Evaluating and Comparing Seismicity Rate Models in the Low-Strain-Rate Otago Region, Aotearoa, New Zealand

Abstract

Developing seismicity rate models (SRMs) in low-strain-rate regions is particularly challenging due to the limited availability of data to forecast future earthquakes. Here, we use the New Zealand Community Fault Model (NZ CFM) to evaluate three fault-based SRMs for the low-strain-rate Otago region: an inversion fault model (IFM) used in the New Zealand National Seismic Hazard Model 2022 (NZ NSHM 2022), a synthetic earthquake catalog generated by the physics-based Rate-and-State Earthquake Simulator (RSQSim), and stochastic catalogs that use a priori defined renewal processes and on-fault magnitude–frequency distributions (MFDs). Our analysis indicates that the IFM resolves relatively high rates of Mw ≥ 7:5 multifault ruptures in Otago, while the RSQSim catalog favors segmented Mw 7.0–7.4 ruptures. This leads to RSQSim suggesting higher seismic hazard estimates in Otago than the IFM at low probabilities of exceedance; however, this discrepancy is small relative to other sources of uncertainty within the NZ NSHM 2022. To compare these SRMs against instrumental seismicity, we use the constraint that no Mw ≥ 5 earthquakes were recorded in the Otago study area between 1951 and 2021. These 70 years of quiescence can be replicated by the RSQSim and stochastic catalogs with characteristic on-fault MFDs, but not with stochastic catalogs that implement either Gutenberg–Richter on-fault MFDs or the NZ NSHM 2022 geodetic model slip-rate estimates. Comparisons to the NZ NSHM 2022 distributed seismicity models indicate that a least-information uniform rate zones (URZs) negative binomial forecast aligns better with the NZ CFM-based SRMs than forecasts using a URZ-Poisson or hybrid model. Paleoseismic records from Otago suggest 10–100 ka spatiotemporal migrations of fault activity; however, this is not replicated by the RSQSim catalog. Collectively, these results highlight the challenges and opportunities of developing SRMs in low-strain-rate regions.