Processing of near-field earthquake accelerograms

The Near-Field pulse-like velocity and displacement time histories associated with a strong earthquake can greatly affect a wide range of different types of structures [I]. The important long period components in NEAR-FIELD earthquake accelerograms, however, are normally eliminated or distorted by conventional data processing which is based on band-pass filtering; for example, using an Ormsby filter with a low frequency cut-off of 0.2-0.4 Hz. A new data processing technique has been proposed by Iwan and Chen [2] to recover the long period components from NEAR-FIELD earthquake accelerograms. This technique is based on the inverse of the data recording and retrieving procedure, which includes appropriate instrument correction according to the instrument type, and baseline correction without band-pass filtering. It may be briefly summarized by the following steps: I .Apply a least-mean-square linear fit to the uncorrected accelerograms to eliminate any uncertainty in the instrument centering. 2.Apply instrument correction to compensate for the fact that the transfer function of the transducer is not flat over the entire frequency band. 3.Integrate the instrument corrected acceleration time history to obtain a raw velocity time history, assuming zero initial velocity. A trapezoidal integration rule and a central difference differentiation scheme are used herein. 4.Apply a segmented polynomial baseline fit to the raw velocity time history to remove any non-physical trends. Since the ground velocity physically begins at zero and ends at zero, the baseline is fitted to the initial and final portions of the raw velocity time history. These two polynomials are connected by the lowest order (smoothest) polynomial baseline connection that continuously connects the initial and final portions of the accelerogram. The objective of baseline correction is to diminish the long-period noise or drift introduced in the signal recording and playback process. 5.Integrate the baseline corrected velocity time history to obtain a displacement time history, assuming zero initial displacement. Differentiate the baseline corrected velocity time history to obtain the corrected acceleration time history. A Matlab package has been written to fulfill this data processing procedure. The package is also able to plot acceleration, velocity, and displacement time histories, to plot horizontal displacement particle trajectories, and to compute and plot Response Spectra and Drift Demand Spectra [3]. There are a total of 15 Matlab routines (functions) in this package. Sample Matlab programs are contained in Appendix 1. SAMPLELM and SAMPLE2.M demonstrate the usage of the routines. Appendix 2 lists all routines. A floppy disk containing all routines is included with this report