V3A BPG FRA Final FMMP 2

Mekon
ng Riverr Commission
Flood
d Managem
ment and Mitigation Program
mme
Strructura
al Meaasures and Flood PProofin
ng
in the Lo
ower Mekon
M ng Basiin
Best Practtice Guide

B G eliness for
ood R
Flo Risk Asses
A smen
nt
Volum
me 3A
May 2010
2
Published in Phnom Penh, Cambodia in September 2013 by the Mekong River Commission,
Office of the Secretariat in Phnom Penh
Citation:
Royal Haskoning, Deltares, UNESCO‐IHE, The Flood Management and Mitigation Programme,
‘Component 2: Structural Measures and Flood Proofing in the Lower Mekong Basin’, May 2010,
Final Report, Volume 3A “Best Practice Guidelines for Flood Risk Assessment”. 158 pp.
Opinions and interpretations expressed are those of the authors and may not necessarily reflect
the views of the MRC Member Countries
Editors: Ms. Tiffany Hacker, Dr. David Lampert, Mr. David Smith
Editors have applied, to the extent possible, the MRC standard for names of rivers, villages,
districts and provinces. However some names in maps, figures and tables could not be timely
adjusted as a result of the picture‐format used by the authors.
© Mekong River Commission

Office of the Secretariat in Phnom Penh (OSP)
576, National Road #2, Chak Angre Krom,
P.O. Box 623, Phnom Penh, Cambodia
Tel. (855‐23) 425 353. Fax (855‐23) 425 363
Office of the Secretariat in Vientiane (OSV)
Office of the Chief Executive Officer
184 Fa Ngoum Road,
P.O. Box 6101, Vientiane, Lao PDR
Tel (856‐21) 263 263. Fax (856‐21) 263 264
Website: www.mrcmekong.org
Email: [email protected]
MRC Flood Management and Mitigation Programme Component 2: Structural Measures and Flood Proofing
TABLE OF CONTENTS
1 INTRODUCTION .................................................................................................................... 3
1.1 Guide to the reporting structure of the Flood Management and Mitigation
Programme ‐ Component 2, Structural Measures and Flood Proofing .......................... 3
1.2 Best Practice Guidelines for Integrated Flood Risk Management .................................. 4
1.3 Concepts of Flood Risk Assessment in the Best Practice Guidelines .............................. 4
1.3.1 Flood risk and flood risk management .......................................................... 4
1.3.2 Flood risk assessment .................................................................................... 5
1.4 Basis of the Flood Risk Assessment Guidelines .............................................................. 7
1.5 Use of the Flood Risk Assessment Guidelines ................................................................ 7
1.6 The Best Practice Guidelines and project phases/stages ............................................... 8
2 FLOOD HAZARD ASSESSMENT ............................................................................................ 11

2.1 Introduction.................................................................................................................. 11
2.2 Classification of type of flood ....................................................................................... 11
2.3 Tributary flood hazard assessment .............................................................................. 15
2.3.1 General ........................................................................................................ 15
2.3.2 Flood hazard derived from rainfall extremes .............................................. 15
2.3.3 Example of flood hazard derived from rainfall extremes ............................ 18
2.3.4 Flood hazard determined from observed flows .......................................... 23
2.3.5 Flood hazard derived from regional flood statistics .................................... 26
2.4 Mainstream flood hazard assessment.......................................................................... 28
2.4.1 General ........................................................................................................ 28
2.4.2 Data collection ............................................................................................. 28
2.4.3 Database development and conduct of field visits ..................................... 30
2.4.4 Data validation and processing ................................................................... 31
2.4.5 Hydrological hazard assessment in terms of flood peak water levels
and flood volumes ....................................................................................... 31
2.4.6 Flood hazard assessment............................................................................. 32
2.4.7 Example: Mekong near Nakhon Phanom (Thailand) ................................... 33
2.5 Combined flood hazard assessment............................................................................. 37
2.5.1 General ........................................................................................................ 37
2.5.2 Data collection ............................................................................................. 38
2.5.3 Database development and conduct of field visits ..................................... 39
2.5.4 Data validation and processing ................................................................... 39
2.5.5 Hydrological hazard assessment ................................................................. 39
2.5.6 Flood hazard assessment using a hydraulic model ..................................... 41
2.5.7 Example: combined floods in the lower Xe Bang Fai River Focal Area
(Lao PDR) ..................................................................................................... 42
2.6 Flood Hazard Assessment in the Mekong Delta ........................................................... 51
2.6.1 General ........................................................................................................ 51
2.6.2 Data collection and conduct of field visits ................................................... 51
2.6.3 Database development ............................................................................... 53
2.6.4 Hydrodynamic model development ............................................................ 54
2.6.5 Flood hazard assessment............................................................................. 58
2.7 Flood hazard mapping .................................................................................................. 58
2.7.1 General ........................................................................................................ 58
2.7.2 Transfer of modelling results (A) ................................................................. 59
2.7.3 Creation of spatial data (B) .......................................................................... 60
2.7.4 Creation of flooding maps (C) ...................................................................... 62
2.7.5 Data sources ................................................................................................ 64
Guidelines Flood Risk Assessment ‐i‐ May 2010

