Age Uncertainty in Recurrence Analysis of Paleoseismic Records

Abstract

Determining the aperiodicity of large earthquake recurrences is key to forecast future rupture behaviour. Aperiodicity is classically expressed as the coefficient of variation of recurrence intervals, though the recent trend to express it as burstiness is more intuitive and avoids minor inaccuracies. Due to the underestimation of burstiness in records with a low number of recurrence intervals, the paradigm is to obtain long paleoseismic records with many events. Here we present a suite of synthetic paleoseismic records designed around the Weibull and inverse Gaussian distributions that demonstrate that age uncertainty relative to the mean recurrence interval causes overestimation of burstiness. The effects of over-and underestimation interact and give complex results for accurate estimates of aperiodicity. Furthermore, we show that the way recurrence intervals are sampled from a paleoseismic record can have strong influences on the resulting statistic and its implication for probabilistic seismic hazard assessment. Comparing values of burstiness between paleoseismic records should therefore be done with caution.

Validation of seismic hazard curves using a calibrated 14 ka lacustrine record in the Eastern Alps, Austria

“…5 ). Because the number of interevent times is rather large in unit III (24), age uncertainty plays a smaller role 13 . The shortest interevent times found in unit III are 8 years, thus supporting the assumption that the lacustrine paleoseismic record only contains mainshocks.…”

“…B values larger than 0 indicate a “bursty” recurrence pattern, meaning that the standard deviation of interevent times is larger than their mean. We follow the classification of Kempf and Moernaut 13 and use the term “bursty” for B > 0.05, “aperiodic” for − 0.05 < B < + 0.05, “weakly periodic” for − 0.33 < B < − 0.05 and “strongly periodic” for B < − 0.33.…”

“…However, these (1) often fail to capture the long (> 500–1000 years) and variable recurrence times of the largest earthquakes in a region, especially in slowly deforming intraplate regions, and (2) often are used as model input data, and therefore cannot be used for model validation 9 , 12 . Short individual (paleo-) earthquakes chronologies can produce biased recurrence statistics 13 , 14 and make it hard to assess hazard map performance 7 . A common approach is therefore to integrate observations from several sites and substitute time for space 15 – 17 .…”

Earthquake forecasting from paleoseismic records

“…An alternative to renewal processes for modelling paleoseismic records uses a Long-Term Fault Memory Model 11 , which assumes that the timing of future events is dependent on not only the time elapsed since the most recent event (as in renewal models) but also the previous inter-event times. However, when considering global dataset and synthetic earthquake records there appears to be no significant correlation or anti-correlation between successive inter-event times for the vast majority of the earthquake records 9 , 12 . Therefore, in this study we only focus on renewal processes.…”

Recognising surface versus sub-surface deformation of soft-sediments: Consequences and considerations for palaeoseismic studies

“…Van Loon et al, 2016, Belzyt et al, 2021, we have stressed the importance of recognising erosive surfaces and sedimentary caps that link MTDs with the surface. Sub-surface deformation of the sedimentary sequence clearly leads to issues with simple age-depth correlations where the timing of seismic events is estimated from the level (depth) of sediment that they affect and bracketed by dated horizons (see Moernaut, 2020 andKempf andMoernaut, 2021 for reviews). The consequences of sub-surface intrastratal deformation is that the link with the age of the sediment is removed and it therefore cannot be assumed that deformation affecting stratigraphically lower beds is older and triggered by older earthquakes.…”