Landslide dams have been recognised as significant components of multi-hazard cascading systems, linking slopes and rivers. Despite the potential for catastrophic consequences, landslide dam breaching and evolution remain under-researched and poorly understood, often due to the remoteness of large volume, valley-blocking landslides and the general lack of high resolution pre- and post-failure survey data. The Hapuku Rock Avalanche presents a unique opportunity to study landslide dam evolution and breaching timelines due to the accessibility of the site and the availability and resolution of pre- and post-failure remote sensing data. Field observations and mapping, sampling, geophysical surveying, and 27 remote sensing surveys from 2016 to 2022 have provided detailed data on the dam. The Hapuku landslide was the largest rock avalanche triggered by the 2016 Mw 7.8 Kaikōura earthquake sequence, occurring ∼9 km upstream of the main highway and rail corridor on the South Island of New Zealand. It dammed the Hapuku River, which rapidly formed a lake behind the 80 m-high deposit. Four major erosion events and three significant partial breach events, identified through observations and remote sensing data differencing, resulted in water outflow from the lake, significant erosion of the dam and deposition of sediment into the river. The partial breaches correspond with less than 1 in 10-year rainfall events in 2017 and 2018, and the first occurred 141 days after dam formation. Seepage and internal erosion of the dam were observed to be progressing upstream before the partial breaches, in which water overtopped the dam. The third partial breach event, 2 years after dam formation, was the most significant erosional event in the last 6 years. The dam has eroded episodically and more locally since 2018, and the degree of erosion appears to be decreasing with time, despite more intense storms. A small lake remains. The evolution of the Hapuku Rock Avalanche dam emphasises the complexity of dam and breaching evolution, which are often oversimplified.
Read full abstract