The spatiotemporal distribution of the background seismicity identified by different declustering methods is analysed in the case of Northeastern Italy. In particular, the performance of two declustering methods is compared: the Nearest Neighbour (NN) algorithm, developed by Zaliapin and Ben-Zion (2016), and the Stochastic Declustering (SD) algorithm, by Zhuang et al. (2004). Many studies assume that the background seismicity is a realization of a homogeneous Poisson process over time, but not necessarily in space (e.g. the well-known Epidemic Type Aftershock Sequence - ETAS - model by Ogata, 1998). In this study a statistical analysis is performed to assess whether the background seismicity identified by each declustering method has the spatio-temporal features typical of such a Poisson process. Different spatiotemporal measures are applied to assess the performance of the aforementioned declustering methods in removing the spatiotemporal clustering patterns from the full catalogue. The Allan Factor (Allan, 1966) and the Markov Modulated Poisson Process (Rydén, 1996) are used for the analysis of time patterns, while the Morisita Index (Morisita, 1959) is used for space clustering study (Benali et al., 2020, 2022, and references therein). Our results show that the time correlation and the space clustering are reduced, but not totally eliminated, in Northeastern Italy declustered catalogues. Moreover, we found that the Markov Modulated Poisson Process with multiple states is a more suitable background seismicity model than the traditional homogenous Poisson process. The obtained results support consistency and effectiveness of the adopted declustering methods, particularly within the identified short to intermediate time ranges. However, none of these declustering methods totally eliminated the spatiotemporal clustering from the whole earthquakes catalogue. The natural clustering of seismicity in space eventually complicates the interpretation of identified residual spatial correlation and the assessment of any declustering approach.
Statistical analysis of declustered catalogues in Northeastern Italy
E Varini;
2023
Abstract
The spatiotemporal distribution of the background seismicity identified by different declustering methods is analysed in the case of Northeastern Italy. In particular, the performance of two declustering methods is compared: the Nearest Neighbour (NN) algorithm, developed by Zaliapin and Ben-Zion (2016), and the Stochastic Declustering (SD) algorithm, by Zhuang et al. (2004). Many studies assume that the background seismicity is a realization of a homogeneous Poisson process over time, but not necessarily in space (e.g. the well-known Epidemic Type Aftershock Sequence - ETAS - model by Ogata, 1998). In this study a statistical analysis is performed to assess whether the background seismicity identified by each declustering method has the spatio-temporal features typical of such a Poisson process. Different spatiotemporal measures are applied to assess the performance of the aforementioned declustering methods in removing the spatiotemporal clustering patterns from the full catalogue. The Allan Factor (Allan, 1966) and the Markov Modulated Poisson Process (Rydén, 1996) are used for the analysis of time patterns, while the Morisita Index (Morisita, 1959) is used for space clustering study (Benali et al., 2020, 2022, and references therein). Our results show that the time correlation and the space clustering are reduced, but not totally eliminated, in Northeastern Italy declustered catalogues. Moreover, we found that the Markov Modulated Poisson Process with multiple states is a more suitable background seismicity model than the traditional homogenous Poisson process. The obtained results support consistency and effectiveness of the adopted declustering methods, particularly within the identified short to intermediate time ranges. However, none of these declustering methods totally eliminated the spatiotemporal clustering from the whole earthquakes catalogue. The natural clustering of seismicity in space eventually complicates the interpretation of identified residual spatial correlation and the assessment of any declustering approach.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
