Probability Modelling Of PGA Amplification Factor Corrected By Site Conditions

Peak ground-motion acceleration?PGA?is a fundamental parameter in seismic design codes,directly representing the seismic loading in the aseismic design for engineering structures.Moreover,PGA also plays an important role in earthquake early warning system and fast seismic intensity evaluation.According to seismic damage investigation,seismic wave can be affected by shallow soil layers through which it propagates,which mainly shows amplification and filtering effect on the seismic wave components of different frequencies.Therefore,even in the area with similar seismic source distance,the ground motion and PGA can be different that the distribution of seismic damage will be different.Therefore,estimating PGA of a specific site is necessary for the aseismic design engineering.With the concepts of risk assessment and probabilistic seismic risk analysis in engineering are widely accepted,PGA prediction should made under probability frame.Nevertheless,such research has not been much reported in China and abroad.To solve the problems above,tens thousands of earthquake records,including surface and downhole accelerograms,recorded by 40 strong-motion stations from Ki K-net seismic station array in Japan are collected.The statistics of PGA amplification factor f _PGAare systematically studied that a demonstration that f _PGAbasically is lognormally distributed is made.The correlations between mean and standard deviation of f_PGA with respect to PGA_R,which is the peak acceleration value of downhole acceleration records,are established.The mean and standard deviation of f _PGA depend on site characteristic parameters which can be used to determine the regressive coefficients for the proposed probability models of f _PGA.Following the f _PGA probability models,ground surface PGA corrected by specific site characteristic parameters can be predicted under different probability levels.The main work and salient results can be outlined as follows.?1?Using the Japanese KiK-net strong-motion array,40 strong-motion stations are selected and the seismic records recorded by these stations are collected according to proposed data selection standards.The parameters such as VS30,VS20,VSe and soil depth D,which are used in site classification in seismic design codes in China and other countries are chosen to represent site characteristics.The distribution of the site characteristic parameters for the selected stations are illustrated.?2?After defining the PGA amplification factor f _PGA,the probabilistic distribution of f _PGA for the selected stations are carefully studied that a lognormally distribution feature of f _PGA is delineated with respect to a specific ground shaking intensity input.That is,a log-normal probability density function can be used to simulate f _PGA.A linear correlation of mean and standard deviation of f _PGA depending on log?PGA_R?is proposed.?3?The regressive coefficients of the relationships between the mean and standard deviations of f _PGA and log?PGA_R?are found to be closely dependent on site characteristic parameters.However,an individual site characteristic parameter like V_S30,V_S20,V_Se or soil thickness D,poorly correlates with the statistic parameters,while satisfactory correlations can be drawn with respect to linear combinations of V_S30,V_S20,VSe or soil thickness D.Regressing on the data,the statistic parameters of f _PGA are calculated according to linear combination of site characteristic parameters so as to build the probability density function of the log-normal distribution model of f _PGA.?4?A probability model of f_PGA is proposed.Using the f _PGA probability model,the ground surface PGA can be predicted under different probability levels for a specific site.The feasibility,reliability and reasonability of the PGA probability model are testified using recorded ground motions.By the calculation framework,the predictions of ground surface PGA,corrected by site characteristic parameters,under different probability levels are made to meet the demanding of risk analysis in engineering practice,which paves the way for site-corrected surface PGA predictions in earthquake precursory system and fast seismic intensity scale estimation.