ABSTRACT:

Recently developed methods for wave propagation analysis termed as SIORM and NIOM are compared using earthquake ground motion records of Samukawa vertical array in Japan. Three components of the earthquakes M6.5 of February 20, 1990 and M5.9 of February 2, 1992 are used in this analysis. For each component, P- and S- portions are analyzed separately at five elevations of the vertical array. Both methods provide identical models of wave propagation and illustrate stable peaks corresponding to propagation of Pand S- wave. The incident phase of P- and S- wave and its reflection from the ground surface are clearly depicted. Amplification of wave in shallow layers is observed in the results of both methods. Moreover, the ratio of the amplitudes of incident and reflected wave models provides some idea about damping in shallow ground layers.

INTRODUCTION

Among the methods used for evaluating the ground motion variations, cross-correlation has been the most common. However, the shape of cross-correlation function is greatly influenced by the wave propagation properties. The resolution of cross-correlation peaks may be poor for actual earthquake ground motion records when the concentration of power is within small time lags (Haddadi and Kawakami, 1998). The recently developed methods of simplified input output relation method, SIORM, (Kawakami & Bidon, 1997) and normalized input-output minimization method, NIOM, (Kawakami and Haddadi, 1998; Haddadi and Kawakami, 1998) give simple models of wave propagation and are shown to be more reliable than conventional correlation and impulse response functions. The methods are effective in detecting the arrival times of incident and reflected waves and in showing the amplification property of shallow layers. This paper applies SIORM and NIOM methods to strong ground motion records of Samukawa vertical array in Japan and compare the models of wave propagation obtained by those methods.

This content is only available via PDF.
You can access this article if you purchase or spend a download.