Thesis (Ph.D.)--Drexel University, 2010. / Includes abstract and vita. Includes bibliographical references (leaves 449-468).

Thesis (Ph. D.)--Drexel University, 2006. / Includes abstract. Includes bibliographical references (leaves 230-242).

Remote sensing continues to be an invaluable tool in earthquake damage assessments and emergency response. This study evaluates the effectiveness of multilayer feedforward neural networks, radial basis neural networks, and Random Forests in detecting earthquake damage caused by the 2010 Port-au-Prince, Haiti 7.0 moment magnitude (Mw) event. Additionally, textural and structural features including entropy, dissimilarity, Laplacian of Gaussian, and rectangular fit are investigated as key variables for high spatial resolution imagery classification. Our findings show that each of the algorithms achieved nearly a 90% kernel density match using the United Nations Operational Satellite Applications Programme (UNITAR/UNOSAT) dataset as validation. The multilayer feedforward network was able to achieve an error rate below 40% in detecting damaged buildings. Spatial features of texture and structure were far more important in algorithmic classification than spectral information, highlighting the potential for future implementation of machine learning algorithms which use panchromatic or pansharpened imagery alone. / Master of Science / Satellite imagery can help emergency managers to better respond to the aftermath of deadly earthquakes. This study evaluates the effectiveness of machine learning algorithms in detecting earthquake damage caused by the 2010 Port-au-Prince, Haiti event. Additionally, textural and structural features of high resolution black and white imagery are investigated as key variables for damage classification. We found that each of the algorithms achieved nearly a 90% wide area damage density match using the United Nations Operational Satellite Applications Programme (UNITAR/UNOSAT) dataset as validation. The most effective algorithm achieved an error rate below 40% in detecting damaged buildings. Spatial features of texture and structure were far more important in classification than spectral information, highlighting the potential for future implementation of machine learning algorithms which use panchromatic or pansharpened imagery alone. This would reduce data requirements and allow response resources to be allocated over space quicker and more efficiently.

The insurance industry relies on both commercial and in-house software packages to quantify financial risk to natural hazards. For earthquakes, the initial loss estimates from the industry’s catastrophe risk (CAT) models are based on the probabilistic damage a building would sustain due to a catalog of simulated earthquake events. Based on the occurrence rates of the simulated earthquake events, an exceedance probability (EP) curve is calculated, which provides the probability of exceeding a specific loss threshold. Initially these loss exceedence probabilities help a company decide what insurance policies are most cost efficient. In addition they can also provide insights into loss predictions in the event that an actual natural disaster takes place, thus the insurance company is prepared to pay out their insured parties the necessary amount. However, there is always an associated uncertainty with the loss calculations produced by these models. The goal of this research is to reduce this uncertainty by using Bayesian inference with real time earthquake data to calculate an updated loss. Bayes theory is an iterative process that modifies the loss distribution with every piece of incoming information. The posterior updates are calculated by multiplying a baseline prior distribution with a likelihood function and normalization factor. The first prior is the initial loss distribution from the simulated events database before any information about a real earthquake is available. The crucial step in the update procedure is defining a likelihood function that establishes a relative weight for each simulated earthquake, relating how alike or dislike the attributes of a simulated earthquake are to those of a real earthquake event. To define this likelihood function, the general proposed approach is to quantify real time earthquake attributes such as magnitude, location, building tagging and damage, and compare them to an equivalent value for each simulated earthquake from the CAT model database. In order to obtain the simulated model parameters, the catastrophe risk model is analyzed for different building construction types, such as steel and reinforced concrete. For every model case, the loss, peak ground acceleration per building and simulated event magnitude and locations are recorded. Next, in order to calculate the real earthquake attributes, data was collected for three case studies, the 7.1 magnitude 1997 Punitaqui, the 8.8 magnitude 2010 Chile earthquake and the 6.7 magnitude 1994 Northridge earthquake. For each of these real earthquake events, the magnitude, location, peak ground acceleration at every available accelerometer location, building tagging and qualitative damage descriptions were recorded. Once the data was collected for both the real and simulated events, they were quantified so they could be compared on equal scales. Using the quantified parameter values, a likelihood function was defined for each update step. In general, as the number of updates increased, the loss estimates tended to converge to a steady value for both the medium and large event. In addition, the loss for the 6.7 and 7.1 event converged to a smaller value than that of the 8.8 event. The proposed methodology was only applied to earthquakes, but is broad enough to be applied to any type of peril.

In this thesis, stochastic methods are utilized to improve a familiar comprehensive probabilistic model to obtain realistic estimates of the earthquake insurance premium rates in different seismic zones of Turkey. The model integrates the information on future earthquake threat with the information on expected earthquake damage to buildings. The quantification of the future earthquake threat is achieved by making use of the seismic hazard analysis techniques. Due to the uncertainties involved, the hazard that may occur at a site during future earthquakes has to be treated in a probabilistic manner. Accessibility of past earthquake data from a number of different data sources, encourages the consideration of every single earthquake report. Seismic zonation of active earthquake generating regions has been improved as recent contributions are made available. Finally, up-to-date data bases have been utilized to establish local attenuation relationships reflecting the expected earthquake wave propagation and its randomness more effectively. The damage that may occur to structures during future earthquakes involves various uncertainties and also has to be treated in a probabilistic manner. For this purpose, damage probability matrices (DPM), expressing what will happen to buildings, designed according to some particular set of requirements, during earthquakes of various intensities, are constructed from observational and estimated data. With the above considerations, in order to demonstrate the application of the improved probabilistic method, earthquake insurance premium rates are computed for reinforced concrete and masonry buildings constructed in different seismic zones of Turkey.

