HISTORY OF
ENGINEERING-BASED
EARTHQUAKE CASUALTY
MODELING
Research Participants & Sponsor
• Hope A. Seligson, Kimberley I. Shoaf,
Corinne Peek-Asa, And Maya MahueGiangreco
• Support for this research was provided
by National Science Foundation Grant
Numbers
CMS-9900062 and CMS-0085314
Definitions
• Engineering-based earthquake casualty models predict
building damage-related casualties (and in some cases,
other types of casualties). These models have typically
been developed by engineers from limited anecdotal,
historical data (not from epidemiological studies, nor
involving health-related researchers).
• These models are typically used for emergency response,
planning and mitigation by government agencies at
various levels, but are less useful for health preparedness
planning.
1970’s: NOAA Scenarios
•
•
•
•
•
NOAA published scenarios in 1972 (SF Bay area) and 1973
(LA area) that estimated building-related casualties.
Tabulated aggregate damage and casualty statistics for
historic earthquakes.
Used generalized casualty rates per 100,000 population
based on previous earthquakes.
“Mystery” ratios of 4:1 serious injuries (i.e., requiring
hospitalization) to deaths and 30:1 minor injuries to deaths.
Included estimates for sidewalk deaths and freeway collapse.
The final results were judgment-based, scenario specific
casualty estimates, rather than a broadly applicable casualty
estimation methodology
1980’s: ATC-13 and Expert Opinion
• In 1985, the Applied Technology Council
(ATC-13) took a more comprehensive look at estimating
building damage for classes of structures using expert
opinion.
• Percent damage and damage state for 17 structural classes
are estimated from Modified Mercalli Intensity (MMI),
similar to earlier work by Whitman, et. al (1974).
• Mean casualty rates associated with damage states, were
applied to the exposed population. Rates based on historic
EQs, previous models and “judgmental evaluation”
• “Mystery” ratios still in use.
ATC-13 Casualty Rates
Damage
State
Slight
0-1
Minor
Serious
Deaths
Injuries
Injuries
3/100,000 1/250,000 1/1,000,000
Light
1-10
3/10,000
1/25,000
1/100,000
Moderate
10-30
3/1,000
1/2,500
1/10,000
Heavy
30-60
3/100
1/250
1/1,000
Major
60-100
3/10
1/25
1/100
100
2/5
2/5
1/5
Destroyed
Range
Rate=30A Rate = 4A Rate = A
Note: for light steel and wood-frame construction,
multiply all numerators by 0.1
1990’s: “State of the Art” Computer
Models - HAZUS (NIBS/FEMA)
• Uses advanced ground motion parameters and detailed
engineering analyses to determine building damage
states and associated damage state probabilities.
Represents a significant advance in the automated
application of loss estimation techniques.
• Indoor and outdoor casualty rates by damage state and
model building type, based on ATC-13 and “limited
historical data” for 4 injury severity levels:
»
»
»
»
Injuries requiring basic medical aid
Hospitalized
Life threatening Injuries
Deaths
HAZUS® Earthquake Loss Estimation Methodology Indoor Casualty Rates (HAZUS®99, SR-2)
Damage
State
Slight
Moderate
Extensive
Complete
(No
Collapse)
Complete
(With
Collapse)
Severity 1 (%)
CASUALTY SEVERITY LEVEL
Severity 2 (%)
Severity 3 (%)
Severity 4 (%)
0.05
0.2 – 0.25
(URM* = 0.35)
1.0
(URM = 2.0)
5.0
(URM = 10.0)
0
0.025 – 0.030
(URM = 0.40)
0.1
(URM = 0.2)
1.0
(URM = 2.0)
0
0
(URM = 0.001)
0.001
(URM = 0.002)
0.01
(URM = 0.02)
0
0
(URM = 0.001)
0.001
(URM = 0.002)
0.01
(URM = 0.02)
40.0
20.0
5.0
(LRWF* = 3.0,
MH* = 3.0,
SLF* = 3.0)
10.0
(LRWF = 5.0,
MH = 5.0,
SLF = 5.0)
Notes:
URM = unreinforced masonry,
LRWF = low-rise wood frame,
HR URMI = high rise steel or concrete frame structures with URM Infill walls,
MH = mobile home,
SLF = steel, light frame,
HR PC = high rise precast concrete structures
More “State of the Art” Computer Models:
EPEDAT
• EPEDAT (Early Post-Earthquake Damage Assessment Tool) was
developed by ABS Consulting/ EQE International for the CA
Office of Emergency Services. It is a GIS-based program designed
to produce regional damage and casualty estimates for emergency
response and planning purposes.
• For casualty models, Beta distribution applied to ATC-13 and
Whitman casualty rates to distribute casualties within range of
potential damage in each damage state (i.e., more injuries with
more damage in a given damage state).
Current Research: Earthquake Data and
Opportunities for Improvement
•
•
•
•
•
NSF (and other funding) allowed researchers (inter-disciplinary team
from UCLA, LA County DHS, and ABS/EQE) to collect and
correlate data from the Northridge and other earthquakes:
building characteristics and damage data
coroner’s data
hospital admission data
ED logs
Survey data on damage and injuries
Research goal: capitalize on the high-quality data to improve the way
engineering-based models estimate building-related casualties, and
make the results more meaningful to health care providers.
