Catastrophe Risk Analysis: A Case Study
Adityam Krovvidi
General Manager
Workshop on Disaster Management
October 1, 2004, New Delhi
Delivering a world of solutions
Delivering a world of solutions
www.rmsi.com
Introduction
Historical Perspective
Modeling Catastrophes
Case Study
Conclusions
Application Software
Development
Delivering a world of solutions
Presentation outline www.rmsi.com
Introduction
The objectives are to discuss:
– fundamentals of catastrophe risk analysis,
– profile of natural Cat risks in India and
– tools and techniques available to assess Cat risks
Definition of Catastrophe
– Sudden disaster that causes many people suffer
(Oxford dictionary)
Catastrophic perils
– Natural hazards: cyclones, earthquakes & floods
Cataclysmic events
– causing disasters in
terms of wide spread
loss of life & property
Delivering a world of solutions
Introduction www.rmsi.com
Introduction
Historical Perspective
Historical Perspective
Modeling Catastrophes
Case Study
Conclusions
Application Software
Development
Delivering a world of solutions Source: IS: 1893-2002
Presentation outline www.rmsi.com
Historical Cyclones
Events catalogue
– RMSI compiled and cleaned (major source: IMD)
» Track data: 1891-2000
» Parametric data: 1950-2000
• Central pressure, Forward velocity, Radius to max wind
Frequency and severity
– Av. 5-6 per year in NIO (global av. = 80)
– 2-3 make landfall on Indian coast
» CAT 5/4:3/2:1/0 = 1:4:9
» East & west coasts ratio = 10:1
Consequences
– 1975-2001 casualty data suggests (source: EM-DAT,
Universite Catholique de Louvain, Belgium)
» Av. 1460 per year
» Av. 474 per event
– 2001 projected economic losses
» 1999 Ersama, Orissa (CAT 5) = USD 1.5 billion
» 1990 Machalipatnam, AP (CAT 3) = USD 1.1 billion
» 1977 Chirala, AP (CAT 5) = USD 1.0 billion
Delivering a world of solutions
Historical Perspective www.rmsi.com
Historical Earthquakes
Events catalogue
– RMSI updated, cleaned and completed
» Sources: ISET, USGS
» Dates back from 2001 through history of known events
Frequency and severity (catalogue 1800-2001)
– Av. 2-3 per year with M=>5.0
– Indicative return periods for India
» 5.08.0 = 30 years
Consequences
– 1975-2001 casualty data suggests (source: EM-DAT,
Universite Catholique de Louvain, Belgium)
» Av. 2924 per year
» Av. 2010 per event
– 2001 projected economic losses
» 2001 Bhuj, Gujarat (M6.9) = USD 3.0 billion
» 1993 Latur, Maharashtra (M6.3) = USD 225 million
Delivering a world of solutions Source: IS: 1893-2002
Historical Perspective www.rmsi.com
Historical Floods
Events catalogue
– RMSI compiled 1896-1996
» Sources: CWC, UNESCO
» Gaps and heterogeneities exist
Frequency and severity (catalogue 1800-2001)
– Major floods 3-4 every year (catalogue 1980-2001)
Consequences
– 1975-2001 casualty data suggests (source: EM-DAT,
Universite Catholique de Louvain, Belgium)
» Av. 1297 per year
» Av. 340 per event
– 2001 projected economic losses (source: CWC)
» Average USD 280 million per year (1953-2000)
» 1998 Ganga & Brahmaputra; UP, WB, Assam, Bihar =
USD 1.2 billion
» 1986 Godavari, AP = USD 872 million
Source: Vulnerability Atlas
Delivering a world of solutions
Historical Perspective www.rmsi.com
Introduction
Historical Perspective
Modeling Catastrophes
Modeling Catastrophes
Case Study
Conclusions
Application Software
Development
Delivering a world of solutions
Presentation outline www.rmsi.com
Modeling Catastrophe Risk
RMSI probabilistic risk modeling framework
– Comprises 4 standard modules
– Modules work in a funnel fashion
» Output of one is input of next
Example of cyclone risk modeling follows…
Delivering a world of solutions
Modeling Catastrophes www.rmsi.com
Modeling Cat Risk – Stochastic Module
Coastline segmentation
– The 50 nmi gates capture the complex
orientations
Simulation of events on each gate
– Develop CDFs for cyclone parameters
» Central pressure
» Forward velocity
» Angle of landfall
– Stratified sampling of CDFs
– Events defined by random matching of
parameters
– Pattern matching with historical tracks
4800
Andhra Pradesh
Stochastic events
Delivering a world of solutions
Modeling Catastrophes www.rmsi.com
Modeling Cat Risk – Hazard Module
Georgiou’s (1985) windfield model
Model parameters
– Pressure drop
– Forward velocity
– Track angle with site
– Radius to max wind
– Distance to site
Calibration of coefficients
