Recent global tsunami events including the 2004 Indian Ocean tsunami and the 2011 Great East Japan tsunami, have reinforced the importance of improving scientific understanding of tsunami hazard risk assessment and implementation of tsunami risk management. During an earthquake-tsunami event, evacuation is considered the primary risk reduction strategy for preventing casualties. Empirical data on evacuation behaviour and movements during and after past tsunami events has the potential to inform future disaster risk management and evacuation response planning. To date however, the majority of tsunami evacuation behaviour research has focused on specific aspects of evacuation response such as tsunami preparedness, risk awareness, or evacuation intentions. There have been limited contributions to capturing a comprehensive overview of a ‘real-event’ complete evacuation process to enhance knowledge on response to warnings and evacuation decision-making, evacuation movements, and additional activity after evacuation. Moreover, tsunami evacuation modelling has been an increasingly applied risk management tool to assess evacuation potential for at risk communities however, very few publications have used ‘real-event’ evacuation parameters to improve realistic evacuation modelling outputs.

In this Masters of Science thesis a New Zealand case study is used to address this gap in global tsunami research literature with an analysis of the evacuation of Kaikōura residents following the 2016 Kaikōura earthquake and tsunami, which is used to inform the development of a network-based evacuation model. A survey was developed with disaster risk management practitioners involved during the response phase of the 2016 Kaikōura earthquake to enhance understanding of: • Local awareness of tsunami hazard and the need to evacuate before and after the 2016 Kaikōura earthquake • The influence of risk awareness on evacuation-decision-making by residents • Which warnings prompt evacuation actions • Evacuation dynamics including origin-route-destination and congestion barriers • Earthquake and tsunami preparedness actions taken before and after the 2016 Kaikōura earthquake.

A total of 1055 surveys were distributed in Kaikōura township, with 131 returned (12.4% return rate). Key survey findings indicate that prior to the 2016 Kaikōura earthquake, respondents had a ‘good’ (44%, n=55) or ‘very good’ (25%, n=31) level of knowledge on tsunami hazard and the need to evacuate. This contributed to a large proportion (79%, n=88) of respondents who reported that they evacuated due to the potential tsunami threat during and immediately after the 2016 Kaikōura earthquake. The vast majority (95%) of those living in the tsunami evacuation zone at the time (n=64) reported that they evacuated. Almost half (45%) of respondents who were residing outside tsunami evacuation zones (n=42) evacuated, with most reporting that they did so because of ‘severe shaking and aftershocks’. Reported congestion barriers included bridge failures, fallen lamp post, and road damage due to subsidence caused by the earthquake. These barriers slowed down evacuation time of some residents, particularly along Churchill Street and Killarney Street, adding minutes to their total evacuation time. More than half (55%) of respondents evacuated prior to the first tsunami wave was recorded after the earthquake (within 10 minutes). Ninety percent of respondents reported they evacuated by the time at which the tsunami’s highest point was measured on the Kaikōura sea level gauge (40 minutes after the earthquake). The duration time at which survey respondents reported they stayed at their evacuation point ranged from 30 minutes to 48 hours. The most common reported time was 6 hours (18%). Over half (52%, n=45) of respondents reported they returned from their evacuation point when they felt it was safe to do. Thee three commonly reported preparedness actions residents have made following the 2016 Kaikōura earthquake include, prepared an emergency kit with essential items (57%, n=70), discussed an evacuation plan with friends and family or household members (54%, n=67), and prepared a go bag containing essential items (43%, n=53).

Evacuation behaviour and movements from the survey results informed key parameters for developing a network-based vehicular model. The model utilises Kaikōura’s road network and evacuation density to predict evacuation times and congestion potential for future events. Key findings under an ‘ideal’ night-time scenario predicted a total evacuation time of 12:05 minutes for Kaikōura Township to evacuate to safety. This time estimate is a concern for future local-source tsunami threats which could result in tsunami waves arriving at the coast in a matter of minutes. Households in South Bay, located along Avoca Street (west of Kaikōura Peninsula), and north of the township are expected to take the longest time to evacuate (greater than 5 minutes). The response actions reported by Kaikōura residents during the 2016 Kaikōura earthquake aligned well with the results of the network-based evacuation model. Evacuation times, congestion pin points and safe zone capacity of the evacuation model were consistent with the 2016 response actions. However, further analysis could refine the evacuation model to include evacuation delay time, congestion restrictions, and incorporation of both pedestrian and vehicle evacuation.

This thesis A) advances the understanding of a complete evacuation response process of a local-source earthquake and tsunami event in New Zealand and B) develops a network-based vehicular evacuation model informed by realistic evacuation behaviour data and movements of residents following the 2016 Kaikōura earthquake, which can be used to inform evacuation planning for Kaikōura and other coastal communities exposed to local-source tsunami. The methods and results presented in this thesis highlights the critical need for such work to be conducted for communities exposed to tsunami hazard risk, specifically areas susceptible to local source tsunami that would require immediate evacuation. Understanding real-event evacuation response provides valuable opportunities to analyse and inform disaster risk management and emergency response planning to increase national and global tsunami resilience.