PALEOSEISMOLOGY OF AN IMBRICATE REVERSE FAULT SYSTEM, SOUTH ISLAND, NEW ZEALAND: MAXIMUM MOMENT MAGNITUDES

Multi-fault earthquakes have greater potential moment magnitudes (M_W) than those involving a isolated fault or fault segment. Historical ruptures on reverse faults and recent stress modelling suggest that traditional length-displacement-magnitude scaling law estimates may underpredict M_W for synchronous ruptures on imbricate faults. We present a paleoseismic study of two imbricate reverse faults, the Fox Peak and Forest Creek faults in the central South Island of New Zealand, and model the maximum M_W for the fault system. The data from 5 paleoseismic trenches, over 100 survey lines, neotectonic mapping and structural modelling show that these two faults may have ruptured together during their MRE and penultimate events (c. 2500 and 5000 years ago, respectively). While the faults have only moderate slip rates (c. 1 mm/yr) and recurrence intervals (c. 3000 years) compared to the faster slipping faults of the New Zealand plate boundary, each individual fault is capable of producing a c. M_W 7.2 earthquake when rupturing in isolation. We use Monte Carlo simulations with variables constrained by field measurements, regional geophysical data, and an exponential, distance-based jumping probability to derive probability distributions of M_W allowing for synchronous Fox Peak-Forest Creek fault earthquakes. The results of the Monte Carlo simulation show that allowing for combined Fox Peak-Forest Creek fault earthquakes causes an average M_W increase of 0.15-0.20 over that of an isolated earthquake (a 50-100% increase in seismic moment). The relative probabilities of the combined and isolated ruptures are distinguishable in the 'peaks' of the distributions. Listric fault geometries, as observed in field mapping and seismic surveys, cause an even larger magnitude increase. Coulomb stress modelling can help inform which magnitude distribution should be used for hazard analysis. Paleoseismic studies that do not take into account subsurface fault linkage (e.g. fault imbrication), fault-to-fault jumping probabilities, and listric fault geometries will underestimate the magnitude potential of a fault system. Our methodology can readily be applied to other imbricate reverse fault systems around the world.