HR: 14:55h
AN: S23C-05 [Abstracts]
TI: Extending Probabilistic Seismic Hazard Analysis to Portfolio and Regional Loss Estimation: A San
Francisco Bay Area Example
AU: * Luco, N
EM: nluco@usgs.gov
AF: USGS, PO Box 25046, MS 966, Denver, CO 80225
United States
AU: Wesson, R L
EM: rwesson@usgs.gov
AF: USGS, PO Box 25046, MS 966, Denver, CO 80225
United States
AU: Perkins, D M
EM: perkins@usgs.gov
AF: USGS, PO Box 25046, MS 966, Denver, CO 80225
United States
AU: Karaca, E
EM: ekaraca@usgs.gov
AF: USGS, PO Box 25046, MS 966, Denver, CO 80225
United States
AU: Gupta, N
EM: niting@usc.edu
AF: SCEC, USC, Los Angeles, CA 90089
United States
AU: Cao, T
EM: Tianqing.Cao@conservation.ca.gov
AF: CGS, 801 K St, Sacramento, CA 95814
United States
AU: Harmsen, S C
EM: harmsen@usgs.gov
AF: USGS, PO Box 25046, MS 966, Denver, CO 80225
United States
AU: Field, E H
EM: field@caltech.edu
AF: USGS, S Wilson Ave, Pasadena, CA 91106
United States
AU: Petersen, M D
EM: mpetersen@usgs.gov
AF: USGS, PO Box 25046, MS 966, Denver, CO 80225
United States
AB:
For single-family woodframe dwellings in the nine-county San Francisco Bay Area, we demonstrate a new method for computing a
portfolio or regional loss exceedance probability curve (or ``loss curve'' for short) as an extension of conventional
Probabilistic Seismic Hazard Analysis (PSHA). The method makes use of probabilistic loss given ground shaking relations
(``damage functions'' or ``fragilities'') that are available from various sources, including the CUREE-Caltech Woodframe
Project, HAZUS, and our own relations. Of note, our method can accommodate loss given ground shaking relations that require
multiple measures of ground shaking intensity (i.e., vector hazard), such as the HAZUS relations that use 0.3- and 1.0-second
spectral acceleration.
Whereas PSHA calculates the ground shaking hazard at a site independently of that at any other site, the method we
demonstrate takes into account the correlation in ground shaking across sites that can be affected by a common earthquake
event. This is done by separating the inter-event and intra-event variabilities in ground shaking for a given earthquake
magnitude, source-to-site distance, etc. Not including the correlation in ground shaking underestimates the probabilities of
observing extreme (large or small) aggregate losses.
Just as the ground shaking hazard curves computed via PSHA are uncertain due to various modeling assumptions, so are the
corresponding loss curves. Using Monte Carlo simulation, we quantify the epistemic uncertainty in the loss curves that is
induced by uncertainties in the seismicity and ground shaking models - i.e., the characteristic earthquake magnitude, the
fault slip rate, and the appropriate ground shaking attenuation relation. In addition, we include the epistemic uncertainty
brought about by different loss given ground shaking relations.
While the method we demonstrate can be computationally demanding, particularly if the uncertainty in a loss curve is to be
quantified, we show that it is amenable to parallel computation.
DE: 9810 New fields (not classifiable under other headings)
SC: Seismology [S]
MN: Fall Meeting 2005