Course Unit 3 Flood Risk
Flood Risk Assessment, Uncertainty, Management H.P. Nachtnebel
Structure of presentation
Objectives and introduction
Methodological concept
Risk assessment (options and uncertainties)
Risk management and mitigation strategies
Conclusions
Objectives
Demonstration of flood risk assessment and of flood risk management options
considering trends and
various sources of uncertainties
Observations: Flood damages
are the most frequent and costly natural hazard (Jongman et al., 2012; UNISDR, 2011).
the respective economic damages are about $19 billion/a (Kundzewicz, 2010) and more than 115 million people/a are affected globally
have increased in most regions of the world during the last decades (de Moel et al., 2009, Barredo, 2009; Bouwer et al., 2010; Kreft, 2011; UNISDR, 2011).
This fact is surprising because many countries, especially in Europe, have annually invested substantial amounts in
physical flood protection measures, such as levees, dykes and flood detention reservoirs.
Observations: Flood damages
are the most frequent and costly natural hazard (Jongman et al., 2012; UNISDR, 2011).
the respective economic damages are about $19 billion/a (Kundzewicz, 2010) and more than 115 million people/a are affected globally
have increased in most regions of the world during the last decades (de Moel et al., 2009, Barredo, 2009; Bouwer et al., 2010; Kreft, 2011; UNISDR, 2011).
This fact is surprising because many countries, especially in
(Munich Re)
Flood trends
lack of evidence and thus low confidence regarding the sign of trend in the magnitude and/or frequency of floods on a global scale over the instrumental record.
With high confidence, floods larger than recorded since the 20th century occurred during the past five centuries in northern and central Europe, the western
Mediterranean region and eastern Asia.
(5th IPCC Assessment report)
The Risk Management Cycle
Consultation Risk Analysis
Recovery and Post Disaster
Works Flood Event
Management Flood
Prepardness
What is Reliability, Failure, Risk ?
Reliability: the probability, that a system serves its purpose
Failure: the probability that a system does not serve its purpose
Risk: The potential for realization of unwanted, adverse consequences from a hazard to human life, health,
property, or the environment
Definitions: Reliability and Failure
Resistance (Design Level) Load Q X
Q is a random variable with pdf f(Q) Reliability: Z(X*)
X* f (Q)dQX*
Some Definitions
Hazard
Consequences (Damages)
unwanted, adverse consequences from a hazard to human life, health, property, or the environment
Adverse consequences, especially for human health and life, the environment, cultural heritage, economic activity and infra-structure (EU-FRD 2007)
The damage D(Q) is based on exposure and vulnerability:
Consequences (Damages)
unwanted, adverse consequences from a hazard to human life, health, property, or the environment
Adverse consequences, especially for human health and life, the environment, cultural heritage, economic activity and infra-structure (EU-FRD: 2007)
The damage D(Q) are based on exposure and vulnerability:
exposure of populations and property (who and what)
Consequences (Damages)
unwanted, adverse consequences from a hazard to human life, health, property, or the environment
Adverse consequences, especially for human health and life, the environment, cultural heritage, economic activity and infra-structure (EU-FRD: 2007)
The damage D(Q) are based on exposure and vulnerability:
exposure of populations and property (who and what)
Definition of the risk
Floods (load or hazard) Q and probability distribution function (pdf) f(Q)
Loss function (potential damages) D (Q)
Risk R is an expectation value
Damage function dependent on Q flood probability
0
) ( )
(
() f Q D Q dQ
R
Flood Risk Assessment
What is a flood ?
