Case Study Of Probabilistic Seismic Hazard Analysis Method Based On Multi-model

One of the focused issues in current seismology is how to estimate the seismichazard level of a region usingreasonable and effective probabilistic seismic hazardanalysis (PSHA) method. This method is based on multi-model, which assimilatesplenty oflatest research results of related fields, has beenwidely accepted and appliedin the word. This thesis attempts to make a preliminary systematic research on thismethod by the combination of theoretical and case analyses.They include processingand analysis of the earthquake catalogs, determining seismicity parameters, seismicitymodels, magnitude distribution models, choice of attenuation relationship, correctionof uncertainty and seismic hazard analysis. In addition to introducing this method, thiswork alsotries to make exploration in some aspects.The contents of the thesis aresummarized below.(1) Preparation of a unified catalog in moment magnitudethrough magnitudeconversion relationshipIn seismic hazard analysis by the probabilistic method,a spatially smoothed modeland background model must be established based on earthquake catalogues. Atpresent, such data are usually involved in different magnitude scales describing sizesof events. Since the attenuation relationship this work uses is based on momentmagnitude, a homogeneous catalog in moment magnitudeshould be prepared inadvance. This work downloads the seismic data during period1907-2012from theInternational Seismological Center (ISC), containing the earthquake magnitudes ofdifferent scales released by a variety of sources. This studytakesMW,GCMTas thereference magnitude, then a host of empirical relationships between magnitude scalesfrom other institutions with MW,GCMTare derived through the least squares technique.The process, for choosing magnitude scale to a seismic event, follows a rule that givespriority to the one with a larger correlationcoefficient and smaller residual standarddeviation in relevant regression relationship. Indirect conversion relationships andrelationships derived from global seismic data serve as supplement when lacking ofdirect conversion relations or the level of goodness of fit is too poor.Consequently,this work successfully gets a homogeneous catalog in momentmagnitude for study area. (2) Analysis of magnitude completeness (MC) by combination of three methodsA spatial and temporal windowing method is applied to remove foreshocks andaftershocks, thenthe study area is divided into5smaller zones on the basis of spatialdifference in tectonic setting and seismicactivity. The analysis of magnitudecompleteness (MC) for each zone is done by the combination of three methods:cumulative number of earthquakes plotted against time, maximum curvature (MAXC)and goodness of fit text (GFT). The seismicity parameters of zones are calculated withthe maximum likelihood method on the basis of completeness analysis of the catalogat last. Results indicate that though each method has its own advantages anddrawbacks, the optimum MCcan be obtained when it is used in combination with othermethods. Graphs of time and cumulative number of earthquakes can be used toidentify the starting time of the high-quality catalog, which is the basis of the MAXCand GFT, but it could be easily affected by seismicity fluctuationsover time. Becauseof poor data quality in the study region, the error of MAXC is so big that it can onlybe used as supplement for two other methods and the fit-level of GFT can be as largeas approximately80%in zones. Although the maximum of GFT may not correspondto MC, it can be always obtained from the extreme point of GFT. The MCof5zonesduring the period of1964-2012generally occurs aroundMW4.8, and the b value variesfrom1.136to1.514.(3) The PSHA method based on the multi-modelConsideringthe characteristics of different sources, this work uses three kinds ofmodels to describe the seismic hazard of the study area, which are the spatiallysmoothed model, background model and characteristic model. Through reasonablecombination, two sets of schemes withreasonable weights are formed by themodels.Then the final results of seismic hazard analysis are obtained. During theprocess of calculation, thiswork chooses the attenuation relationship of NextGeneration of Ground-Motion Attenuation Models (NGA) issued by the US and takesfull account of uncertainties of the fault parameters and the attenuation relationship ina reasonable way. Because the method needs not to divide potential sources, so theuncertainty of it can be successfully avoided and the seismic hazard of faults wouldnot be diluted at the same time.(4) Analysis of the results of seismic hazard evaluationThe results shows that the PSHA method based on multi-model can well reflectthe seismic activity level and seismic hazard level of the study area. From seismic zoning, the peak ground acceleration (PGA) generally exhibits azonal distributionalong the fault strike, and PGA diminishes in the direction perpendicular to the faultstrike. The characteristic and effect of the fault source are fully expressedin theseismic hazard analysis. Comparing the contributions of the grid source and the faultsource to the exceeding probabilityfinds that when PGA is low,the largest contributionto exceed probability is the grid source, next is the fault source which is close to thesite. But with the increase of PGA, the contribution of the fault source becomesincreasingly prominent. When PGA increases to a certain degree, the fault closest tothe site becomes the dominant factor in probability contribution. The above rule is inline with the understanding of different seismogenic structures and meets the physicallaw of ground motion attenuation with distance.（5）Development of the seismic risk analysis programThe PSHA method based on multi-model, which is composed by lots of modules,has a very complex theory system. This work successfully develops the relatedprogram on the platform of mapbasic under the guidance of experts and scholars. Thenew program has been applied to the seismic hazard assessment of some majorprojects abroad, and the results have been approved by the national earthquake safetyevaluation committee. Though the program is still not perfect, it would have someapplication values as it canbe improved with reference to the latest internationaladvancesin this aspect.This thesis presents theoretical and case studies of the PSHA method based onmulti-model.The results have certain reference values for improvementof probabilisticseismic hazard analysis in our country.