SITE EFFECTS BY H/V RATIO: COMPARISON OF TWO DIFFERENT PROCEDURES

In this article, H/V ratio are evaluated using different spectral techiniques applied to both earthquake and microtremor data (Nakamura technique). In particular, in order to avoid numerical instability, two different numerical techiniques are taken into account: (a) a smoothing procedure applied to the spectra of the seismogram components and (b) a regularization method applied to the H/V ratio (Landweber scheme). The data set consists of more than 70 earthquake events recorded by three component sensors displaced in the town of Fabriano (Central Italy) during the Umbria-Marche sequence started on September 1997. The local magnitudes range between 2.7 and 4.4, while the epicentral distances range between nearly 30 and 60 km. The stations were set to continuous recording so that a huge amount of microtremors was stored. The results are compared in terms of predominant frequencies and amplification levels in order to point out the influence of the adopted methods. The H/V ratio provides similar results if applied to a smoothed version of both earthquake and microtremor spectra, confirming that Nakamura technique is a cheap and a fast method to collect information on the site amplification effects. Moreover, the results relevant to earthquake data seem not to depend on the method used to stabilize the H/V ratio, whereas those relevant to microtremor data does. The explanation of this fact is suggested by the behaviour of the Landweber filter showing that the predominant frequency detected by means of microtremor data lies in a high instability region of the spectra.