– Historical storms reconstruction
Directional roughness
Peak gust wind speed at site
Validation
Delivering a world of solutions
Modeling Catastrophes www.rmsi.com
Modeling Cat Risk – Vulnerability Module
Domestic
Published Inventory
Research
Vulnerability
+ Base +
Functions
vulnerability Engineering
Damage data Function + review
MDR (%)
from (Vul. Atlas, IS
Event recon. (composite)
codes)
+ +
Peakgust
Benchmark
Reported loss
curves
data
(Intl. experience)
Delivering a world of solutions
Modeling Catastrophes www.rmsi.com
Modeling Cat Risk – Financial Module
Exposure
– Total value or replacement cost of assets that is at risk
– Valuation at 2001 prices
Loss = Exposure x MDR
Event Loss Table (ELT)
– Event
– Loss
– Probability
Financial analyses
Delivering a world of solutions
Modeling Catastrophes www.rmsi.com
Introduction
Historical Perspective
Modeling Catastrophes
Case Study
Conclusions
Application Software
Development
Delivering a world of solutions
Presentation outline www.rmsi.com
Case Study - Introduction
A World Bank initiative
Study objectives
– Risk identification
– Risk assessment
– Risk representation
Scope
– Four states: AP, OR, GJ, MH
– Three perils: Cyclone, Earthquake, Flood
– Assets: Housing and key public infrastructure
Model resolution: Block
Deliverables
– A comprehensive report
– Detailed exposure & loss results
– Risk maps
Delivering a world of solutions
Case Study www.rmsi.com
Case Study – Cyclone Model Validation
4,000 Modeled Loss (Crore Rs.)
3,500 Observed Loss (Crore Rs.)
3,000
2,500
2,000
1,500
1,000
500
-
1977 Chirala 1979 Ongole 1990 1999 Orissa
Machilipatinam
Delivering a world of solutions
Case Study www.rmsi.com
Case Study – Hazard Map
Cyclone hazard map
– Wind speed
– Rainfall
– Storm surge
Return periods
Delivering a world of solutions
Case Study www.rmsi.com
Case Study – Exposure Map
Exposure mapping
– Housing
– Public infrastructure
» Roads & bridges
» Educational institutions
» Medical facilities
Block and district levels
Delivering a world of solutions
Case Study www.rmsi.com
Case Study – AAL Map
Average annual loss (AAL)
– AAL is the expected loss per year when
averaged over a very long period
Block and district levels
Delivering a world of solutions
Case Study www.rmsi.com
Case Study – EP Plots
Exceeding Probability (EP)
– EP curves are cumulative distributions that
show the probability that losses will exceed a
certain amount
AEP/OEP
Delivering a world of solutions
Case Study www.rmsi.com
Case Study – Loss Cost
Risk modelers consider loss cost as AAL per thousand dollars of exposed value.
The major advantage of loss cost over AAL is that it can be compared across
perils, coverages, geographies, etc.
FL JP AP GJ OR
Cyclone 1.56 0.27 1.55 0.76 3.22
CA JP GJ MR
Housing damage potential Earthquake 2.51 1.67 0.52 0.05
compared globally
OR cyclones have damage
potential double that of Florida
hurricanes
AP cyclones have the same
potential as Florida and GJ’s
potential is one-half of AP
GJ earthquakes have damage
potential 10 times more than
that of Maharashtra. However, it
is 3 and 5 times lower than
Japan and California
Delivering a world of solutions
Case Study www.rmsi.com
Case Study – PML
There is no common approach or unified definition to evaluate probable
maximum loss (PML). Since developing economies cannot afford to plan for a
high risk tolerance a 150 year PML is suggested
PML as % of exposure
– AP = 2.1%
– GJ = 2.1%
– MR = 0.1%
– OR = 3.2%
GJ needs $1 billion for Cat
risk preparedness, closely
followed by AP
Delivering a world of solutions
Case Study www.rmsi.com
Introduction
Historical Perspective
Modeling Catastrophes
Case Study
Conclusions
Application Software
Development
Delivering a world of solutions
Presentation outline www.rmsi.com
Conclusions
India has catastrophe risks comparable to developed nations
The recent events – 1998 floods, 1999 Orissa cyclone & 2001 Gujarat
earthquake – are a case in point
Potential for a one billion dollar (Rs.4,500 crore) loss is high
Cat risk models estimate potential losses reasonably well
Fortunately data and capabilities are available within the country
Industry efforts are the need of the hour to take the World Bank’s
initiative forward
Delivering a world of solutions
Conclusions www.rmsi.com
Delivering a world of solutions
info@rmsi.com
www.rmsi.com
Delivering a world of solutions
www.rmsi.com