NCEER-90-0009 | 1/30/1990 | 66 pages |$10
TOC: The table of contents is provided.
Keywords: Computer Codes, Modeling, Ground Motion, Regression Analysis, Autoregressive Moving Average Techniques, Time Series Histories, Seismic Loading, and Accelerograms.
Abstract: This paper presents the final report of a study on the modeling of earthquake ground motion in seismically active zones. The first part of the study focuses on the use of parametric time series methods to model earthquakes. The modeling parameters used in the analysis of the accelerograms are related to physical variables such as earthquake magnitude, epicentral distance and site geology and a set of regression relations derived for the region. This enables the simulation of site-faithful time histories for the regions without ground motion records. The modeling of the earthquakes is done using a multivariate Autoregressive Moving Average process. The second part of the study deals with the extension of this procedure to model multiple event earthquakes. Studies are conducted on multiple event earthquakes in Mexico and Japan and suitable regression relations between the modeling parameters and the site conditions are developed. This report also includes an alternative procedure to model the time series based on the digital simulation of the power spectral density function. The fundamental advantage of these parameters on these procedures is that the time invariant parameters for both the power spectral density and the ARMA models can be easily related to physical variables at the site. Further, confidence intervals for the modeling parameters can be used to generate an ensemble of earthquakes corresponding to the mean, mean plus one standard deviation etc. for design purposes.