GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 2.1 191 EARTHQUAKE SOURCE PARAMETERS ESTIMATED FROM DIRECT S-WAVES AND CODA ENVELOPES ANALYSIS PERFORMED IN CENTRAL ITALY D. Bindi 1 , P. Morasca 2 , D. Spallarossa 3 , M. Picozzi 4 , K. Mayeda 5 , J. Roman-Nieves 5 , J. Barno 6 , W.R. Walter 6 1 GFZ, Potsdam, Germany 2 INGV, Milan, Italy 3 UNIGE, Genoa, Italy 4 University of Naples Federico II, Italy 5 AFTAC, Patrick AFB, FL, USA 6 LLNL, Livermore, CA, USA The assessment of source parameters over a wide magnitude range is a fundamental issue in understanding the relationships between small and large events. However, effects of seismic wave propagation in attenuating media and limitations in the monitoring configuration, hamper our ability to discriminate between physical contributions to the observed source variability from uncertainties of the methods and data used to estimate the values. Moreover, since the impact of anelastic attenuation increases with increasing source-to-site distance and, alongwith near-surface attenuation effects, filters more severely high frequencies, the estimation of high-frequency source parameters is particularly challenging for small events (Abercrombie, 1995). Several studies investigated the limitations introduced by attenuation effects and by the limited available bandwidth on the estimation of source parameter for small earthquakes (e.g., Kwiatek and Ben- Zion, 2016; Abercrombie et al., 2017). Recently, Bindi et al. (2020) investigated the reliability of the source parameters estimated in central Italy for magnitudes in the range from 1.5 to 6.5, using synthetic spectra generated for the same source-station geometry of an actual data set. The large redundancy in the earthquake and station sampling allowed for retrieving reliable source parameter down to magnitude about 1.8, with the near-surface attenuation parameter k 0 being the most impacting parameter for the determination of the corner frequency of small events. Various approaches have been proposed throughout the years to estimate source parameters, such as empirical Green’s function analysis based on direct waves to remove path and site effects (Prieto et al. , 2004; Ide et al. , 2003; Abercrombie, 2015) or coda waves (Mayeda and Walter, 1996; Mayeda et al. , 2003, 2007, Walter et al. , 2017), and data-driven spectral amplitude decomposition approaches (e.g., Andrews, 1986; Edwards et al. , 2008; Malagnini et al. , 2011; Oth et al. , 2011; Trugman and Shearer, 2017). Several investigations (e.g., Eken et al. , 2004; Gök et al. , 2016; Morasca et al. , 2005; Yoo and Mayeda, 2013; Holt et al. , 2021; Shelly et al. , 2022) demonstrated the ability of methods based on coda envelope measurements to obtain stable source measurements over a large frequency range thanks to the stable properties of coda waves (e.g., Aki, 1969; Aki and Chouet, 1975; Mayeda and Malagnini, 2010). The low sensitivity to source and path heterogeneity of coda waves, in fact, allows the assumption of simple 1D models to adequately describe a region using local to regional events recorded from as few as one station (Mayeda et al. , 2003). The Central Italy region is ideal for comparing the outcomes of different approaches as it is characterized by high-quality data including recent well-recorded seismic sequences such as L’Aquila (2009) and Amatrice-Norcia-Visso (2016-2017). Morasca et al. (2022) applied the coda method by Mayeda et al. (2003) using the Coda Calibration Tool (CCT), a freely available Java- based code (https://github.com/LLNL/coda-calibration-tool) to obtain a regional calibration for Central Italy for estimating stable source parameters. For validating the source parameters through a comparison of the results achieved when different techniques are applied to different portions of the seismogram, the authors also performed a spectral amplitude decomposition (GIT, Generalized Inversion Technique) and investigated the strengths and limitations of the two methodologies. A data set including ~5000 earthquakes and more than 600 stations was considered for running the GIT analysis, while the authors used a small subset of 39 events spanning 3.5<Mw<6.3 and 14 well-distributed broadband stations for calibrating the model for