Define the flood probability
Define the flood impacts (exposure, vulnerability)
Estimate the risk
Identify risk reduction measures
Risk Elements
A hazardous event
A cumulative distribution function F(Q)
The consequences (damages, victims,..) expressed by D(Q)
F (Q)
Q
Potential Damages D (Q)
Q X*
old
Q* Q*
Hazard Analysis
Estimation of the frequency and magnitude of flood events
Annual flood series (annual flood maxima)
Partial duration series (all events above a threshold level)
What is a flood ? Annual series
The largest value in each year constitutes a flood event
What is flood ? Partial duration series
All peak above a threshold level are identified as floods But ensure independency among events
z.B. 1991 a minimum time interval among events
Estimation of the flood probability
Relationship between flood peak and probability
f(Q) is the density function of flood events and F(Q) is the cumulative distribution function (the integral of f(Q))
f(Q)
Q Q
0
F(Q) 1
Estimation of flood probability
A set of flood events has been measured
A model is selected (e.g. Gumbel distribution)
Fitting of the model to data
Estimating the magnitude of rare events
Estimating the uncertainty
Gumbel distribution
2-parametric (a and c which are related to mean and std. deviation
double exponential distribution
leftside bounded, right side unbounded
Equation of a straight line
log _ 1 _
1 )
( take the
e T x
x
F c
xT a
e
T
) 1 ( _ _
log_
_ 1 _
1
ln
take the and multiply by
e c T
x a T
T c
x
a T 1
1 ln ln
A simple example given a data set
Ranking of floods in descending order Assign a return period T
Plot these data in diagram
year 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
Qmax 342 415 199 278 512 333 395 607 212 437
Rank 6 4 10
8 2 7 5 1 9 3
T 1,7 2,5 1 1,3
5 1,4
2 10 1,1 3,3
Probability paper (model) graphical approach
Wahrscheinlichkeitspapier für Gumbel-Verteilung
0 20 40 60 80 100
-2 -1 0 1 2 3 4 5 6 7
reduzierte Variable yT
X
1.001 1.01 1.1 1.2 1.5 2 3 4 5 10 25 50 100 200 300 400 500 1000
Wiederkehrintervall
0.1 1 50 75 80 90 96 98 99Unterschreitungswahrscheinlichkeit [%]99.8 99.9
Modus Mittel
200 300 400 500 600 700
800 900 1000 1100
925
??
Q(m3/s)
General statements
Sample size (# of observed flood events) is in general small, such as several decades.
Thus, extrapolation (estimation of rare events with return period T) should not exceed 3 times the length of
observation
Provide information about the uncertainty in estimating a rare event
Numerical approach:
Estimation of X
Tand Uncertainty DX
Tyear 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
Qmax 342 415 199 278 512 333 395 607 212 437
Rank 6 4 10
8 2 7 5 1 9 3
T 1,7 2,5 1 1,3
5 1,4
2 10 1,1
3,3 928 m³/ s
3 , 128
* 323 , 4 373
x
T x K T s
x ( )*
Estimation of parameters of the sample:
x, sx
x= 373 (m3/s), sx= 128,3 (m3/s)
Estimation of the magnitude of flood with return period T is similar to estimation of a quantile in a normal distribution
K(T) is tabulated for Gumbel
sample size K(T)
Numerical approach:
Estimation of the uncertainty of the estimate
The estimation of any parameter is associated with uncertainty (parameter has always a probability distr.)
More data would be helpful to reduce this uncertainty
It is easier (less unertain) to estimate a mean value than a „rare event“, e.g. xT.
The larger T the larger is the uncertainty
To estimate confidence intervals the confidence level
Numerical approach:
Estimation of the uncertainty of the estimate
specifies the confidence level, usually 95 %
For a standardized normal distribution 95% of the values are within +/- 1,96
For a Gumbel distributed variable 95% of the values are within
All the values are already known and we obtain
n u s
x s
u
xT ()* T T ()*T x
* 2
1 , 1
* 14 , 1
1 T T
T K K
s m
s
m ³ / 409 ³ / 928
78 , 208
* 960 ,
1
928
How to Evaluate the Damages ?
Typology of flood damages
(Messner et al. 2006, Penning-Rowsell et al. 2003, Smith and Ward 1998)
Measurement
Tangible Intangible
Form of damage
Direct
Physical damage to assets:
Buildings Contents Infrastructure
Loss of life Health effects
Loss of ecological goods
Impact Assessment (ex post and ante)
Ex post:
after a flood event document who and what has been hit how strongly by the flood
Make an inventory of all documented damages (fatalities, losses,..)
