Ground motion prediction equations (GMPEs) can be considered as one of the fundamental components of any seismic hazard analysis (SHA), providing results which are essential for earthquake engineering as well as for seismic risk evaluations. Over the past decades, thanks to the availability of larger calibration datasets, empirical equations have undergone great advances in the derivation of more and more complex functional forms. Nevertheless, despite these major developments, there are still too few observations in near-field regions in order to give a well constrained prediction. On the other hand, relative close earthquakes (Repi < 10 – 20 km) with relatively high magnitudes (Mw > 5.5) are usually dominating the hazard, especially in Italian regions. In this framework, the present work aims at proving the reliability of several ground motion models in near-field regions, by testing them against a set of accelerometric records mainly provided by NESS2.0 dataset. To this end, as a subset of observations of engineering interest, records are selected within a distance of 20 km from the source. In addition, in order to provide a more comprehensive picture, analyses are extended to BB-SPEEDset: a synthetic database of broadband ground motions. Evaluation of residuals are therefore extensively addressed in order to highlight, if any, systematic overprediction (or underprediction) of GMPEs with respect to records and simulations, with the final goal of creating a ranking of the performance of GMPEs in near-field regions. Finally, a probabilistic SHA is performed for the sites of Catania and Siracusa, with the aim of showing, once again, the fundamental role of the near-field data in hazard analyses.
Le relazioni di attenuazione (RdA) rappresentano uno degli ingredienti fondamentali in ogni analisi di pericolosità sismica, dove vengono impiegate per fornire una stima del livello di scuotimento atteso al sito. Grazie al continuo arricchimento delle basi di dati accelerometriche, le RdA sono ad oggi costituite da forme funzionali sempre più complesse, in grado di descrivere l’attenuazione in funzione di un numero sempre più cospicuo di predittori. Nonostante i grandi progressi avvenuti negli ultimi anni, ci sono però ad oggi troppe poche osservazioni in campo vicino per ottenere una predizione che sia veramente rappresentativa del moto sismico in prossimità della faglia. D’altra parte, però, sono proprio i terremoti registrati nelle vicinanze della sorgenti sismogenetiche (Repi < 10 – 20 km), con magnitudo relativamente elevate (Mw > 5.5), a dominare la pericolosità al sito, soprattutto in Italia. Il presente lavoro si colloca quindi in questo contesto con lo scopo di verificare l’affidabilità di alcune relazioni di attenuazione in condizioni near-field, tramite un campione di registrazioni accelerometriche fornite principalmente dal dataset NESS2.0. Al fine di definire un sotto campione di osservazioni di interesse ingegneristico, si considerano solamente i dati registrati entro 20km dalla sorgente. Oltre alle registrazioni accelerometriche l’analisi comprende un set di simulazioni numeriche, denominato BB-SPEEDset. Vengono quindi effettuate una serie di analisi dettagliate dei residui, per verificare l’eventuale sovrastima/sottostima sistematica di osservazioni o simulazioni, allo scopo di creare una classifica delle RdA maggiormente performanti in condizioni di campo vicino. Infine, si propone un’applicazione ad uno studio di pericolosità sismica per i siti di Catania e Siracusa al fine di quantificare l’influenza delle RdA sugli spettri di risposta probabilistici, e dimostrare, ancora una volta, il ruolo dominante dei dati near-field nelle analisi di pericolosità.
Earthquake ground motion prediction models in the near-source region and their effect on probabilistic seismic hazard analyses
Chiecchio, Angela
2019/2020
Abstract
Ground motion prediction equations (GMPEs) can be considered as one of the fundamental components of any seismic hazard analysis (SHA), providing results which are essential for earthquake engineering as well as for seismic risk evaluations. Over the past decades, thanks to the availability of larger calibration datasets, empirical equations have undergone great advances in the derivation of more and more complex functional forms. Nevertheless, despite these major developments, there are still too few observations in near-field regions in order to give a well constrained prediction. On the other hand, relative close earthquakes (Repi < 10 – 20 km) with relatively high magnitudes (Mw > 5.5) are usually dominating the hazard, especially in Italian regions. In this framework, the present work aims at proving the reliability of several ground motion models in near-field regions, by testing them against a set of accelerometric records mainly provided by NESS2.0 dataset. To this end, as a subset of observations of engineering interest, records are selected within a distance of 20 km from the source. In addition, in order to provide a more comprehensive picture, analyses are extended to BB-SPEEDset: a synthetic database of broadband ground motions. Evaluation of residuals are therefore extensively addressed in order to highlight, if any, systematic overprediction (or underprediction) of GMPEs with respect to records and simulations, with the final goal of creating a ranking of the performance of GMPEs in near-field regions. Finally, a probabilistic SHA is performed for the sites of Catania and Siracusa, with the aim of showing, once again, the fundamental role of the near-field data in hazard analyses.| File | Dimensione | Formato | |
|---|---|---|---|
|
Chiecchio_Angela_TesiLM.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Tesi di Laurea Magistrale
Dimensione
18.29 MB
Formato
Adobe PDF
|
18.29 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/173119