The Method Research Of Probabilistic Seismic Hazard Analysis Based On Adaptively Smoothed Seismicity Model And 3-Dimensional Fault Model

As probabilistic seismic hazard analysis(PSHA)directly provides seismic design parameters for engineering constructions,its importance is self-evident.The key scientific problem of relevant research lies in,how to optimize parameters and models to better depict the pattern of long-term seismic activity and the characteristics of earthquake damage,thus to improve the reliability of results.In this research area,the methods and technical ideas currently used in China are significantly different from the method based on smoothed seismicity and3-dimensional(3-D)fault model proposed by United States Geological Survey(USGS),and there are apparent differencies in characteristics and applicabilities between the two methods.Therefore,I began to study advanced theories and technologies abroad under instructions of advisors,and apply the methods on domestic research areas and make comparisons with that used in China.Based on this,I try to make some improvements and innovations,and have made some research progresses and achievements.Around on the scientific problem of‘optimizing parameters and models to improve the reliability of results',detailed works are as follows:at first,systematic research has been performed on the most commonly used seismic activity parameter,b-value,and the results of which will be directly used on subsequent seicmic activity models.Based on this,the two most important models employed in the study,adaptively smoothed seismicity model and 3-D fault model have been built,and special attention has been paid on improvement of the former.Combined with the NGA West2 attenuation relationships,these models are used to perform PSHA on Namcha Barwa and surrounding region.Additionaly,our procedure was compared with that used in domestic,and the characteristics of the two methods are summarized.The last part of this dissertatio applied the results of PSHA to analysis landslide stabilility,which belongs to interdiscipline.All my works in this study are associated with the theme of PSHA,although the connection from different parts would be varying.What's more,PSHA is a kind of method and technology,and thus must be universal and repeatable,so my study lays emphasis on method research and is not limited on one research area,which is different from description of geologic features.Details are as follows:(1)The maximum likelihood estimation of b-value in magnitude-frequency relation and analysis of its influence factors:Chapter 2 focuses on the estimation of b-value and services as basis for follow-up studies.B-value in the magnitude-frequency relationship(G-R)is one of the most important parameters in seismicity study and seismic hazard analysis,and the most commonly used techniques to simulate it are the least square approach and the maximum likelihood method.Least square method is simple and easy to apply,thus widely used in China.However,many researches show that there exist some limits in this technique.Therefore,under certain conditions,the maximum likelihood method is recommended as an effective substitution or supplementary.Previous researchers have provided many equations on maximum likelihood estimation of b-value,based on varying implicit assumptions and different ways of solution.I first presented a brief overview for these equations,then provided classification and summary depending on taking account of following factors or not:the effect of binned magnitude,finite maximum magnitude,using unequal observation periods for different magnitude intervals,and analytic solution.Following this,we analyzed six influential factors:binning magnitude,measurement errors of magnitude,sample size,magnitude span,minimum completeness magnitude and fore-and aftershocks.At last,we provide reasonable suggestions for properly using these equations.This study contributes to more accurately understand and use different formulas of estimating b-value by maximum likelihood technique,which could be used as reference for peers.(2)The improvement and application of the adapatively smoothed seismicity modelIn chapter 3,I try to make some improvements on the adapatively smoothed seismicity model proposed by Helmstetter et al.(2007),and apply it on long-term probabilistic forecast for M?5.0earthquakes in the eastern Tibetan Plateau.The main improvements focus on two aspects:First,I divide the study region into grids and then estimate seismic parameters for every grid,thus reflecting the spatial variation of these parameters better.Second,earthquake records are used by dividing magnitude intervels,and for every interval,the initial time of high quality catalog is estimated accordingly.At last,I systematically tested the appearance of models with different input parameters.Results show that the number of events occurred in the future could be perfectly forecasted by our models.Either with longer time-span of input catalog or with smaller target events,the forecast could be improved.For target earthquakes?M 5.0,Gaussian kernel performs slightly better than Power-law.In comparison with models that simply select events?M 3.5,the approach using multiple magnitude intervals does have better performance,although the advantage is not apparent in our study area.Possible reasons are the low quality and short time duration of the input earthquake catalog,resulting in a very limited increase of input events.In contrast with models employing space uniform parameters,our models significantly enhance the forecasts.(3)Probabilistic seismic hazard analysis based on adaptively smoothed seismicity model and3-D fault model:Chapter 4 takes the zoning map of Namche Barwa and its surrounding area as an example,to introduce the method and theory of probabilistic seismic hazard analysis based on adapatively smoothed seismicy and 3-D fault model.The study area situates on the collision zone between the Indian and Eurasian plates,suffering from not only ordinary shallow crust earthquakes but also the subduction composed by Himalaya Main Forward Fault(MFT).Based on the characteristics of different seismic sources,five models were used in this study:adapatively smoothed seismicy model,smoothed seismicy model with fixed radius,background model,3-D fault model,and 3-D subduction model.What's more,as 3-D models must employ suitable attenuation relations,traditional equations using attenuation distances of epicenter or focal distance are not useful.Therefore,five relations from NGA West2 are applied for the first four models,and four newest relations are used for the subduction model.At last,I compared and analyzed the results and sumarized the features of our method.Results show that 3-D fault model is able to show the characteristics of active faults on seismic hazard analysis very well,such as faults'contribution on different probabilities of exceedance,and the hangwall-effect of thrust faults.(4)Application of the results of probabilistic seismic hazard analysis on the landslide stability analysis:Landslide has been one of the most powerful secondary disasters of earthquakes,whereas the results of probabilistic seismic hazard analysis(PSHA)usually represent the comprehensive assessment of the potential seismogenic ability of a place.Chapter 5 tries to apply the PSHA results to landslide stability analysis,thus used as guidance for landslide resistant design and engineering construction in this area.The 2015 M_w 7.8 Nepal earthquake caused severe damage to Tatopani Border Station and the large paleo-landslide on the west of it.The station situates right on the deposition of the toe of the paleo-landslide,thus is very vunerable to future landslide or debris flow.Based on detailed fieldwork in the geological disasters around the paleo-landslide,we proposed an evolutionary model for it.At last,several groups of PSHA results of 10-percent probability of exceedance in 50 years were inputed to GeoStudio,to simulate the stability of the paleo-landslide under different seismic motion inputs.Results show that the paleo-landslide is stable without effect of earthquake.However,when the peak ground acceleration(PGA)exceeds0.35 g,the landslide will be reactivated and produce large newborn landslides.Our study could be used as important reference for earthquake resistant design of the border station and other infrastructural development along the Bhote Koshi Valley.From the foregoing,this study successfully built a complete technical system of probabilistic seismic hazard analysis based on adapatively smoothed seismicy and 3-D fault model for the first time in China.The main part of the approach was borrowed from the National Seismic Hazard Map of United States,but I have made some innovations and improvements technically.This study has a very good prospect on the utilization of the products of urban active fault detection in China,and is hugely useful on plenty of projects abroad.What's more,this work also contributes to the technical system of seismic hazard assessment in China.