Aftershock Sequences and Poisson Distribution
Abstract
The evaluation of temporal independence of events is a major issue in seismic hazard assessment. Seismic hazard analysis performed by the Cornell approach is often undermined by the scarcity of the number of major earthquakes, due to both the limited dimension of seismic sources and the short time-span of seismic catalogs. Therefore the adoption of Poisson distribution (that entails temporal independence of events and a given rupture behavior of the seismic sources) is rarely sufficiently validated by real data.
The analysis of aftershock sequences, characterized by a large sample of events, is helpful for understanding the temporal behavior of rupture processes. In the present study, we focus on the temporal independence of events. A number of authors keep this point of view: "Because aftershocks are a consequence of a mainshock, they cannot constitute a sample of statistically independent events", this sentence is rather controversial and may confuse causality and statistical independence. We analyzed several aftershock sequences and investigated the temporal independence by the fit of data to several negative exponential distribution of interarrival times. We found that the minimum magnitude threshold (MMT) plays a relevant role. The aftershock sequences analyzed show a similar behavior; the main aspects are: 1) as MMT increases, the degree of temporal independence increases; above a given threshold (but still considering a sequence composed by a large number of events) we can observe complete independence; 2) assuming a low MMT, the sequences appear no longer constituted by independent events, in particular the deviation from independence is significant, both in very low and very high interarrival time ranges. According to our results, the choice of MMT value is relevant to assess if aftershock sequences follow Poissonian models. As an example, let us consider a mainshock 7.1: setting MMT of M=3, the sequence shows a large departure from the independence; on the contrary, filtering with high-pass MMT of M=4, the sequence fits well the Poisson distribution.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.S41H0626D
- Keywords:
-
- 7299 General or miscellaneous;
- SEISMOLOGY