GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 2.2 GNGTS 2024 SDQ: a new tool for the evaluaHon of seismic- accelerometric data quality F. Varcheja 1 , M. Massa 1 , R. Puglia 1 , P. Danececk 2 , S. Rao 2 , A. Mandiello 2 , D. Piccinini 3 1 INGV, sezione di Milano, Italia 2 INGV, ONT, Osservatorio Nazionale Terremo9, Roma, Italia 3 INGV, sezione di PISA, Italia Abstract In recent years, many works have focused their abenCon on the issue of data processing and verificaCon procedures, in parCcular of strong-moCon data because of their fundamental role in the case of strong earthquakes At the naConal level in Italy, there are currently two web portals available for checking seismic data quality: EIDA Italia ( hbps://eida.ingv.it/it/getdata ) and ISMDq ( hbp://ismd.mi.ingv.it/quality.php ). The EIDA Italia (Danecek et al., 2021) employs the data quality tools directly from the ORFEUS website (hbp://www.orfeus-eu.org/data/ eida/quality/), while ISMDq (Massa et al., 2022) is a recent system for dynamic quality check both of conCnuous data stream and earthquake data. In this work, we introduce the SDQ (Seismic Data Quality) project, a new open-source Python tool freely available and downloadable. It is designed for the automaCc monitoring of sismo- accelerometric staCons by analyzing both events - selected based on magnitude and distance - and conCnuous data streams. Regarding earthquake data, the quality of individual waveforms is assessed by comparing the ground moCon parameters derived from co-located accelerometers and velocimeters. SDQ operates by uClizing a simple external input file containing the INGV event idenCfier, staCon, and network codes. SDQ is organized into three main phases: acquisiCon, pre-processing, and processing. In the acquisiCon phase, event informaCon, staCon metadata, and waveforms are downloaded from FDSN (hbps://www.fdsn.org/ ) web services ( hbps://www.fdsn.org/ webservices/). IniCally, the waveforms are analyzed to idenCfy event and pre-event noise windows. Then, the data are converted into physical units, accounCng for pre-filter parameters and exclusion condiCons. Finally, each single waveform is assigned to a quality class ranging from A (excellent) to D (data to be rejected) based on Cme- and frequency- dependent algorithms. Thresholds for classificaCon were empirically obtained by combining visual signal inspecCon and staCsCcal analysis. Tests were performed considering about 15,000 waveforms in the Cme interval from January 2012 to June 2023, encompassing all 6- channel staCons of the IV (NaConal Seismic Networks ( hbps://www.fdsn.org/networks/ detail/IV/)) and MN (MedNet network, hbps://www.fdsn.org/networks/detail/MN/) , both managed by the Italian NaConal InsCtute for Geophysics and Volcanology (INGV, hbps:// www.ingv.it/) .