Thesis (Ph. D.)--University of Washington, 2006. / Vita. Includes bibliographical references (leaves 355-372).

There is a great amount of building stock built in earthquake regions where earthquakes frequently occur. It is very probable that such buildings experience earthquakes more than once throughout their economic life. The motivation of this thesis arose from the lack of procedures to determine the change in building capacity as a result of prior earthquake damage. This study focuses on establishing a method that can be employed to determine the loss in the building capacity after experiencing an earthquake. In order to achieve this goal a number of frames were analyzed under several randomly selected earthquakes. Nonlinear time-history analyses and nonlinear static analyses were conducted to assess the prior and subsequent capacities of the frames under consideration. The structural analysis programs DRAIN-2DX and SAP2000 were employed for this purpose. The capacity curves obtained by these methods were investigated to propose a procedure by which the capacity of previously damaged structures can be determined. For time-history analyses the prior earthquake damage can be taken into account by applying the ground motion histories successively to the structure under consideration. In the case of nonlinear static analyses this was achieved by modifying the elements of the damaged structure in relation to the plastic deformation they experience. Finally a simple approximate procedure was developed using the regression analysis of the results. This procedure relies on the modification of the structure stiffness in proportion to the ductility demand the former earthquake imposes. The proposed procedures were applied to an existing 3D building to validate their applicability.

When an earthquake happens, one of the most important tasks of disaster managers is to conduct damage assessment; this is mostly done from remotely sensed data. This study presents a new method for building detection and damage assessment using high-resolution satellite images and LiDAR data from Port-au-Prince, Haiti. A graph-cut method is used for building detection due to its advantages compared to traditional methods such as the Hough transform. Results of two methods are compared to understand how much our proposed technique is effective. Afterwards, sensitivity analysis is performed to show the effect of image resolution on the efficiency of our method. Results are in four groups. First: based on two criteria for sensitivity analysis, completeness and correctness, the more efficient method is graph-cut, and the final building mask layer is used for damage assessment. Next, building damage assessment is done using change detection technique from two images from period of before and after the earthquake. Third, to integrate LiDAR data and damage assessment, we showed there is a strong relationship between terrain roughness variables that are calculated using digital surface models. Finally, open street map and normalized digital surface model are used to detect possible road blockages. Results of detecting road blockages showed positive values of normalized digital surface model on the road centerline can represent blockages if we exclude other objects such as cars.

This study mainly addresses the problem of estimating the prior earthquake damage on the response of reinforced concrete structures to future earthquakes. The motivation has arisen from the heavy damages or collapses that occurred in many reinforced concrete structures following two major earthquakes that recently occurred in the Marmara Region, Turkey. The analysis tool employed for this purpose is the package named IDARC2D. Deterioration parameters of IDARC&#039 / s hysteretic model have been calibrated using a search method. In the calibration process experimental data of a total of twenty-two beam and column specimens, tested under constant and variable amplitude displacement histories, has been used. Fine-tuning of deterioration parameters is essential for more realistic predictions about inelastic behavior and structural damage. In order to provide more realistic damage prediction, three ranges of parameters are proposed. Some damage controlling structural parameters have been assessed via a large number of two-dimensional section analyses, inelastic time history and damage analyses of SDOF systems and seismic vulnerability analyses of reinforced concrete buildings. Inelastic time history and damage analyses of numerous SDOF systems have been carried out to determine whether the loading history has an effect on damage and dissipated hysteretic energy. Then this emphasis is directed to the analyses of MDOF systems. In the analyses of the SDOF systems, various forms of constant and variable amplitude inelastic displacement reversals and synthetic ground motions composed of one of the four earthquake records preceded or followed by its modified records acted as a prior or successive earthquake, have been used. The analyses of two five-story R/C buildings have been caried out using synthetic accelerograms comprised of base input provided by the two recorded ground motions. It is shown that both damage progression and cumulative hysteretic energy dissipated along a path seem to depend on the number and amplitude of cycles constituting the path. However, final damage and accumulated hysteretic energy dissipated along a loading path are independent of the ordering of the same number and amplitude cycles along the path. There is a nonlinear relationship between the earthquake excitation intensity and final damage attained in the end. Increase in the acceleration amplitude leads to exponential increase in damage. As the prior earthquake intensity increases the damage from the succeding main earthquake decreases. A definite ground motion acting as prior and successive earthquake causes substantially different amount of damage. Prior earthquake damage does not substantially affect the maximum drift response in future larger earthquakes. A MDOF frame type structure with aprior damage suffers less overall damage in an earthquake in comparison with the one without a prior damage.