3 FLOOD DAMAGE ASSESSMENT........................................................................................... 69

3.1 Introduction ................................................................................................................. 69
3.2 Classification of flood damages and benefits ............................................................... 69
3.2.1 Flood damages ............................................................................................ 69
3.2.2 Flood benefits .............................................................................................. 71
3.3 General approach to economic damage assessment................................................... 72
3.3.1 Scope of the analysis and level of detail ..................................................... 72
3.3.2 Direct damage assessment .......................................................................... 73
3.3.3 Damage functions........................................................................................ 74
3.3.4 Relative and absolute damage assessments ............................................... 76
3.3.5 Assessment of indirect damages ................................................................. 77
3.4 Damage models ............................................................................................................ 78
3.5 Economic and financial issues in damage assessment ................................................. 79
3.5.1 Financial vs. economic valuation ................................................................. 79
3.5.2 Estimating values of assets .......................................................................... 79
3.5.3 Time value of money ................................................................................... 80
3.6 Loss of life and health effects ....................................................................................... 80
3.6.1 Introduction................................................................................................. 80
3.6.2 Loss of life .................................................................................................... 81
3.6.3 Health effects and social and environmental impacts ................................ 81
3.7 Presentation of damage assessments results .............................................................. 82
3.8 Data collection.............................................................................................................. 83
3.9 Flood damage assessment in the Mekong River Basin ................................................ 86
3.9.1 Scope and selected areas ............................................................................ 86
3.9.2 Approach and methodology ........................................................................ 86
3.9.3 Damage categories ...................................................................................... 87
3.9.4 Data collection and surveys......................................................................... 87
3.9.5 Derivation of damage functions .................................................................. 92
3.9.6 Creation of damage maps ........................................................................... 95
3.9.7 Results of damage assessments .................................................................. 96
4 FLOOD RISK ASSESSMENT ................................................................................................ 103

4.1 Introduction ............................................................................................................... 103
4.2 General approach ....................................................................................................... 103
4.3 Presentation of the results of risk assessments ......................................................... 104
4.3.1 Economic risk............................................................................................. 104
4.3.2 Risk to life .................................................................................................. 106
4.4 Risk reduction measures and risk evaluation ............................................................. 107
4.4.1 Effects of anti‐flooding and flood preparation measures on risk .............. 107
4.4.2 Basics of cost‐benefit analysis ................................................................... 108
4.4.3 Economic optimisation .............................................................................. 110
4.4.4 Risk valuation for loss of life ...................................................................... 111
4.4.5 Risk matrix ................................................................................................. 111
4.5 Risk assessment for the LMB ...................................................................................... 112
4.5.1 Introduction and approach........................................................................ 112
4.5.2 Results of risk assessment ......................................................................... 113
5 REFERENCES ..................................................................................................................... 117
Guidelines Flood Risk Assessment ‐ ii ‐ May 2010

