"He was keen to go beyond just knowing about building damage and focus in on more of the likely community impacts, including the likely habitability – or the ability of residents being able to return home - of residential areas potentially inundated by tsunami," GNS risk specialist Finn Scheele said.
"This was a really important lesson from the Canterbury earthquakes and so disaster risk scientists have been working hard since to model more useful and usable impacts to help decision-makers."
In their just-published study, Scheele and colleagues used three modelled scenarios.
The most likely large event to affect Christchurch – with a 0.04 per cent chance of occurring in a given year, or a 3.4 per cent chance of happening in someone's lifetime – was a tsunami forming from a quake on the South American subduction zone.
Another large but less likely scenario was an event like an 1868 tsunami, which reached New Zealand but caused limited damage due to much lower levels of development and population at the time.
The third scenario – and the one most likely overall – was a tsunami about half the size of the large South America tsunami.
Focused on a low-lying area of Christchurch, taking in about 15,400 homes, the study used a specially-designed model looking at how many houses would be left habitable, and how many people would require shelter.
"We used 2013 Census data for demographic information of the population within the areas affected by tsunami inundation," Scheele said.
"For example, across many previous disasters internationally it has been observed that renters are more likely to be displaced than home owners, and low income households are more likely to require emergency sheltering."
"Each of the factors included have an influence on the model to various degrees, but physical factors, such as building damage and utility outage, are the most important."
Under the two large scenarios, more than 14,000 people were estimated would be displaced on the first day of the event – with between 1700 and 1800 needing shelter.
"This would be due to evacuation orders by authorities, and the actual number in a real event may differ because the inundation extent will be unknown before the tsunami arrives."
Four days later, up to around 10,000 people would still estimated to be out of their homes, with more than 1200 still requiring temporary shelter.
After a month, about two-thirds of all affected residents would be able to return to their homes – but the rest would still be displaced.
In the smaller modelled event, about 69 per cent of people were estimated would be able to return to their homes after a week – and 86 per cent within a month – but more than 600 would still need shelter on the first day.
Scheele said people being forced out of their homes would have ongoing implications for disaster response and recovery.
But he cautioned the model results were to assist planning, and shouldn't be taken as the expected outcome for an actual event.
"For emergency management planning the modelling helps provide guidance on the number of people who will need ongoing accommodation after a tsunami and some guidance as to how long it may take before residents can return to their homes," he said.
"Tsunami are complex events, and this model can assist decision-makers by quantifying and mapping potential impacts over time."
Scheele said the modelling methods could be applied to other coastal communities at risk – and research was also under way to adapt it to other hazards like earthquakes and eruptions.