Simultaneous Algebraic Reconstruction Technique (SART) for Retrieving Shear-Wave Quality Factor Qs Profiles Using Seismic Noise

Session: Near-Surface Effects: Advances in Site Response Estimation and Its Applications
Type: Oral
Date: 4/28/2020
Time: 04:30 PM
Room: 110 + 140
Description: 

Over the last decades seismic noise analysis have gained a significant role in the estimation of the local seismic response. This is due to rapidity of data acquisition, non-invasiveness and low costs this type of computation can offer.

Shear-wave velocity (V_s) and fundamental resonance frequency of sedimentary covers are the parameters generally estimated by seismic noise analysis. V_s, in the specific, is mandatorily required by European and national seismic design codes. However, a more comprehensive information about site response should also include the characteristics of seismic wave attenuation and, in particular, of the shear-wave quality factor (Q_s).

Recently, attempts of retrieving Q_s from ambient noise have been carried out, for example, by Prieto et al. (2009) and Weemstra et al. (2013).

Parolai (2014) proposed an attenuation factor inversion for retrieving the 1D Q_s structure below a site, once the V_s profile is derived from the inversion of dispersion curves. The solution achieved by this approach is based on a Least Squares algorithm with positivity Q_s constraint (e.g. Menke, 1989). Although giving valuable results for the shallowest geological layers, it requires careful trial and error testing of the damping factors for obtaining robust solutions.

In order to improve this inversion, we decided to test Simultaneous Algebraic Reconstruction Technique (SART) as an alternative to the linear inversion method used in Parolai (2014). SART is an iterative reconstruction algorithm. Based on weighted sums of rows and columns of the input data kernel matrix, SART offers the advantage of both an easy implementation and succeeding in fixing a positivity Q_s constraint to the solutions. In this work we present the results of applications of the proposed approach to real test cases and compare them with the Q_s values previously estimated.

Presenting Author: Ilaria Dreossi

Authors