Þann 10. maí kl 14:00 í stofu 157 í VR-II mun Milad Kowsari flytja fyrirlestur um rannsóknir sínar við Umhverfis- og byggingarverkfræðideild Háskóla Íslands.
Fyrirlesturinn ber heitið Dvínun jarðskjálftahreyfinga og áhrif á á jarðskjálftahættu á Norðurlandi (e. Sensitivity of Earthquake Hazard Estimates for the Tjörnes Fracture Zone in North Iceland).
Fyrirlesturinn verður haldinn á ensku.
Milad Kowsari er doktorsnemandi á 3. ári við umhverfis- og byggingarverkfræðideild Háskóla Íslands. Leiðbeinandi hans er Dr. Benedikt Halldórsson, vísindamaður og forstöðumaður rannsókna hjá Rannsóknarmiðstöð í jarðskjálftaverkfræði við umhverfis- og byggingarverkfræðideild Háskóla Íslands, og sérfræðingur í tæknilegri jarðskjálftafræði á úrvinnslu- og rannsóknasviði Veðurstofu Íslands. Í doktorsnefnd Milad sitja enn fremur Dr. Sigurjón Jónsson prófessor í jarðeðlisfræði við King Abdullah University of Science and Technology í Saudi Arabíu, Dr. Birgir Hrafnkelsson prófessor við Raunvísindadeild Háskóla Íslands og Dr. Jónas Þór Snæbjörnsson, prófessor við Háskólann í Reykjavík.
Útdráttur á ensku
Ground-motion models (GMMs) describe various aspects of earthquake ground motion as a function of earthquake magnitude, site-to-source distance, local site conditions and other seismic parameters. Typically, empirical GMMs have been fitted to the recorded data by different regression methods such as the ordinary least-squares, the weighted nonlinear least-squares, the two-stage maximum-likelihood and the random-effects regression methods. However, these regression methods are not able to decrease the bias introduced by the uneven distribution of recordings in regions with sparse data. In many cases, the information contained in the recorded ground-motions is also not sufficient to properly constrain all regression coefficients for a given functional form. To this end, the Bayesian statistical method can be applied to combine prior information of the model parameters with the likelihood of the observed data, summarized in a posterior density. This feature has made it beneficial for the estimation of GMMs in regions like Iceland where due to both limited number of recordings and their narrow earthquake magnitude ranges, magnitude-and-distance dependent saturation terms cannot be constrained by data. Moreover, to account for uneven sampling of the different earthquakes and correlations of the recorded ground-motion from a single event, the aleatory variability is partitioned into inter-event and intra-event components. In this study, a Bayesian random effects model that uses a Markov Chain Monte Carlo algorithm for inference is presented to recalibrate different GMMs. In addition to the inter- and intra-event variability, the correlation coefficients between the model parameters are assessed. The recalibrated GMMs fit the recorded data very well and, due to the uncertainties and correlations between the model parameters now being quantified, they can be used in future seismic hazard analyses in Iceland with much greater knowledge of the quality and confidence in the results.
Um nemandan á ensku
Mr. Kowsari received a B.Sc. degree in Civil Engineering (2010) from Azad University of Bandar Anzali and M.Sc. in Civil Engineering (2012) with focus on Earthquake Engineering, from the University of Kurdistan, Iran where he was awarded the Elite Prize for the best researcher in 2011. His Masters thesis focused on applying Bayesian procedures for the seismic hazard analysis of the metropolitan area of Tehran, Iran. His thesis work and collaborations after his graduation have led to three publications and an additional paper under review in ISI accredited international scientific journals, in addition to several national conference papers, presentations and also an academic book related to seismic hazard analysis. He has excellent academic record in addition to undergraduate teaching experience.