Ex ante:
Derive inundation maps for different hazardous events
Identify exposed number of people and objects
Analyse the vulnerability of people and objects
Estimate potential fatalities, damages
Ex ante procedure
Generation of possible flood events (hydrology)
Establish a DTM
2D-hydraulic model to calculate propagation of flood in the project area
Calculation of inundated area, water depth and flow velocity
Overlay with cadastre map
From Laser scan data to a Digital Terrain
Model (DTM) by mesh generation
Comparing a DTM with Areal Photos
Consideration of Cross Sections
is very helpful in generation the DTM
Application of a Hydraulic Model
Initial conditions: water depth and flow velocity at t=0 at every location is given
Boundary conditions: Inflow hydrograph is given
Model parameters: roughness coefficients for each element are given (estimated)
Results from the Hydraulic Model
Water depth and flow velocity at each location (grid element)
Delineation of inundated areas and boundaries of inundation (basis of exposure)
Which scenarios (discharges) ? EU Flood risk directive
a frequent flood HQ30 a HQ100
an extreme event HQ300
Exposed Objects for HQ 30/100/300
Damage Estimation
Classify objects (one family houses, multi familiy houses, farms, garages, companies, enterprises,
infrastructure…)
Estimate the value of the object and its vulnerability
Companies: need individual analysis
Damages
Damages
Damages
Damages
Damages
Property damages
Building, heating systems, electric and electronic infrastructure.
Vehicles
Goods, products, stock levels Operating equipments, ...
Loss due to service interruption: losses in sales volume and profit Location disadvantages
Environmental consequences
Classification of Damages of Enterprises
Vulnerability of Objects and Uncertainty
On site inspections
Different set of loss functions are available (absolute or relative values)
Damage estimates are subjected to a large uncertainty
Example HOWAS database (Merz et al., 2004)
Interim summary
Procedure for f(Q) and D(Q) has been explained
Estimation variance (uncertainty) has also to be assessed
An Example:
Ex ante flood assessment of a city in Southern Austria
Collecting observations
Generating scenarios
Analysing scenarios
Example
Flood area before implementation of
flood control structures Raab: Qmax = 200 m3/s Rabnitz: Qmax = 40 m3/s probability: ~1/100 p.a.
ZT Turk 1995 & 1997
Development
Land survey 1787
GIS Styria, http://www.gis.steiermark.at/07 -2005
Dykes
Flood reservoir
Reservoir outflow
Inflow to reservoir
Dykes
Flood protection project 97-99
Analysis of the Flood Series
Flood series Feldbach
0 50 100 150 200 250
1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 Annual maxima of the drain Q [ m 3 /s]
HQ
Trend straight
3 /s]
Analysis of the Flood Series
Flood series Feldbach
Flood series Takern
50 100 150 200 250
Annual maxima of the drain Q [ m 3 /s]
HQ
Trend straight
80 100 120 140 160
of the drain Q [ m 3 /s]
HQ
Trend straight
Scenario 2
Flood areas, Depths
Raab: Qmax = 200 m3/s Rabnitz: Qmax = 40 m3/s probability: ~1/100 p.a.
Scenario 3
Existing flood protection Depth of inundation
log jam at the bridge
Scenario 4
inundation area and depth
Raab: Qmax = 245 m3/s Rabnitz: Qmax = 56 m3/s flood probability:~ 1/300
Scenario 5
Inundation area and depth
Raab: Qmax = 310 m3/s Rabnitz: Qmax = 82 m3/s flood probability: ~ 1/1000
Scenario 6
Inundation area and depth
Raab: Qmax = 400 m3/s Rabnitz: Qmax = 97 m3/s flood probability: ~ 1/5000
Exposed Objects for Different Scenarios
# of endangered objects
total
Damage Potential
Method to BUWAL (1999) & BWG (2002)
Converted & discounted Austria p., 2004
Damages in €/building & damages in €/m2
per building per m2
Medium Intensity h> 0,5 m
Classification scheme
Single familiy houses
Appartment buildings Small/med enterprises Industrial firms
stables
Storage buildings
Low Intensity h< 0,5 m
Estimated damages (€)
per building per m2
Damage Potential in Industrial Sector
Damage types
damages of property losses in production
Competition disadvantages subsequent damages
...
Damage Potential in Industrial Sector
Results from interviews 10 companies responded
among them the 4 largest ones:
Management and insurance companies are interested
one company: internal mitigation measures
some of them have an insurance: property and losses in production
sensible topic (image losses when the companies vulnerability would be identified)
difficult to get reliable response from the comapnies
Estimated Total Damages
Interim summary:
A method has been demonstrated to estimate ex ante the flood damage potential
Exposure, damage functions were identified
Often, secondary damages dominate direct damages
What happens after building a levee ?
