Shannon R. Abeling¹, Matthew Cutfield², Dmytro D. Dizhur¹, Nick Horspool3, David M. Johnston³, and Jason M. Ingham¹
1)  Department of Civil Engineering, the University of Auckland
20 Symonds St, Auckland 1010, New Zealand
{sabe321, ddiz001, j.ingham}@auckland.ac.nz
2 )  Holmes Consulting
39 Market Pl, Auckland 1010, New Zealand
MatthewC@holmesgroup.com
3)  GNS Science
1 Fairway Dr, Avalon, Lower Hutt 5011, New Zealand
{N.Horspool, David.Johnston}@gns.cri.nz

Keywords: Casualty modelling, debris, damage photographs, earthquake damage,

Abstract. The importance of an outdoor casualty model for unreinforced masonry (URM) buildings is apparent when considering that 35 of the 39 people killed in the 2011 Christchurch earthquake as a result of URM building failures were not situated within the URM building that failed. The aim of this study was to provide a preliminary outdoor casualty model for New Zealand simple clay-brick URM buildings based on observations from the 2010 Darfield earthquake (Mw 7.1) and the 2011 Christchurch earthquake (Mw 6.2). Damage photographs from 214 URM buildings from the 2010 Darfield earthquake and 442 URM buildings from the 2011 Christchurch earthquake with different levels of strengthening were analysed to investigate building debris characteristics. From the photographs, four hazard areas and three risk states for each hazard area have been defined. The hazard areas are (1) pedestrian hazard area, (2) vehicle occupant hazard area, (3) escaping occupant hazard area, and (4) neighbouring building occupant hazard area. The risk states for each hazard area were defined to classify each building as having an unlikely (risk state = 0), likely (risk state = 1), or near certain (risk state = 2) risk of generating death or serious injury for a person in the hazard area. An equation was generated to estimate the number of casualties for the pedestrian hazard area. Outdoor casualty rates were derived and model uncertainty was analysed. Large uncertainties in the input parameters of the proposed model led to significant uncertainty in model outputs. Multiple areas for improvement of the model were then discussed and further research was proposed to better calibrate the model and reduce uncertainty in its predictions. Despite drawbacks relating to uncertainty, the proposed model provides
demonstrable benefit as a structured, logical framework to help predict outdoor casualties in seismic loss estimation.