Spatial correlation of ground motions in estimating seismic hazards to civil infrastructure - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 70

mean and the median values of possible losses are not equal to each other
in this case.
The within-earthquake variability, for every particular earthquake sce-
nario, leads to fl uctuations of ground motion from one location to another;
stronger than the median estimates at one site and weaker at another site.
Even if the large level of ground motion causes great damage at some loca-
tions, the damage at other locations may be small. Thus, if all of the ground
motion variability is treated as within-earthquake ( ρ η = 0.0; ρ ( Δ ) = 0.0 for

Δ 0 km), the variability of total loss would be smaller than that for the

other extreme case (all the variability is between earthquakes) analyzed
above. However, the median values, which are affected by particularly great
losses, are larger than those for the case of between-earthquake variability
only. Note that the mean and the median values of possible losses are equal
in this case.

>

3.3.2 Recent trends for seismic loss estimation studies
considering ground-motion correlation

The knowledge of the loss distribution about the mean (e.g. variance or
standard deviation) is very important for decision making and mitigation
activities. For example, primary insurers are concerned with the central part
of the distribution (mean and median values), while re-insurers deal mostly
with the tail of the distribution. Several studies analyzed the infl uence of
ground-motion uncertainty and correlation on characteristics of loss distri-
bution for widely located building assets (portfolios) and spatially distrib-
uted structures (lifelines). Specifi c (scenario) earthquakes were considered
by Lee et al. (2004), Lee and Kiremidjian (2007), Molas et al. (2006), Goda
and Atkinson (2009), Crowley et al. (2008a,b), and Sokolov and Wenzel
(2011a). Multiple earthquakes and loss probability curves were analyzed by
Wesson and Perkins (2001), McVerry et al. (2004), Bommer and Crowley
(2006), Park et al. (2007), Goda and Hong (2008b, 2009), and Sokolov and
Wenzel (2011b).
Seismic loss estimation was performed using the techniques based on
different ground-motion parameters, namely: modifi ed Mercalli intensities
(McVerry et al. , 2004; Molas et al. , 2006), peak ground acceleration (Wesson
and Perkins, 2001; Sokolov and Wenzel, 2011a,b), and spectral acceleration
or spectral displacement (Lee et al. , 2004; Crowley and Bommer, 2006; Lee
and Kiremidjian, 2007; Park et al. , 2007; Crowley et al. , 2008a; Goda and
Hong, 2008b), and using various descriptions of loss, i.e, the mean damage
ratio (Wesson and Perkins, 2001; Lee et al. , 2004; Crowley and Bommer,
2006; Lee and Kiremidjian, 2007; Crowley et al. , 2008a; Goda and Atkinson,
2009; Sokolov and Wenzel, 2011b) and monetary loss (McVerry et al. , 2004;
Park et al. , 2007; Goda and Hong, 2008b, 2009; Sokolov and Wenzel, 2011a).