Quantifying co-seismic and post-seismic erosion using geomorphic change detection following the Mw 7.8 Kaikōura earthquake, New Zealand

Mountain belts are built by uplift and deformation, the associated earthquakes commonly triggering widespread and destructive landsliding across the landscape. The quantification of co-seismic landslide erosion is important because post-seismic observations reveal enhanced mass wasting and fluvial transport of sediment following a large earthquake. These cascading processes, conceptualised as the hazard response chain, have implications for both the management of hazards in seismically active regions and long-term landscape evolution. However, direct observations of the geomorphic processes driving the sediment cascade are rare, particularly across different physiographic settings. The aim of this study is to advance our understanding of earthquake induced landscape dynamics by accurately estimating both the volume of co-seismic landslides and the rate at which their sediment is remobilised after an earthquake.

The original contribution of this thesis is to quantify co- and post-seismic surface processes with GCD to reduce the uncertainty when estimating co-seismic and post-seismic erosion. The range and resolution of remote sensing data collected before and after the 2016 Mw 7.8 Kaikōura earthquake, New Zealand, are of a higher spatial and temporal resolution than those available following previous large earthquakes globally. Using these datasets, several new approaches to quantifying co-seismic and post-seismic erosion are presented. These include: i) the use of a pre- to post-seismic difference model to individually estimate volume for the majority of source areas within a co-seismic landslide inventory; ii) the volumetric quantification of co-seismic landslide debris location on hillslopes immediately after the earthquake to determine the portion of sediment delivered off-slope to the fluvial domain; and iii) the quantification of a catchment scale post-seismic sediment cascade to directly observe earthquake impact on landscape processes.

Understanding the influence of local controls advances the understanding of how earthquakes topographically focus denudation and the duration of sediment cascades. The total volume from the > 30,000 co-seismic landslides is 234 +81/-55 (1σ) M m3. The 9 largest co-seismic landslides in bedrock, all >100,000 m2, represented 38% of the total volume. Source material and failure mechanism regulated the evacuation of landslide sediment in the years after the earthquake. Where catchments were dominated by debris (soil and rock) avalanches, the volume of sediment which was post-seismically eroded on hillslopes within 5 years of the earthquake was equal to up to 25% of the total volume of landslide debris. For a singular large rock avalanche, up to 54% of the debris volume was remobilised. However, <10% of the of the debris volume was transported beyond the range front of the Seaward Kaikōura Ranges over these 5 years. In contrast, ≤1% of the volume of large coherent rock slides in Neogene siltstones, sandstones and limestones was remobilised, despite these landslides dominating the total volume of co-seismic landsliding. Therefore, the local physiographic setting regulated how the hillslopes responded to strong ground shaking, the volume and duration of earthquake induced erosion, and the mass balance of earthquakes.