Testing and evaluation of earthquake rupture simulations for New Zealand fault systems and the main Himalayan Thrust

In this thesis, I have evaluated the synthetic earthquakes from RSQSim for New Zealand fault

systems and the Main Himalayan Thrust fault, focusing on their implications for seismic hazard

assessment. RSQSim is a boundary element model that employs an approximation of the

constitutive rate- and state-dependent friction law to simulate seismicity in a diverse tectonic

environment.

Here, I first evaluated the output of first-generation RSQSim for New Zealand against the 2010

New Zealand National Seismic Hazard Model (NZ NSHM), which employed the same 530

faults geometry and slip rates from the NZ NSHM 2010. First, I compared slip rates derived

from the hybrid loading method with those from the NZ NSHM 2010. Then, I compared

seismicity, seismic moment rates and hazard estimates from RSQSim with the NZ NSHM

2010. Finally, I compared the probabilistic seismic hazard estimates for 475- (10% probability

of exceedance in 50 years), 2500-, and 10000-year return periods prepared using rupture sets

from RSQSim with the classical PSH estimates from the NZ NSHM 2010. I found that slip

rates derived from hybrid loading, seismicity, seismic moment rate and hazard estimates from

RSQSim closely align with NZ NSHM 2010 on the South Island, where crustal faults mainly

accommodate the deformations. Discrepancies are higher on the North Island; in this region,

the Hikurangi subduction interface and the crustal faults adjust the plate deformations. It is

presented in Chapter 2.

I then evaluated the synthetic earthquakes from RSQSim generated using the recently

developed ~ 880 fault sources from the NZ Community Fault Model version 1 (NZ CFM v1)

against the NZ NSHM 2022 inversion fault model (IFM), which also used the same fault

sources as the RSQSim. Here, I compared magnitude-frequency distributions (MFDs), seismic

moment rates, and peak ground accelerations (PGAs) at 475, 2500, and 10,000-year return

periods derived from the NZ CFM v1 RSQSim catalogue with results from the NZ NSHM

2022. I performed hazard comparisons using the same ground motion model, so the difference

mainly arises from the source modelling technique only. I found that MFDs and seismic

moment rates from RSQSim lie within NS NSHM 2022 distributions. Furthermore, PGAs

discrepancies (NZ CFM v1 RSQSim - NZ NSHM 2022 IFM) derived from NZ CFM v1

RSQSim catalogues with NZ NSHM 2022 IFM are large (~ ±250%) at a 10,000-year return

period in the low seismicity region (fault slip rate ≤ 0.5 mm/yr) and on the North Island. When

I included the NZ NSHM 2022 distributed seismicity model (DSM) in RSQSim and NSHM

2022, hazard discrepancies at low seismicity regions reduced notably but persisted in the North Island sites overlying the Hikurangi subduction interface. Overall, hazards derived from

RSQSim rupture sets indicated good agreement with NSHM 2022 on the South Island, where

crustal faults primarily accommodate deformations, despite the higher discrepancies on the

North Island NZ CFM v1 RSQSim-derived PGAs lie within the 5th and 95th quantile of those

of the NZ NSHM 2022. I presented this study in Chapter 3.

I then used the RSQSim to simulate earthquakes on the Main Himalayan Thrust (MHT), the

most prominent continental subduction thrust on the Earth. I used rich geologic, seismic and

geodetic datasets to incorporate the MHT’s 3D geometry, slip rates, and frictional into RSQSim

to simulate 10,000 kyr seismicity on MHT. I found that seismicity rates, the spatial distribution

of events and the rupture extent from MHT RSQSim are comparable to historical and

instrumental earthquake records along the MHT. Further, I found that previously inferred low

coupling zones on the MHT act as rupture barriers in the simulations and limit the maximum

magnitude on the MHT to Mw 8.9. I also compared stochastic probabilistic seismic hazard

(PSH) maps derived using the RSQSim catalogue with classical PSH maps. I found that

stochastic-derived PGAs agree with the classical on the MHT. At 2500- and 10,000-year return

periods, 84% of sites have PGA differences below 35%. I presented this in Chapter 4.

Overall, the findings of this thesis illustrate that RSQSim can be used to replicate observed

seismicity and the seismic hazard estimates on the South Island faults where crustal faults

primarily accommodate the deformations. Discrepancies in seismicity, seismic moment rates

and hazard estimates are higher when compared with NSHM on the North Island, where the

Hikurangi subduction interface and the crustal faults together adjust the plate deformations,

suggesting the need for improvement in the RSQSim model and hazard analysis. Similarly, it

can be used on the MHT to reproduce historical and instrumental seismicity. Given the short

instrumental and historical earthquake records for forecasting long-term earthquake rates,

synthetic earthquakes generated by RSQSim offer a viable solution for long-term forecasting

and hazard assessment.