results that need to
remain confidential until then and is not to be copied or distributed. Paper copies are being made
available for review and for the preparation of related conference talks as needed. Holders of the report
agree to not release the results to anyone. Persons needing a copy should contact Chris D. Poland and
cpoland@degenkolb.com."


When the Big One Strikes Again Estimated
Losses due to a Repeat of the 1906 San
Francisco Earthquake
Charles A. Kircher,
a)

M.EERI,
Hope A. Seligson,
b)
M.EERI,
Jawhar Bouabid
c)

M.EERI,
and Guy C. Morrow
d)
M.EERI
This paper presents interim results of an on-going study of building damage
and losses likely to occur due to a repeat of the 1906 San Francisco earthquake
using the HAZUS technology. Recent work by Boatwright et al. (2006) provides
MMI-based ShakeMap estimates of spectral response accelerations derived from
observations of intensities in the 1906 San Francisco earthquake. This paper
calculates damage and loss estimates using those estimated ground motions,
compares the resulting estimates with those calculated using a method parallel
with that of current seismic provisions of building codes for a magnitude M7.9
event on the San Andreas Fault, and contrasts differences in damage and loss
patterns for these two scenarios. The study region of interest comprises 19
counties of the greater San Francisco Bay Area and adjacent areas of Northern
California, covering 24,000 square miles, with a population of over 10 million
people and about $1.5 trillion of building and contents exposure. The majority of
this property and population is within 40 km (25 miles) of the San Andreas Fault.
The current population of this Northern California region is about ten times what
it was in 1906, and the replacement value of buildings is about 500 times greater.
Despite improvements in building codes and construction practices, the growth of
the region over the past hundred years causes the range of estimated fatalities,
approximately 800 to 3,400 depending on time of day and other variables, to be
comparable to what it was in 1906. The forecast property loss to buildings for a
repeat of the 1906 earthquake is in the range of approximately $90 to $120
billion; 7,000 to 10,000 commercial buildings in the region are estimated to be

