Landslide vulnerability: Learnings from the impacts of debris flows on buildings to inform quantitative loss of life risk assessments

L. N. Agnew, R. Hayes, A. S. Miner and A. Hunter

A significant weather event in February 2023 triggered numerous landslides in the townships of Muriwai, Piha and Karekare in the Waitākere area on Auckland’s West Coast. The event caused widespread damage to residential buildings and resulted in two deaths. Most landslides materialised as debris flows. Due to their high velocity and ability to travel long distances, debris flows pose a significant risk to life. Vulnerability can be described as the conditional probability of death (given a landslide impact), or as a conceptual measure of overall human susceptibility, sensitivity or proneness to landslide hazards (human vulnerability). The former is often related to the collapse of occupied buildings and thus is indirectly, a function of structural vulnerability.

This case study found the extent and nature of damage to residential buildings, and therefore vulnerability of the occupants, varied considerably. Typically, extent of damage depended on debris flow volume, and the recency of construction of the building. Building collapse due to inundation was common, however other ‘atypical’ instances of damage included buildings being ‘rafted’ downslope and the occupants escaping unharmed, trees entrained in the top of the debris penetrating the upslope side of buildings, buildings being undermined and remaining intact, and significant volumes of debris resting against the upslope wall but resulting in minimal, if any, damage. This variability indicates there can be significant uncertainty in the outcome of a landslide risk assessment, which can be difficult to quantify.

AGS (2007c) includes a table of example vulnerability values for a limited number of inundation and building damage scenarios, adapted from Finlay et al. (1999). In some instances, there are wide ranges in the data and a practitioner must use judgement to select an appropriate value. Should the building collapse, which occurred with many of the buildings impacted by landslides in Muriwai, Karekare and Piha, AGS (2007c) recommends a value of 1 (death is almost certain) be adopted. A revised and expanded set of vulnerability values adopted for the study site is presented in this paper.

Human behaviour is a key factor of human vulnerability (Pollock & Wartman, 2020) and had a large influence on the resulting number of deaths as many residences that suffered significant damage were either unoccupied or evacuated. Had this not been the case, the survivability in many properties would likely have been low. Records of human behaviour are often poorly persevered during an event (Pollock & Wartman, 2020), and it is difficult to predict the influence of human behaviour in a risk assessment. As such, human behaviour is generally ignored during the risk assessment process in favour of the worst-case scenario approach.