How to be fooled searching for significant variations of the b-value

Marzocchi, Warner; Spassiani, Ilaria; Stallone, Angela; Taroni, Matteo

doi:10.1093/gji/ggz541

Cited by 88 publications

(69 citation statements)

References 67 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…That is to say, the features are computed on windows of n events (with n = 200) and their value is assigned to the nth event at the time t i and windows move of one event at time (i.e., each feature represents a time series). Here, n equal to 200 was selected considering that this parameter is important for obtaining robust estimates of the b-value [60]. For instance, Dresen et al, 2020 [27], used n equal to 150.…”

Section: Features 21 Features Computation and General Working Outlinementioning

confidence: 99%

Forecasting the Preparatory Phase of Induced Earthquakes by Recurrent Neural Network

Picozzi

Iaccarino

2021

Forecasting

View full text Add to dashboard Cite

Earthquakes prediction is considered the holy grail of seismology. After almost a century of efforts without convincing results, the recent raise of machine learning (ML) methods in conjunction with the deployment of dense seismic networks has boosted new hope in this field. Even if large earthquakes still occur unanticipated, recent laboratory, field, and theoretical studies support the existence of a preparatory phase preceding earthquakes, where small and stable ruptures progressively develop into an unstable and confined zone around the future hypocenter. The problem of recognizing the preparatory phase of earthquakes is of critical importance for mitigating seismic risk for both natural and induced events. Here, we focus on the induced seismicity at The Geysers geothermal field in California. We address the preparatory phase of M~4 earthquakes identification problem by developing a ML approach based on features computed from catalogues, which are used to train a recurrent neural network (RNN). We show that RNN successfully reveal the preparation of M~4 earthquakes. These results confirm the potential of monitoring induced microseismicity and should encourage new research also in predictability of natural earthquakes.

show abstract

Section: Features 21 Features Computation and General Working Outlinementioning

confidence: 99%

Forecasting the Preparatory Phase of Induced Earthquakes by Recurrent Neural Network

Picozzi

Iaccarino

2021

Forecasting

View full text Add to dashboard Cite

show abstract

“…The b‐value describes the magnitude‐frequency distribution of earthquakes, whereby a smaller value indicates a higher proportion of large earthquakes with respect to small earthquakes, and vice versa. Variations in the observed b‐value have been attributed to changes in stress conditions and/or rock heterogeneity, and fluid diffusion (Marzocchi et al., 2020).…”

Section: Moment Magnitude and Earthquake B‐valuementioning

confidence: 99%

Mechanism for Deep Crustal Seismicity: Insight From Modeling of Deformation Processes at the Main Ethiopian Rift

Muluneh

Brune

Illsley‐Kemp

et al. 2020

Geochem Geophys Geosyst

View full text Add to dashboard Cite

We combine numerical modeling of lithospheric extension with analysis of seismic moment release and earthquake b‐value in order to elucidate the mechanism for deep crustal seismicity and seismic swarms in the Main Ethiopian Rift (MER). We run 2‐D numerical simulations of lithospheric deformation calibrated by appropriate rheology and extensional history of the MER to simulate migration of deformation from mid‐Miocene border faults to ∼30 km wide zone of Pliocene to recent rift floor faults. While currently the highest strain rate is localized in a narrow zone within the rift axis, brittle strain has been accumulated in a wide region of the rift. The magnitude of deviatoric stress shows strong variation with depth. The uppermost crust deforms with maximum stress of 80 MPa, at 8–14 km depth stress sharply decreases to 10 MPa and then increases to a maximum of 160 MPa at ∼18 km depth. These two peaks at which the crust deforms with maximum stress of 80 MPa or above correspond to peaks in the seismic moment release. Correspondingly, the drop in stress at 8–14 km correlates to a low in seismic moment release. At this depth range, the crust is weaker and deformation is mainly accommodated in a ductile manner. We therefore see a good correlation between depths at which the crust is strong and elevated seismic deformation, while regions where the crust is weaker deform more aseismically. Overall, the bimodal depth distribution of seismic moment release is best explained by the rheology of the deforming crust.

show abstract

“…Given the strong correlation between a-and b-values and that the Poisson hypothesis used in Weichert (1980) approach may be challenging in Ischia (see next Section), we further check the stability of results by estimating the b-value also independently from the annual rate. To this end, we use the method described in Taroni (2021) that allows estimating the b-value with a timevarying magnitude of completeness, including the correction for the binning of the magnitudes (in our case 0.1) and the correction for an unbiased MLE (Marzocchi et al, 2020). We obtain a b-value = 1.18 +/−0.13, very similar to the one previously estimated.…”

Section: Magnitude-frequency Distributionmentioning

confidence: 61%

The Seismicity of Ischia Island, Italy: An Integrated Earthquake Catalogue From 8th Century BC to 2019 and Its Statistical Properties

et al. 2021

View full text Add to dashboard Cite

Ischia is a densely inhabited and touristic volcanic island located in the northern sector of the Gulf of Naples (Italy). In 2017, the Mw 3.9 Casamicciola earthquake occurred after more than one century of seismic quiescence characterized only by minor seismicity, which followed a century with three destructive earthquakes (in 1828, 1881, and 1883). These events, despite their moderate magnitude (Mw < 5.5), lead to dreadful effects on buildings and population. However, an integrated catalogue systematically covering historical and instrumental seismicity of Ischia has been still lacking since many years. Here, we review and systematically re-analyse all the available data on the historical and instrumental seismicity, to build an integrated earthquake catalogue for Ischia with a robust characterization of existing uncertainties. Supported by new or updated macroseismic datasets, we significantly enriched existing catalogues, as the Italian Parametric Earthquake Catalogue (CPTI15) that, with this analysis, passed from 12 to 57 earthquakes with macroseismic parametrization. We also extended back by 6 years the coverage of the instrumental catalogue, homogenizing the estimated seismic parameters. The obtained catalogue will not only represent a solid base for future local hazard quantifications, but also it provides the unique opportunity of characterizing the evolution of the Ischia seismicity over centuries. To this end, we analyse the spatial, temporal, and magnitude distributions of Ischia seismicity, revealing for example that, also in the present long-lasting period of volcanic quiescence, is significantly non-stationary and characterized by a b-value larger than 1.

show abstract

How to be fooled searching for significant variations of the b-value

Cited by 88 publications

References 67 publications

Forecasting the Preparatory Phase of Induced Earthquakes by Recurrent Neural Network

Forecasting the Preparatory Phase of Induced Earthquakes by Recurrent Neural Network

Mechanism for Deep Crustal Seismicity: Insight From Modeling of Deformation Processes at the Main Ethiopian Rift

The Seismicity of Ischia Island, Italy: An Integrated Earthquake Catalogue From 8th Century BC to 2019 and Its Statistical Properties

Contact Info

Product

Resources

About