APPENDICES
Appendix 1 Approach to Flood Damage and Risk Assessment in Europe

Appendix 2 Approaches for the Assessment of Structural Damage to Buildings
Appendix 3 Fundamental issues in Economic Valuation
Appendix 4 Analysis of Human Instability in Flood Flows
Appendix 5 The Best Practice Guidelines and Project Phases/Stages
Appendix 6 Flow Chart to Create Flood Maps with a GIS
LIST OF FIGURES
Figure 1‐1 General scheme for flood risk management. ................................................................. 5
Figure 1‐2 General scheme for flood risk assessment. .................................................................... 6
Figure 1‐3 Scheme for flood risk assessment used in the FMMP‐C2. ............................................. 6
Figure 2‐1 The sub‐basins or sub‐areas in the Lower Mekong River Basin. Letters refer to
the countries in which these sub‐basins are found (M= Myanmar, T = Thailand,
L = Lao PDR, V = Viet Nam, C = Cambodia). .................................................................. 12
Figure 2‐2 Overview of flood types in the Mekong River. ............................................................. 13
Figure 2‐3 EV1 and GEV‐fit to annual maximum daily rainfall at Pleiku for the years 1927‐
2006. ............................................................................................................................. 19
Figure 2‐4 Depth‐duration‐frequency curves for Pleiku. ............................................................... 20
Figure 2‐5 A 2‐hour‐design rainstorm diagram for the Upper Se San River derived from
measures used for Pleiku. ............................................................................................ 21
Figure 2‐6 Principles of the Clark Unit Hydrograph method, shown in various diagrams
that are related to each other. ..................................................................................... 22
Figure 2‐7 Hypothetical scatter plot of data points and fitted regression line showing the
relationship between flood volumes and flood peak water levels. The red
arrow in the Figure shows the deviation from the regression line for a single
observation................................................................................................................... 25
Figure 2‐8 TCEV‐distribution of the ratio of the T‐year return period event to the mean
annual flood for the pooled regional sample (Upper part of LMB, from
Adamson, 2007). .......................................................................................................... 27
Figure 2‐9 Fit of the GEV function to annual maximum daily discharge of the Mekong
River at Nakhon Phanom. ............................................................................................. 33
Figure 2‐10 Peak water level discharge‐Flood volume relation at Nakhon Phanom. ..................... 34
Figure 2‐11 Defined 2D grid of discharges (with return periods 2, 5, 10, 25, 50 and 100
years) and volumes (based on the regression line and lines of deviation from
the regression line). ...................................................................................................... 35
Figure 2‐12 Observed combinations of peak discharge and flow volume (red dots) and
selected 2D grid (blue dots). The green lines show the observed combinations
that most closely reflect the grid combinations. .......................................................... 35
Figure 2‐13 Example of scaling of an observed hydrograph to make it suitable for serving
as one of the grid points in Figure 2‐12. ....................................................................... 36
Figure 2‐14 Derived frequency curve for maximum annual water levels at a location in the
Mekong River floodplain just downstream of Nakhon Phanom. ................................. 37
Figure 2‐15 Development of the flood simulation matrix for combined floods.............................. 41
Figure 2‐16 Hydrological map of Lower Xe Bang Fai River basin. .................................................... 42
Figure 2‐17 Impression of the 3‐D grid for peak discharge and flood volume in the Mekong
River (at Nakhon Phanom) and flood volume in the Xe Bang Fai River (at
Mahaxai). The blue dots only show the ‘outer side’ of the grid; the full grid
contains 90 such dots. .................................................................................................. 44
Figure 2‐18 Scaling of an observed hydrograph for the Xe Bang Fai River at Mahaxai. .................. 45
Guidelines Flood Risk Assessment ‐ iii ‐ May 2010

