Estimation of signal characteristics in seismic ground motions

In this study it is investigated whether signal processing techniques and least-square minimization methods applied to seismic data can identify the deterministic signal components that contribute significantly to the motions. The identification of these deterministic components would then allow the treatment of the remaining part of the time histories as random and result in better descriptions of the stochastic ground motion characteristics. This paper describes the methodology and applies the approach to analytically evaluated seismic ground motions at an array of sensors. Commonly used signal processing techniques are briefly described and compared, and an efficient least-square minimization scheme is introduced. The analysis indicates that the signals can be identified with their frequency, wavenumber, amplitude and phase.