Injury
Pyramid:
Northridge
Earthquake
DOA
27
Die in hospital
6
9
Hospitalized/Trauma Cases
Hospitalized/Non-Trauma
129
8,200
16,400
221,400
Emergency Department
Treat & Release
Out of Hospital
Treat & Release
Injured no
treatment
Application of Northridge Data
• Comparison of model prediction to actual Northridge
data to develop “after-market” modifications that
make results more useful to medical community.
• Translates estimates of “Injuries” and “Deaths” to:
–
–
–
–
–
–
Fatalities (non-hospital, i.e., DOA)
Fatalities requiring hospital care (i.e., ICU)
Trauma cases
Non-Trauma Hospital Admissions
ED Treat & Release
Out of Hospital Treatment
Refinements to EPEDAT’s casualty models
for injury planning and response
Pyramid Level
Recommended Model
Description of Injury
Category
Non-Hospital Fatalities
82% of EPEDAT “best
estimate” of Deaths
DOA
In-Hospital Fatalities
18% of EPEDAT “best
estimate Deaths
Majority require intensive
care
Trauma Cases
6.5% of EPEDAT “best
estimate” Serious Injuries
ISS* >15 (some require
intensive care)
Hospital Admits
(Non-Trauma)
93.5% of EPEDAT “best
estimate” Serious Injuries
ISS 15
Emergency
Department (ED) Treat
& Release
16.5% EPEDAT upper
bound Total Injuries
Extremities (esp. in night),
falls, blunt trauma,
lacerations
Out of Hospital
Treat & Release
33% EPEDAT upper
bound Total Injuries
Similar to ED, may spill
over into ED
Additional Research Products
• A literature review of the medical, epidemiological,
public health, and engineering literature. See:
http://www.ph.ucla.edu/cphdr/projects.html.
• Development of a standardized classification scheme for
all aspects of earthquake-related casualties (e.g., injury
mechanism, building damage). See:
http://www.ph.ucla.edu/cphdr/scheme.pdf
• An integrated review of available casualty and damage
data (e.g., Northridge, Kobe, Nisqually EQs) classified
according to the new classification scheme - in progress.
Conclusions
• Engineering-based casualty models allow for rapid
estimation of regional population impacts for response,
planning and mitigation purposes. While many advances
have been made in the area of loss estimation, casualty
modeling has not received the attention dedicated to the
development of other model components.
• Future enhancement of the such models will benefit
greatly from coordinated data collection and analysis, as
well as inter-disciplinary research incorporating medical
and public health perspectives. This integrated approach
will facilitate the use of data from recent and future
events to refine engineering-based casualty
models.
Standardized Classification Scheme
for Earthquake-Related Injuries
Purpose of Standardized Classification
Scheme
• To establish a systematic, multi-disciplinary
and collaborative approach to the study of risk
assessment, loss estimation for earthquakes
• To create a common language to define the
event, the victims and responses for any given
earthquake
• Reduce variability of data for reported deaths
and injuries
Components of Classification Scheme
Individual
Building
Level
Level
Demographics Building
Description
Injury
Building
Damage
Location
Activity
Hazard Level
Earthquake
Source
Local Site
Hazard
Use of Existing Measures
• Abbreviated Injury Score
• International Classification of Diseases,
9th. Revision
• ATC 20
Hazard Level Variables
• Earthquake Source
–
–
–
–
–
–
–
–
–
Earthquake Name
Event Number or ID
Magnitude
Magnitude Scale
Date
Time
Day of Week
Earthquake Location
Rupture Length
– Rupture Area
– Presence of Surface
Rupture
– Deepest Point of Rupture
– Shallowest Point of
Rupture
– Fault Source
• Local Site Hazard
– Earthquake ground
motion
– Local Site Conditions
Building Level Variables
• Building Description
–
–
–
–
–
–
Structural System
Building Height
Building Size
Building Year
Seismic Design Quality
Debris Generation
Potential
– Occupancy Type
– Estimated Occupancy
– Actual Occupancy
• Building Damage
– Building Safety
Inspection Status
– Safety Tag
– Dollar Damage
– Damage Percent
– Damage State
– Building Collapse
Individual Level Variables
• Demographics
–
–
–
–
–
–
–
Age
Gender
Race/Ethnicity
Level of Education
Occupation
Income
Disabilities and Preexisting Conditions
• Injury Characteristics
– Cause of Injury
•
•
•
•
Relation of EQ
Structural Relatedness
Secondary Hazards
Injury Mechanisms
– Injury Severity
– Treatment
• Level of Treatment
• Immediacy
– Diagnoses
– Costs
• Direct Medical Care Costs
• Indirect Costs
Individual Level Variables, cont.
• Location
– Injured individual’s
physical location
– Injured individual’s
geographic location
• Activity
– Starting position
– Activity