Mark W. Stirling’s research while affiliated with University of Otago and other places

Long recurrence intervals of large earthquakes relative to the historical record mean that geological data are often utilized to inform forecasts of future events. Geological data from any particular fault may constrain the timing of past earthquakes (paleoearthquake data), or simply the time period over which a certain amount of fault displacement has occurred due to one or more earthquakes. These data are typically subject to large uncertainties, and available records often only constrain the timing of a few events. Variability in earthquake inter‐event times (aperiodicity) has been observed for many faults, particularly in low seismicity regions, further hampering the utilisation of small data sets for developing forecasts. A challenge for earthquake forecasting therefore concerns how best to utilize all of the limited available data while fully considering uncertainties. Here we present a concise Bayesian model for developing time‐dependent earthquake forecasts from geological data. Using the additive property of the Brownian passage time distribution, we make inference on the model parameters jointly from paleoearthquake and fault displacement data. Monte Carlo Markov Chain methods are used to sample the posterior distribution of model parameters, which is subsequently used to forecast future earthquake probabilities. The method incorporates data uncertainties and does not rely on a priori assumptions of quasiperiodic earthquake recurrence, allowing application in a wide range of tectonic settings. We demonstrate the method using data from two reverse faults in Otago, southern Aotearoa New Zealand, a region in which aperiodic earthquake recurrence has previously been observed.

Evaluating fault segmentation is important for our understanding of seismic hazard assessment and fault growth. However, it is still unclear what controls if reverse fault earthquakes will rupture across segment boundaries. Here, we combine fault mapping and trench data from the low slip rate (0.04-0.15 mm/yr) multi-segment Nevis-Cardrona Fault (NCF) in the South Island of Aotearoa New Zealand to assess if it has ruptured in single or multi-segment earthquakes during the late Quaternary. Two new trenches on its Nevis segment provide stratigraphic evidence for two surface rupturing earthquakes, which through Optically Stimulated Luminscence dating and OxCal modelling, are constrained to have occurred at 28.9 +12.9 -9.1 ka and 12.8 ± 4.9 ka. The most recent timing is only weakly correlated to surface rupture timings from two trenches along the NCF's NW Cardrona segment. Furthermore, the 2 ± 1 m Nevis segment single event displacements we estimate would be unusually low for a ~85 km long NCF multi-segment rupture. We therefore surmise that late Quaternary NCF surface rupturing earthquakes did not rupture through ~30-50° bends that link these segments. Our trench data and fault mapping also indicate lower slip rates on the Nevis segment than previous studies (0.04-0.1 mm/yr vs 0.4 mm/yr).

Forecasting large earthquakes along active faults is of critical importance for seismic hazard assessment. Statistical models of recurrence intervals based on compilations of paleoseismic data provide a forecasting tool. Here we compare five models and use Bayesian model-averaging to produce time-dependent, probabilistic forecasts of large earthquakes along 93 fault segments worldwide. This approach allows better use of the measurement errors associated with paleoseismic records and accounts for the uncertainty around model choice. Our results indicate that although the majority of fault segments (65/93) in the catalogue favour a single best model, 28 benefit from a model-averaging approach. We provide earthquake rupture probabilities for the next 50 years and forecast the occurrence times of the next rupture for all the fault segments. Our findings suggest that there is no universal model for large earthquake recurrence, and an ensemble forecasting approach is desirable when dealing with paleoseismic records with few data points and large measurement errors.

A seismicity rate model (SRM) has been developed as part of the 2022 Aotearoa New Zealand National Seismic Hazard Model revision. The SRM consists of many component models, each of which falls into one of two classes: (1) inversion fault model (IFM); or (2) distributed seismicity model (DSM). Here we provide an overview of the SRM and a brief description of each of the component models. The upper plate IFM forecasts the occurrence rate for hundreds of thousands of potential ruptures derived from the New Zealand Community Fault Model version 1.0 and utilizing either geologic- or geodetic-based fault-slip rates. These ruptures are typically less than a couple of hundred kilometers long, but can exceed 1500 km and extend along most of the length of the country (albeit with very low probabilities of exceedance [PoE]). We have also applied the IFM method to the two subduction zones of New Zealand and forecast earthquake magnitudes of up to ∼Mw 9.4, again with very low PoE. The DSM combines a hybrid model developed using multiple datasets with a non-Poisson uniform rate zone model for lower seismicity regions of New Zealand. Forecasts for 100 yr are derived that account for overdispersion of the rate variability when compared with Poisson. Finally, the epistemic uncertainty has been modeled via the range of models and parameters implemented in an SRM logic tree. Results are presented, which indicate the sensitivity of hazard results to the logic tree branches and that were used to reduce the overall complexity of the logic tree.

The 2022 revision of Aotearoa New Zealand National Seismic Hazard Model (NZ NSHM 2022) has involved significant revision of all datasets and model components. In this article, we present a subset of many results from the model as well as an overview of the governance, scientific, and review processes followed by the NZ NSHM team. The calculated hazard from the NZ NSHM 2022 has increased for most of New Zealand when compared with the previous models. The NZ NSHM 2022 models and results are available online.

We develop new magnitude–area scaling relations for application in the New Zealand National Seismic Hazard Model 2022 (NZ NSHM 2022) and future applications. A total of 18 published relations are selected, comprising the following tectonic and slip types: crustal strike-slip (seven relations), reverse (two relations), normal (two relations), subduction interface (five relations), and two dip-slip relations to augment the small number of available reverse and normal relations. The scaling relations are evaluated against an instrumental earthquake database flatfile, and scores are provided for each relation. Equations of the form Mw=logA+C are then used to develop mean and bounding relations for the suite of scaling relations. The final set of relations used in NZ NSHM 2022 is adjusted to be consistent with observations of major historical New Zealand earthquakes and U.S. Geological Survey practice. We also provide a second set of Mw=logA+C relations that are absent of these adjustments and so more directly reflect the results of our scoring of the published relations.

Multifault ruptures are common for historical earthquakes, and here we consider their impact on seismic hazard. We compare ground-shaking hazard forecasts from the 2022 Aotearoa New Zealand National Seismic Hazard Model (NZ NSHM 2022), which incorporates many multifault ruptures (referred to as the multifault model) with modeled hazard from a simpler model of characteristic earthquakes on individual faults or fault segments (referred to as the segmented model). The multifault model includes very-low-probability rupture lengths of up to ∼1100 km and a mean of 221–234 km, whereas the segmented model primarily comprises rupture lengths of <200 km (mean, 43–51 km) and the maximum of 414 km. The annual rates of Mw 6.9–7.5 earthquakes are more than an order of magnitude higher for the segmented model (0.132–0.24/yr; recurrence times ∼4–7 yr) than the multifault model (0.027/yr; recurrence times 37 yr). Conversely, the rates of earthquakes are similar for segmented and multifault models at Mw>7.5 (0.018–0.031/yr; recurrence times 32–56 yr). Despite differences in rupture lengths and annual rates of earthquakes, the calculated ground-shaking hazard at 10% probability of exceedance (PoE) in 50 yr for the segmented model differs by <55% compared with the multifault model for 95% of sites across Aotearoa New Zealand. For 50% of sites, the modeled hazard differs by <20% between the two models. If a distributed seismicity model (DSM) is included in the hazard calculations, 95% of sites differ in modeled hazard by <18%, and 50% of sites differ by <2.2%. In most areas, seismic hazard at 10% PoE in 50 yr is greater for the segmented model than the multifault model, with notable exceptions along the central Alpine fault in the western South Island and the Taupō volcanic zone in the central North Island.