Land use will change
More people will settle in the former flood plain
More houses will be built
The value of properties increases
The damage potential increases
Risk is changing with time
What happens when land use changes (e.g. population density increases)
f (Q)
Q
Damage potential D (Q)
Q X*
old new
65
Risc curves
Q Damage D (Q)
F(D’>D) Q
Risk curves
Flood probability f(Q)
Q
Damage potential D(Q)
Q
X*(T)
Damage potential D(Q) Prob(Damage>D)
1/T*
Risc curve without a levee with a levee
with intensified land use
Design level X*
Consequences
The expected damages may be larger after implementation of flood protection measures
Land management and development strategies are required
Reliability of protective measures
Consequences
The expected damages may be larger after implementation of flood protection measures
Land management and development strategies are required
Safety of levees ?
Protective structures may fail already before the critical load is reached
Risk Management
Risk management compares different alternatives, quantifies them and ranks them
Assist in selecting a preferred alternative
EU-FDR asks for measures which
Reduce existing risks
Avoid the emergence of new risks
The EU-FDR asks for consideration of non-strcutural and structural measures
Design of protective structures
Specification of failure levels
No protection for agricultural land
Protection of residential areas at least against HQ100
Protection of densily populated areas against HQ300
Protection of sensible infrastrcture against HQ300
Leads to the definition of the residual or remaining risk
Definition of the remaining risk
Design level for a dyke X* (resistance)
Remaining risk R (X*) because of exceedance of X*
Loss function flood probability
X* is the design value
*
) ( )
(
*) (
X
dQ Q
D Q
f X
R
Design of protective structures
Specification of failure levels
No protection for agricultural land
Protection of residential areas at least against HQ100
Protection of densily populated areas against HQ300
Protection of sensible infrastrcture against HQ300
Leads to the definition of the residual or remaining risk
Risk based design
Any protective measure has costs and reduces damages
Minimize total costs: Min { C(h) + R(h)} h*
Options for Risk Mitigation
Possible decisions refer to
Reducing damages Actions Ai to control D(Q):
• Revise building codes
• Harmonisation of risk maps with local/ regional development
• Early warning systems
• Raising awarness about risk exposure
• Avoid secondary damages
Options for Risk Mitigation
Possible decisions refer to
Changing pdf Actions Ai to control f(Q):
• Increase natural retention capacity
• Consider surface and groundwater systems
• Reduction of the uncertainty in f(Q)
• Consideration of human interventions
• Consideration of sediment transport and discharge
Options for Risk Mitigation
Possible decisions refer to
Changing protection level Actions Ai to control X*:
•Increase the reliability of the resistance
•Temporary protection systems
•Dikes require spillways like dams to protect the dike from collapse and to
ensure a controlled flooding and drainage
New Orleans
Options for Risk Mitigation
Possible decisions refer to
Changing protection level Actions Ai to control X*:
•Increase the reliability of the resistance
•Temporary protection systems
•Dikes require spillways like dams to protect the dike from collapse and to
ensure a controlled flooding and drainage of the floodplain
New Orleans
Options for Risk Mitigation
Possible decisions refer to
Changing protection level Actions Ai to control X*:
•Increase the reliability of the resistance
•Temporary protection systems
•Dikes require spillways like dams to protect the dike from collapse and to
ensure a controlled flooding and drainage
New Orleans
Options for Risk Mitigation
Possible decisions refer to
Changing protection level Actions Ai to control X*:
•Increase the reliability of the resistance
•Temporary protection systems
•Dikes require spillways like dams to protect the dike from collapse and to
ensure a controlled flooding and drainage of the floodplain
New Orleans
Options for Risk Mitigation
Possible decisions refer to
Risk transfer Actions Ai to control R(X*):
• Insurance system vs catastrophic funds
• Clear seperation of responsibilities among individual and public authorities
• Risk zonation and individual responsibilities
Conclusions
Strategies are needed which are reasonable in the short and the mid term
Conclusions
Strategies are needed which are reasonable in the short and the mid term
+ communication of hazards
+ removing of highly vulnerable objects (hospitals, Kindergarden, chemical firms
Conclusions
Strategies are needed which are reasonable in the short and the mid term
+ communication of hazards
+ removing of highly vulnerable objects (hospitals, Kindergarden, chemical firms + Improving the reliability of systems
+ integration of spillways into dikes
+ restriction on land use in riverine areas
Conclusions
Strategies are needed which are reasonable in the short and the mid term
+ communication of hazards
+ removing of highly vulnerable objects (hospitals, Kindergarden, chemical firms + Improving the reliability of systems
+ integration of spillways into dikes
+ restriction on land use in riverine areas