Regional earthquake induced landslide susceptibility : lessons from the 2016 Mw 7.8 Kaikōura earthquake.

Abstract

Coseismic landslides are one of the most impactful secondary hazards resulting from large earthquakes. Determining the physical factors that dictate landslide hazard during earthquakes is critical to improving landslide susceptibility models and hazard plans. The 2016 Mw 7.8 Kaikōura earthquake in the northeast South Island of New Zealand ruptured more that 20 on- and off-shore faults, and triggered more than 30,000 landslides. The exceptional documentation of faults and landslides from the event and the diversity of lithology and topography in the Kaikōura region present an opportunity to better understand seismic hazard and landslide susceptibility. This thesis presents a detailed analysis of landslide susceptibility from the Kaikōura earthquake with a primary focus on two factors which influence the distribution of landslides from the Kaikōura earthquake: surface fault ruptures and coastlines.

Distance to a surface fault rupture, which may include the influence of both ground motion and physical properties of the fault zone, exerts one of the strongest empirical controls on the distribution of earthquake induced landslides from the Kaikōura earthquake. Quantifying coseismic deformation fields about faults is one way to define a categorical predictor for the fault zone in landslide susceptibility analysis. Distributed displacement around the Papatea fault in 2016 was characterised using 3-D displacement fields to estimate the potential extent of the fault zone. The magnitude of displacement was compared to currently defined ‘fault avoidance zones’ which capture the majority of, but not all, significant distributed displacement from the Kaikōura earthquake. The same estimation of distributed displacement was then measured across 13 additional fault ruptures from the Kaikōura earthquake to determine the width of off-fault displacement and deformation, which is a proxy for the fault damage zone around surface fault ruptures. A higher density of earthquake induced landslides observed within this zone across ruptures cannot be fully explained by the attenuation of shaking in current strong ground motion models.

Comparative logistic regression models of regional landslide susceptibility were developed to further examine the relative influence of ground motion and the fault damage zone as well as the effect of other factors like distance to the coastline. Models confirm the observed contribution of the fault damage zone around faults but also suggest that strong ground motion plays a more significant role in the overall distribution of landslides. Improved ground motion models, based on more robust observational ground motion data and quantification of near-fault site effects, will be key to furthering our understanding of landslide susceptibility around faults.

There was an order of magnitude greater number of landslides from the Kaikōura earthquake on slopes within 1 km of the coast as compared to slopes from 1 to 3 km from the coast. Coastal slopes introduce a variety of site-specific landslide forcings, for example wave action, and susceptibility factors, like increased moisture, but these factors are rarely incorporated into regional earthquake induced landslide susceptibility studies. Comparative logistic regression models of landslide susceptibility suggest that slope angle dominates both inland and coastal landslide susceptibility in the Kaikōura region. Average slope is steeper along the uplifted Kaikōura coast – a legacy effect of past wave action, erosion, and relict landsliding.

While most coastal slopes in the Kaikōura region are buffered from wave action by uplifted shore platforms, at Conway Flat, south of Kaikōura, coastal cliffs are exposed to wave action at high tide. The edge of the coastal cliff top was repeatedly mapped at Conway Flat using a 72- year record of historical aerial imagery and recent lidar. In this location, the Kaikōura earthquake produced nearly 25% of cumulative coastal cliff retreat over the last 72 years. More than 50% of retreat at Conway Flat since 1950 can be traced back to strong ground motion in earthquakes. Estimates of strong ground motion recurrence and potential coseismic retreat can be used alongside current estimates of cliff retreat to gauge the influence of earthquakes on steep coastline evolution globally.

The results of this work have improved our understanding of coseismic landslide susceptibility on a regional scale and offer some insights into the fundamental processes governing slope failure in large earthquakes. These types of contributions help to bridge the divide between practical and fundamental science aims (e.g., between empirical and physics-based, or regional and site-specific models) and represent critical steps towards increasing resilience to earthquakes.