The Spatial Patterns Of
Earthquake Casualties
(Damages) And Social
Vulnerability
Zahra Golshani
Natural Resource & Environmental Science
University of Illinois
Feb,17,2009
Introduction
Natural hazard Loss reduction through
mitigation, preparedness and recovery programs
Social factors play significant role in determining
population vulnerability to hazard (not just
physical nature of hazard)
Increasing disparities in wealth and socio-
economic status increases the potential loss for
greater human loss
Introduction (continue)
Risk modeling has been limited to physical
aspect
Developing integrated models of risk assessment
that would incorporate the social and economic
consequences of earthquakes
Incorporating indirect/non-structural loss to
current models
Objective/Questions
An attempt to assess vulnerability in
spatial term (including both social
vulnerability and physical damage)
Testing damage & social vulnerability
relationship
Does it confirm the literature?
Creating social risk map
Conceptual model
Cutter-1996
Geographic
Risk Context Biophysical
Vulnerability
Hazard
Place Vulnerability
potential
Social
Mitigation Social
Fabric Vulnerability
Defining Social Vulnerability
Some fundamental factors that influence
social vulnerability includes:
Lack of access to resources, including
information and knowledge
Limited access to political power and
representation
Weak building or weak individuals
(Blaiki et al.1994;cutter 1997;Mileti 1999)
Measures Of Social Vulnerable
Population
Characteristics Variable
Differential access to resources Vulnerable-POP
& information/greater (by Age)
susceptibility due to physical Female population
weakness
Non-white (Race)
Low education
Wealth or poverty Per capita Income
Median house value
Data Description
Social Vulnerability Data : all was obtained from
census manipulations were done to obtain the
rates
Vulnerable-POP (people under 14 & over 70)
Female population
Non-white (Race)
Low education (no school through 6 grade population)
Per capita Income
Median house value
Average dollar value of residential damage in
zipcode-89
Data Description-2
some change due to data issues
Casualty (only 33 casualty in Northridge so it was not
possible to do meaningful analysis with that data.
The Turkey casualty data--- only 5% avaliable (census
data are aggregated)
Therefore damage data for Northridge earthquake
was used: Average estimated damage for single and
multi family at zipcode level (Obtained from CPS Report: California
Policy Seminar )
Procedures
Creating damage map (Total damage for
single and multi family houses as the
Physical vulnerability indicator)
Social vulnerability maps (using different
variables)
Compare the results
Aggregate the social map and compare
the results
Method
Using GIS to manipulate and join data
from different sources
Using GIS to create single and integrated
maps and compare the results
Case Study
Northridge earthquake (modest
earthquake)
Jan 17,1994
Physical Damage Map
(LA County-zipcode level)
$ value
Distribution of data is
not normal
Skewed to the right
287 out of 312
zipcode had less than
$1 million estimated
damage
Damage & Poverty-Rate Maps
A closer look
ZIP TotalDam PovertyPCT
90013 26,385 58.12%
90021 22,500 56.38%
90017 15,602 50.45%
90058 11,000 47.77%
90813 26,000 45.63%
Damage & Non_White Rate Maps
A closer look
ZIP TotalDam NonWhitePCT
90305 5,895 95.23%
90008 8,490 93.88%
90043 6,204 91.59%
90047 5,356 91.57%
90746 400 86.84%
Damage & Household size Maps
Closer look
ZIP TotalDam HHSize
90262 - 4.70
90011 10,507 4.65
91733 - 4.60
91744 22,286 4.59
90221 2,000 4.57
Damage & Female population Maps
A closer look
ZIP TotalDam POPFemale
90201 - 52,334
90650 62,889 52,145
90011 10,507 49,331
90280 - 48,548
90250 1,667 48,297
Damage & Low Education
population Maps
A closer look
ZIP TotalDam LowEdu
90011 10,507 20,892
90201 - 16,680
91331 193,814 16,284
90280 - 14,606
90255 13,556 14,090
Household Median value and
Vulnerable population results
ZIP TotalDam HouseMedValue ZIP TotalDam POP_VUL
90014 10,133 35,600 90201 - 37,762
93523 - 45,800 90011 10,507 36,949
93591 - 77,300 90650 62,889 34,649
92301 - 78,800 90280 - 32,891
93560 - 85,400 91331 193,814 32,776
ZIP TotalDam IncomePerCap
90058 11,000 7,359
90813 26,000 7,567
90001 5,775 7,632
90033 11,685 7,775
90003 5,075 7,804
Comparison of social &physical
vulnerability maps
Social vulnerability & damage
relationship (significant at 10%)
Regression Statistics
Multiple R 0.100613275
R Square 0.010123031
Adjusted R Square 0.00685611
Standard Error 0.066538826
Observations 305
ANOVA
df SS MS F Significance F
Regression 1 0.013718993 0.013718993 3.098646142 0.079366325
Residual 303 1.341506846 0.004427415
Total 304 1.355225839
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 0.032705178 0.012492977 2.617885149 0.009292053 0.008121198 0.057289159 0.008121198 0.057289159
Socio_Vul_Rank -0.041997263 0.023858053 -1.760297174 0.079366325 -0.088945714 0.004951187 -0.088945714 0.004951187
Observations/implications
Modeling the spatial pattern of social
vulnerability with GIS does not contradict
the literature
Implications on natural hazard planning
Identifying high risk locations (physical
damage) -mitigation
Targeting social vulnerable groups in
response, relief and recovery phases
Future direction
Running different kind of regressions
including spatial regression on data
Using other physical vulnerability
indicators such as distance to epicenter,
distance to Fault, peak ground
acceleration
Using other variables such as injury
conclusion
The benefit of integration
Improving risk assessment models
Better and more efficient natural hazard
planning
Questions & Comments??
Thanks for your time
have a great afternoon