The use of empirical vulnerability functions to assess the response of buildings to tsunami impact: Comparative review and summary of best practice

The increased availability of post-tsunami damage data over the past decade has led to a growing inventory of empirical vulnerability functions that model the damage response of buildings to tsunamis. At present, these functions are highly varied with respect to the building typologies and damage states that they represent, making it difficult to apply them effectively in tsunami vulnerability and loss assessments. This paper provides a detailed review of the methods that have been used to develop empirical vulnerability functions and identifies the key factors that are contributing to the variation seen in existing functions. Categorisation and direct comparison of existing functions show that the variation in previous work is due to significant differences in the accuracy of the hazard assessment techniques that are used, inconsistent and poorly defined damage state and building typology classification systems, the use of error-prone and incorrectly applied statistical methods and the use of post-event data that is not representative of the building damage occurring in study areas. As a result of these findings, recommendations for future work include: (1) using more accurate hazard assessment methods; (2) applying more consistent damage state classification methodologies; (3) implementing more precise building typology classification systems; and (4) ensuring that the tsunami-impacted buildings used to generate vulnerability functions are representative of the full damage and hazard range occurring in impacted areas.