Poster #266, Earthquake Forecasting and Predictability (EFP)

b-value of what? Complex behavior of the magnitude distribution during and within the 2016/17 Central Italy sequence

Marcus Herrmann, Ester Piegari, & Warner Marzocchi
Poster Image: 

Poster Presentation

2021 SCEC Annual Meeting, Poster #266, SCEC Contribution #11624 VIEW PDF
The Magnitude–Frequency-Distribution (MFD) of earthquakes is a major ingredient for earthquake forecasting and seismic hazard models. It is typically modeled with the Gutenberg–Richter (GR) relation and its b-value (slope of the GR relation) to infer the occurrence rate of large earthquakes from small ones. The b-value is often observed to vary in space and time. Resolving those variations properly is critical to understanding the underlying physical processes, but several aspects affect this ability (e.g., technical issues, data quality, incompleteness, sampling/processing methods, and expert judgment).

Here we argue that a complex sequence with multiple ruptured faults can f...
urther bias MFD and b-value analyses: If the internal structure and evolution of a sequence is not properly considered, a sectionally and/or temporally varying MFD could be neglected, leading to inappropriate inferences. To make MFD analyses more physically meaningful, we introduce a new perspective by isolating seismicity in spatial clusters based on hypocenter density and dividing them into temporal periods defined by the largest events, accounting for short-term incompleteness. The Central Italy 2016/17 sequence is used as an example due to its complex structure, cascading evolution, and the availability of a high-resolution catalog.

The obtained clusters reveal a complex structure and behavior. The MFD among and within the clusters varies spatially and temporally, but may occasionally remain constant in time. For instance, after the Mw6.0 Amatrice event and prior to the Mw5.9 Visso event, the b-value differed largely among the three main clusters. Afterward, but still prior to the largest Mw6.5 Norcia event, the b-value further increased in that structure where Norcia eventually occurred. Norcia did not alter the MFD of the main clusters further; their b-value remained high. In a subhorizontal structure at depth, the b-value remained constantly high throughout the sequence.
Our results suggest a complex MFD behavior during a sequence, questioning interpretations based on an average b-value over the whole sequence (little representational power) or for arbitrary data selections (insufficient or unphysical resolution). Our findings raise the question of the appropriate spatiotemporal scale for resolving b-value variations during a sequence. Properly accounting for small-scale variations potentially improves forecast performance.