Figure 2‐19 GEV‐fit to marginal distributions of annual maximum discharge and flood
volume at Mahaxai (Xe Bang Fai River). ....................................................................... 46
Figure 2‐20 Flood volume – peak discharge relation for Nakhon Phanom and flood
volumes relation between Mahaxai and Nakhon Phanom, without and with Se‐
related lines. ................................................................................................................. 47
Figure 2‐21 Xe Bang Fai River hydraulic model with boundary conditions. .................................... 48
Figure 2‐22 Derived frequency curve for maximum annual water levels at a location in the
floodplain of the Xe Bang Fai River. ............................................................................. 49
Figure 2‐23 Flood depth and extent map of the Lower Xe Bang Fai River, for a return
period of T = 100 years. ................................................................................................ 50
Figure 2‐24 Hydraulic infrastructure of the Mekong Delta. ............................................................ 52
Figure 2‐25 The schematisation of the Mekong Delta using the hydrodynamic delta model
(ISIS) involves a large number of hydraulic nodes downstream of Kratie, next
to floodplain ‘reservoirs and river cross‐sections. ....................................................... 53
Figure 2‐26 Annual maximum water discharges at Stung Treng, 1910‐2006. ................................ 55
Figure 2‐27 Annual flood volumes at Stung Treng, period 1910‐2006............................................ 55
Figure 2‐28 Seasonal hydraulic inflow to the Tonle Sap Lake, estimated from 1910‐2006,
with recorded data for 1997‐2004. .............................................................................. 56
Figure 2‐29 Average monthly flow regime of the Mekong River at Stung Treng and of the
Tonle Sap Lake as percentage of the annual totals. ..................................................... 56
Figure 2‐30 Water level map, based on interpolation of ISIS nodes. .............................................. 58
Figure 2‐31 Water depth map, generated from a water level map subtracting the DTM
(ground) levels.............................................................................................................. 58
Figure 2‐32 Depiction of how water or flood depth is obtained by subtracting the ground
level (DTM) from the water level (hydraulic simulation + GIS interpolation). ............. 59
Figure 2‐33 Example of ISIS data imported into Access, with node information and water
levels for different scenarios, various return periods (probabilities), and
different periods of the year. ....................................................................................... 60
Figure 2‐34 ISIS schematization. ISIS calculates with (among other data) reservoir nodes (+)
and cross‐section nodes (dots). A spatial component of ISIS contains cross‐
sections (blue lines) and reservoirs (purple polygons). ................................................ 61
Figure 2‐35 Example of a raster, with the reservoir raster (angular areas) superimposed on
the point‐based raster.................................................................................................. 63
Figure 2‐36 Example of a flood depth map, here in the Mekong Delta for a flood with a
return period of 25 years. ............................................................................................ 64
Figure 3‐1 General approach to damage estimation, with the relationship between land
use and flood depths used to calculate damages. ....................................................... 73
Figure 3‐2 Example of a stage damage function for houses in the Netherlands (Kok et al.,
2005). Note that the curves for building damage and contents show these as
percentages of the total damage. ................................................................................ 75
Figure 3‐3 Schema of structural flood damage analysis. ............................................................... 76
Figure 3‐4 Absolute and relative damage approaches. ................................................................. 77
Figure 3‐5 Example of a mortality function derived for large‐scale flooding in the
Netherlands (Jonkman, 2007). ..................................................................................... 81
Figure 3‐6 Damage intensity per hectare of the flooding of South Holland province in the
Netherlands. ................................................................................................................. 83
Figure 3‐7 The absolute and relative flood damage assessment approaches require
different data sets. ....................................................................................................... 91
Figure 3‐8 District flood damage curves for three categories of damages in Koh Andet
District in Cambodia. .................................................................................................... 93
Figure 3‐9 Example of a damage function for housing, relating housing damage (as % of
the total value) to floodwater depths. ......................................................................... 94
Figure 3‐10 Relative flood damage curves for rice paddies, Cambodia. ......................................... 95
Guidelines Flood Risk Assessment ‐ iv ‐ May 2010

