Bayesian formulation and leveraging physics-based simulated synthetic catalogs to expand the limited datasets of historical catalogs. The technique is based on a two-step Bayesian update that uses the synthetic catalog to perform a first parameter estimation and then uses the historical catalog to further calibrate the parameters. We applied this technique to analyze the occurrence of large-magnitude interface earthquakes along 650 km of the central subduction zone in Peru, located offshore of Lima. We built 2,000-years-long synthetic catalogs using quasi-dynamic earthquake cycle simulations based on the rate-and-state friction law. The validity of the synthetic catalogs was verified by comparing their annual magnitude exceedence rates to those of recorded seismicity and their predicted areas of high interseismic coupling to those inferred from geodetic data. We show that when the Bayesian update uses the combination of synthetic and historical data, instead of only the historical data, it reduces the uncertainty of model parameter estimates by 45% on average. Further, our results show that the time-dependent seismic hazard estimated with the both datasets is 40% smaller than the one estimated with only the historical data.
... Finally, the paper evaluates the implications of including synthetic catalogs to determine large-earthquake recurrence and time-dependent seismic hazard in the region. Ceferino et al. (2020Ceferino et al. ( , 2017 presented a probabilistic formulation for modeling time and space interactions between earthquake mainshocks. The following subsections briefly describe the earthquake rupture model for completeness. ...
This paper presents a robust parameter estimation technique for a probabilistic earthquake hazard model that captures time and space interactions between earthquake mainshocks. The approach addresses the existing limitations of parameter estimation techniques by developing a Bayesian formulation and leveraging physics-based simulated synthetic catalogs to expand the limited datasets of historical catalogs. The technique is based on a two-step Bayesian update that uses the synthetic catalog to perform a first parameter estimation and then uses the historical catalog to further calibrate the parameters. We applied this technique to analyze the occurrence of large-magnitude interface earthquakes along 650 km of the central subduction zone in Peru, located offshore of Lima. We built 2,000-years-long synthetic catalogs using quasi-dynamic earthquake cycle simulations based on the rate-and-state friction law. The validity of the synthetic catalogs was verified by comparing their annual magnitude exceedence rates to those of recorded seismicity and their predicted areas of high interseismic coupling to those inferred from geodetic data. We show that when the Bayesian update uses the combination of synthetic and historical data, instead of only the historical data, it reduces the uncertainty of model parameter estimates by 45% on average. Further, our results show that the time-dependent seismic hazard estimated with the both datasets is 40% smaller than the one estimated with only the historical data.
... Moreover, the model introduces a spatial correlation formulation that represents the stress interactions among neighboring zones in the tectonic region. A 1-D version of the model was originally proposed by Ceferino, Kiremidjian, and Deierlein (2017), and the 2-D version of the model and an in-depth study of its properties was presented in USA, ggd@stanford.edu Additionally, this paper describes two approaches for estimation of the model parameters. ...
... Model discretization of a 1-D representation of a tectonic fault. Extracted fromCeferino et al. (2017). ...
Earthquake rupture occurrence modeling is the basis of seismic risk and Performance-based Earthquake Engineering (PBEE). This paper summarizes a probabilistic formulation for modeling time and space interactions of earthquake rupture occurrence in a tectonic region. The formulation represents the elastic-rebound behavior in the tectonic plates and models the stress interaction of neighboring areas of the faults through spatial correlation. The paper also describes two methods for the estimation of the model parameters. The first method is a simple approach that estimates the parameters in pairs and then calibrates the spatial correlation parameter. The second one uses a Bayesian update to estimate all parameters at simultaneously. Both approaches are applied to model the rupture occurrence of large interface earthquakes on the subduction zone along the Coast of Lima, Peru. The Bayesian update is demonstrated to be a more reliable estimation technique of the two approaches as it predicts a hazard rate that is closer to the data. The simple approach overpredicts the hazard rate by more than 25% for areas where data are very sparse.
Hospital systems play a critical role in treating injuries during disaster emergency responses. Simultaneously, natural disasters hinder their ability to operate at full capacity. Thus, cities must develop strategies that enable hospitals’ effective disaster operations. Here, we present a methodology to evaluate emergency response based on a model that assesses the loss of hospital functions and quantifies multiseverity injuries as a result of earthquake damage. The proposed methodology can design effective plans for patient transfers and allocation of ambulances and mobile operating rooms. This methodology is applied to Lima, Peru, subjected to a disaster scenario following a magnitude 8.0 earthquake. Our results show that the spatial distribution of healthcare demands mismatches the post-earthquake capacities of hospitals, leaving large zones on the periphery significantly underserved. This study demonstrates how plans that leverage hospital-system coordination can address this demand-capacity mismatch, reducing waiting times of critically injured patients by factors larger than two.
This article presents a probabilistic formulation for modeling earthquake rupture processes of mainshocks. A correlated multivariate Bernoulli distribution is used to model rupture occurrence. The model captures time interaction through the use of Brownian passage-time distributions to assess rupture interarrival in multiple sections of the fault, and it also considers spatial interaction through the use of spatial correlograms. The correlograms represents the effect of rupture nucleation and propagation. This model is proposed as an attractive alternative to existing probabilistic models because it (1) incorporates time and space interactions of mainshocks, (2) preserves the marginal distributions of interarrival times after including spatial rupture interactions, that is, model consistency, and (3) has an implicit physical interpretation aligned with rupture behavior observations. The proposed model is applied to assess the occurrence of large interface earthquakes in the subduction fault along the coast of Lima, Peru. The model matches well both the annual magnitude exceedance rates and the average seismic moment release in the tectonic region. The Akaike information criterion (AIC) test confirms that our model performs statistically better than models that do not capture earthquake space interactions. AIC also shows that the spherical correlogram outperforms the exponential correlogram at reproducing earthquake data. Finally, time-dependent seismic hazard in the region is calculated, and the results demonstrate that by accounting for recent earthquake occurrences, the inclusion of time-dependent effects can reduce the 30 yr seismic hazard by a factor of 4.
