Seismic risk management of insurance losses using extreme value theory and copula - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 760

least for one variable, whereas the latter gives a more localised description
of data pairs that are both extreme.
28.3 Insurer's earthquake risk exposure modelling
In this section, a regional seismic loss model for conventional wood-frame
houses in south-western British Columbia, which was developed by Goda
et al. (2011), is focused on. The model is fully probabilistic and evaluated
using Monte Carlo simulation; the analysis framework is illustrated in Fig.
28.1. In the following, key features of the model components are presented
(i.e. seismic hazard analysis and generation of spatially correlated seismic
intensities in Section 28.3.1, and seismic vulnerability and loss assessment
of wood-frame houses in Section 28.3.2). Interested readers are recom-
mended to refer to Goda et al. (2011) for more detailed descriptions.
28.3.1 Seismic hazard analysis and generation of spatially
correlated seismic intensities
Probabilistic seismic hazard analysis offers a rational framework to describe
elastic seismic demand in terms of peak ground motions and response
spectra, and addresses key uncertainties and dependence in earthquake
occurrence (both temporal and spatial), earthquake magnitude, rupture
characteristics, and ground motion intensities (see Chapter 1). The typical
outputs are given by seismic hazard curves and seismic hazard deaggrega-
tion. For Canada, details of the national seismic hazard maps, which form
the basis of seismic design provisions of the current National Building Code
of Canada (NBCC2005), were given by Adams and Halchuk (2003). In this
study, an updated seismic hazard model for western Canada (Goda et al. ,
2010; Atkinson and Goda, 2011) is used to refl ect newly available seismo-
logical information and models. The key improvements include: (i) conver-
sion of different magnitude scales into a uniform moment magnitude scale
in the earthquake catalogue; (ii) re-evaluations of magnitude-recurrence
relations for different earthquake sources based on a longer and homoge-
neous catalogue up to the end of 2008; (iii) use of newer ground motion
prediction equations with proper distance measure conversions (note that
this enables more fl exible adjustment for different site conditions using the
average shear wave velocity in the uppermost 30 m V S30 ); and (iv) consid-
eration of probabilistic mega-thrust Cascadia scenario events (that can be
as large as M w 9.0 earthquake, similar to the 2011 Tohoku earthquake in
size). Overall, adoption of the updated seismic hazard model instead of the
NBCC2005 model decreases uniform hazard ordinates at short vibration
periods (0.1-0.5 s) by 10-30% for the 2% probability of exceedance in 50