a)
Principal, Kircher & Associates, 1121 San Antonio Rd, Suite D-202, Palo Alto, CA 94303
b)
Technical Manager, ABSG Consulting Inc., 300 Commerce Dr., Suite 200, Irvine, CA 92602
c)
Project Director, PBS&J Inc., 415 Fox Chase Circle, Alpharetta, GA 30005
d)
Vice President, Risk Management Solutions, Inc., 7015 Gateway Blvd., Newark, CA 94560 closed due to serious damage; and about 160,000 to 250,000 households
calculated to be displaced from damaged residences. Losses due to fire following
earthquake, as well as losses to utility and transportation systems, would be in
addition to these estimates.
INTRODUCTION
The great earthquake of April 18, 1906, caused widespread damage to San Francisco and
other Bay Area locales, ranging from as far north as Mendocino County to as far south as
Monterey County. The literature for many years has reported approximately 700 to 800
deaths (see for example Freeman, 1932, p. 8) although some recent studies (Hansen et al.,
1989) suggest the life loss may have been approximately four times greater. Direct economic
losses to buildings in San Francisco were about $400 million (Steinbrugge, 1982, p. 298).
Most of these losses were due to the three-day conflagration following the earthquake that
burned over 500 downtown blocks.
In 1906, about 390,000 people lived in San Francisco, and less than 1 million people
lived in the greater San Francisco Bay Area (U. S. Census Bureau, 1995). Today, the number
of San Franciscans has more than doubled, and the Bay Area population has increased ten
fold. In 1906, few buildings were over 10 stories in height; ferryboats crossed the bay; and
horses were still a viable means of transportation. Today, tall buildings and large bridges
spanning the bay define the skyline of San Francisco. Over time, unreinforced masonry
buildings and other highly vulnerable structures have been strengthened, or replaced, by
more seismically resistive construction. However, considering the growth of the region, have
improvements in seismic resistance been sufficient to offset increased risk due to a much
larger population and greatly appreciated property value? This paper explores that question,
and related seismic risk questions, by estimating damages and related losses likely to occur to
the greater Bay Area due to a repeat of the 1906 San Francisco earthquake.
The damage and loss estimates presented in this paper are interim results (as of January
2006) of an on-going 06 earthquake loss study, and these estimates include preliminary
values of direct damage to buildings due to ground shaking and ground failure, as well as
economic and social losses due to these damages. While attention is often focused on the
loss estimates of a study, it is important to document the methods and data used in calculating
those losses. Thus, this paper provides, within the limits of its allotted space in this 1906 earthquake centennial theme issue of Spectra, a relatively extensive discussion of how the
study was conducted.
STUDY OBJECTIVE, SCOPE AND APPROACH
The objective of this study is to comprehensively estimate potential losses to the greater
Bay Area region due to a repeat of the 1906 San Francisco, considering direct damages to the
regions infrastructure caused by earthquake ground motions and ground failure, and induced
(or secondary) damages caused by fire, inundation, hazardous material release, and debris
generation. Infrastructure includes all buildings, essential facilities, and lifeline systems of
the region. This study relies primarily on the Earthquake Model of the HAZUS technology
(NIBS, 1997, 2005, Kircher et al., 2006) to accomplish this scope, since the model provides
the necessary methods for estimating earthquake damage and loss, and default inventory data
describing the infrastructure and demographics of the region. For full appreciation of the
discussion of approach and methodology used for this study, a general familiarity of the
HAZUS Earthquake Model is necessary. This can be obtained from the references above.
Interim results presented in this paper are limited to estimates of direct damage to
buildings due to ground motion and ground failure, and related losses. Building damage and
loss methods are sophisticated in their consideration of building type and material, height,
and design vintage in assessing seismic performance, provided these data are known or can
be obtained (or inferred) from sources such as United States Census Bureau, Dun &
Bradstreet (business-related information), or county tax assessor files. Default inventory
databases of HAZUS are quite extensive but still have inherent limitations. For example, the
demographics of a given census tract are known quite well from census data, and the use or
occupancy and the exposure (i.e., value of buildings) are known reasonably well from census,
Dun & Bradstreet and Means cost data. However, the model building type, which defines the
structural system, is typically not known.
Default inventory databases of HAZUS infer model building type from an assumed
distribution by square footage of the different building types, given occupancy (referred to
herein as the building mapping scheme), and are based largely on the occupancy-building
type relationships of Earthquake Damage Evaluation Data for California, ATC-13 (ATC,
1995). Other key building performance properties that are not known, in general, include
building height and seismic design level. The HAZUS default building-mapping-scheme assumes all model building types to be of low-rise construction and to have a seismic design
level based on a generic mix of buildings (based on an assumed typical distribution of
building age). The assumption that all buildings are low-rise can cause very poor estimates
of damage and loss (e.g., for tall buildings in downtown San Francisco). Likewise, the
generic mix of buildings can also result in very poor estimates of damage and loss, if
buildings (e.g., in the census tract of interest) are significantly older, or significantly newer,
than that assumed by the typical age distribution. Further, default inventory data does not
provide model building types for seismically retrofitted buildings.
A significant effort in this study is the improvement of default data describing building
inventory of the 19-county study region. Specifically, the default mapping scheme is
replaced by 22 custom mapping schemes that better describe actual combinations of model
building type by height and seismic design level throughout the 19-county study region.
These inventory improvements are based on evaluations of building age and density data by
census tract and tax assessor data obtained from an on-going study by Applied Technology
Council for the Community Action Plan for Seismic Safety (CAPSS) of the City and County
of San Francisco (ATC, 2005). Default building properties are also modified to better
represent damage and loss for the most vulnerable building types (e.g., URM, non-ductile
concrete, and soft-story buildings), and new retrofitted model building types are developed to
estimate damage and loss for those model building types that have been seismically
strengthened (e.g., URM buildings).
The study region inventory as well as loss results were reviewed for reasonab