Figure 3‐11: Steps (simplified) to create maps based on absolute and relative damage –
flood depth curves........................................................................................................ 97
Figure 3‐12 Example of a damage map based on relative damage curves, showing total
damage to rice paddy fields in 3 districts and their communes in Cambodia.............. 99
Figure 3‐13 Example of a damage map based on relative damage curves, showing damages
to rice paddy fields per hectare in 3 districts in Cambodia and their communes. ..... 100
Figure 4‐1 Conceptual approach for determination of flood risk. ............................................... 104
Figure 4‐2 Relationship between return period and damage (left) and probability of
flooding and damage (right). ...................................................................................... 105
Figure 4‐3 The area below the curve, the total risk or expected damage (for example, in
USD/year), can be approximated by summing the areas of the rectangles, A, B,
b, C, c, D, d, E and e. ................................................................................................... 106
Figure 4‐4 Example of an FN curve. ............................................................................................. 107
Figure 4‐5 Effects of measures on flood risk. .............................................................................. 108
Figure 4‐6 Principle of economic optimisation. ........................................................................... 110
Figure 4‐7 Societal risk of fatalities due to flooding and the limit line of acceptable harm. ....... 111
Figure 4‐8 Structure of a risk matrix. ........................................................................................... 112
Figure 4‐9 Approach for estimation of the risk. .......................................................................... 112
Figure 4‐10 Probability damage curve for three Cambodian districts........................................... 113
Figure 4‐11 Flood risks by district for selected areas in Cambodia and Viet Nam. ....................... 114
LIST OF TABLES
Table 2‐1 Overview of flood types per sub‐area (sub‐basin) ....................................................... 14
Table 2‐2 Parameters of EV1 and GEV distributions fitted to annual maximum daily
rainfall at Pleiku and rainfall values (mm) for selected return periods. ....................... 19
Table 2‐3 Short duration rainfall (mm) at Pleiku derived from daily rainfall extremes. .............. 20
Table 2‐4 Computation of design storm ordinates for 15 min intervals. ..................................... 21
Table 2‐5 Parameters of the GEV distribution fitted to annual maximum daily discharge
of the Mekong River at Nakhon Phanom and discharges (m3/s) for selected
return periods............................................................................................................... 34
Table 2‐6 Average monthly and annual flow (MCM) of the Xe Bang Fai River at Mahaxai
and Mekong River at Nakhon Phanom......................................................................... 42
Table 2‐7 GEV‐parameters, flood peak discharges and flood volumes (June‐November)
for distinct return periods in the Xe Bang Fai River at Mahaxai. .................................. 44
Table 2‐8 Overview of water level monitoring stations at coastal boundaries in Viet Nam. ....... 57
Table 3‐1 Classification of flood damages. ................................................................................... 69
Table 3‐2 Distribution of damage over damage categories for a flood scenario for South
Holland province in the Netherlands (Jonkman et al., 2008). ...................................... 83
Table 3‐3 Overview of Flood Damage Data .................................................................................. 84
Table 3‐4 Observed rank share of damages in the focal areas .................................................... 87
Table 3‐5 Data sources used for mapping damages in the Cambodian Mekong Delta. .............. 90
Table 3‐6 Damage data for different return periods for Koh Andet District. Damage is
shown in million USD.................................................................................................... 96
Table 4‐1 Illustrative summary of damage and probability calculations for a range of
flood scenarios. .......................................................................................................... 104
Table 4‐2 Results of risk calculations for the three focal areas in Cambodia. ............................ 113
Guidelines Flood Risk Assessment ‐v‐ May 2010

MRC Flood
d Management and
a Mitigation Programme
P Com
mponent 2: Strucctural Measures and Flood Prooffing
Guidelines Flood Risk Asseessment ‐ vi ‐ May 2010

ABBREVIATIONS AND ACRONYMS
ADCP Acoustic Doppler Current Profiler (Acoustic Doppler Profiler); an

instrument used to measure the velocity of water across an entire
water column
ArcGIS ArcGIS is a complete software system for designing and managing
solutions through the application of geographic knowledge
ARF Area Reduction Factor (hydrology)
BCM Billion Cubic Meters
BDP Basin Development Planning
BPG Best Practice Guidelines
CBA Cost Benefit Analysis
d/s Downstream
DACA Damage and Casualties Assessment project for the Lower Mekong
Basin using the HIS‐SSM for assessing damages and casualties
DEM Digital Elevation Model (see also DTM), is a digital model or 3D
representation of a terrain's surface created from terrain elevation
data
DSF Decision Support Framework
DTM Digital Terrain Model (see also DEM)
EC European Commission
EU European Union
EV1 Extreme Value type 1 distribution (hydrology)
EXCIMAP European Exchange Circle on Flood Mapping
FEMA Federal Emergency Management Agency of the USA
FHA Flood Hazard Assessment
FMM Flood Management and Mitigation
FMMP‐C2 Flood Management and Mitigation Programme, Component 2
(MRC)
FN curve Societal risk is usually defined as the relationship between the
probability of a catastrophic incident, expressed as the average
frequency with which it can be expected to occur, and its
consequences. It is usually represented as an F‐N curve in which F is
the annual frequency and N represents the number of casualties.
FRA Flood Risk Assessment
FV Future Value (economic analysis)
GEV Generalised Extreme Value distribution (hydrology)
GIS Geographic Information System
HAZUS Software model used for risk assessment analysis of potential losses
from floods, hurricane winds and earthquakes (by FEMA)
HH Household(s)
HIS‐SSM Hydrological Information System: a damages and casualties
assessment module
HYMOS An information system for water resources management
IDW Inverse Distance Weighting method: an interpolation method to
obtain a continuous GIS‐raster on the basis of data points (nodes),
assigning the most weight to nearby points by using their distance
to the point to calculate(see also NN)
IFRM Integrated Flood Risk Management
ISIS Hydrodynamic simulator for modelling flows and levels of water in
open channels and estuaries
IUH Instantaneous Unit Hydrograph (hydrology)
Guidelines Flood Risk Assessment ‐ vii ‐ May 2010

