2.3 Geologic recurrence values
We summarize geologic recurrence values for each fault section in Table 1. Two primary types of recurrence data are depicted: recurrence inferred from land-level changes, and recurrence assumed from tsunami deposits. The sensitivity of both types of data to earthquake magnitude is unknown, and various combinations of slip, magnitude, and location likely influence land-level change and tsunami generation. Only the largest (Mw ≥ 8.5) events may leave unambiguous records: for example, the Mw 8.2 Chignik rupture generated a negligible near- and far-field tsunami and small (< 0.08 m) vertical displacements (Elliott et al., 2022; Ye et al., 2022), less than the theoretical detection limit of 0.1- 0.2 m discussed by Shennan et al. (2016). Until more is known about the sensitivity of land-level change and tsunami recorders to earthquake rupture characteristics in the AASZ, we assume that the geologic data records earthquakes ≥ Mw 8.5.
Uncertainties are not reported in a standardized way for the geologic recurrence data we summarize here, so we use author-reported recurrence intervals and uncertainties. Where not supplied by the authors, we calculate the mean recurrence interval by dividing n-1 events into the total closed interval (oldest event to most recent event) or n events into the total open interval (oldest event to present day) and assign uncertainty equal to the standard deviation of the mean recurrence value (Table 1). More complicated calculations are possible (Field et al., 2013) but are not yet warranted for the AASZ because of the relative lack of data, and the sometimes disparate approaches and assumptions used for event identification and subduction interface earthquake age estimates. We presume that recurrence calculations are standardized within any particular hazard model framework, and a logic tree approach will be used to propagate uncertainties in recurrence and paleo-event size for classic probabilistic seismic hazard analysis (National Research Council, 1997) or that recurrence values with standardized uncertainty will be used as a constraint in inversion-based PSHA (Field et al., 2020).