Civil Engineering Reference
In-Depth Information
Though PAGER uses simple and intuitive color-coded alerting criteria,
it preserves the necessary uncertainty measures by which one can gauge
the likelihood for the alert to be over- or underestimated. Utilizing the
earthquake impact scale, PAGER's rapid loss estimates, despite their uncer-
tainties, can be used to adequately recommend alert levels and suggest
appropriate response protocols. Requiring or awaiting observations or loss
estimates with a high level of accuracy may delay response and could even-
tually result in increased losses.
31.6
Loss estimation for recent earthquakes
The EIS scheme helps to organize
quantitative
information in a
qualitative
way such that decision makers can take
effective
and
timely
actions under
the conditions of uncertainty. Following a sizable earthquake (with minimum
threshold of
M
w
3.5 within the United States, and
M
w
5.5 worldwide), the
arrival of a ShakeMap automatically triggers the PAGER system to produce
impact alerts.
After one and a half years of operation, we can now provide a summary
of the performance of PAGER's impact-based alerting for a subset of the
events. A complex system such as this requires constant examination,
reanalysis, and updating. The PAGER team has analyzed 746 earthquakes
from September 2010 through September 2011 for potential losses, 70 of
which warranted alerting PAGER email recipients (see
Plate VII
(between
pages 452 and 453)). Of these, 39 had earthquake shaking-based alerts of
'yellow' or greater, indicating an estimated economic loss of 1 million USD
or higher and/or estimated fatalities of one or more. During this period, 707
'green' alerts were generated; in many cases these 'no concern' alerts were
also incredibly valuable for decision-makers, often avoiding unwarranted
attention despite earthquakes with large magnitudes. For the 16 EIS Alert
PAGER runs in this period, the initial Summary Alert level was correct six
times, an over-prediction fi ve times, and an under-prediction fi ve times. The
fi nal Summary Alert level was correct nine times, with most of them resulted
to be high-consequential earthquakes, an over-prediction six times, and an
under-prediction one time. Even when under- and over-predicting, results
were within the neighboring alert level.
Table 31.2 provides a list of earthquakes with estimated and reported
economic losses for some of the signifi cant earthquakes of 2010-2011. There
are large variations in reported economic losses for many earthquakes in
the list.
We used widely cited loss estimates and compared them with the PAGER
estimated median economic losses obtained from the most recent version
of the ShakeMap. In general, the median loss estimates varied from 25%
to 250% of the reported losses and, in most cases, they fall within the proper
Search WWH ::
Custom Search