JICA Japan International Cooperation Agency

LMB Lower Mekong (River) Basin
LMD Lower Mekong Delta
LXQ Long Xuyen Quadrangle (Viet Nam)
MCM Million Cubic Meters
MRC(S) Mekong River Commission (Secretariat)
MRB Mekong River Basin
MSL Mean sea level, the average (mean) height of the sea, with
reference to a suitable reference surface
NN Natural Neighbours method: an interpolation method to obtain a
continuous GIS‐raster on the basis of data points (nodes), assigning
the greatest weight to nearby points by calculating overlapping
areas in Voronoi/Thiessen polygons (see also IDW)
NPV Net Present Value (an economic term)
PDR (Lao) (Lao) People’s Democratic Republic
PoR Plain of Reeds (Viet Nam)
PV Present Value (an economic term)
RFMMP Regional Flood Management and Mitigation Programme
RID Royal Department of Irrigation
RR Rainfall Ratio (hydrology)
SBF Xe Bang Fai River (Lao PDR)
SCS‐CN Soil Conservation Service (USA) Curve Number method (hydrology)
SWAT Hydrological model: river basin scale model quantifying the impact
of land management practices in large, complex watersheds
TCEV Two Component Extreme Value (hydrology)
u/s upstream
UH Unit Hydrograph (hydrology)
UK United Kingdom
UNESCO‐IHE Institute for Water Education (IHE) of the United Nations
Educational, Scientific and Cultural Organization (UNESCO)
USA United States of America
UTM The Universal Transverse Mercator is the geographic coordinate
system, which uses a 2‐dimensional Cartesian coordinate system to
give locations on the surface of the Earth
VRSAP The “Vietnam River Systems and Plains” model is a hydro‐dynamic
model to simulate mainly one‐dimensional hydrodynamic river flow
or quasi two‐dimensional flow on floodplains
WUP Water Utilisation Programme
WUP‐A Water Utilisation Project Component A: Development of Basin
Modelling Package and Knowledge Base
Guidelines Flood Risk Assessment ‐ viii ‐ May 2010

GLOSSARY
Attribute data Also known as "count" data.
Areal rainfall Rainfall in a specific area expressed as an average depth of liquid

water over the area; the average rainfall over the specific area
calculated on timescales on a storm, seasonal, or annual basis.
Backwater The rise in surface elevation of flowing water upstream from and as
a result of an obstruction to flow. The difference between the
observed stage and that indicated by the stage‐discharge relation is
reported as backwater.
Best Practise Guidelines Best Practice Guidelines are guidelines, describing methods or
techniques that have consistently shown results superior to those
achieved with other means, and that are used as a benchmark. Best
Practice Guidelines are also interpreted guidelines describing
processes of developing and following a standard way of doing
things that dedicated organizations can use.
Colmatage The natural deposition of particles suspended in water on land

areas, usually on river floodplains and deltas, coastal lowlands, and
flooded meadows.
Convective rainfall Occurs when rain particles form in the active updraft of a
cumulonimbus cloud, grow primarily by the collection of cloud
droplets (i.e., by coalescence and/or riming) and fall out not far
from their originating updraft.
Damage curve The functional relationship between inundation characteristics
(depth, duration, flow velocity) and damage for a certain category
of risk.
Design rainstorm A critical rainfall event that is used for assessing the flood
hydrograph of a certain return period.
Direct damage All harm which relates to the immediate physical contact of
floodwater to people, property and the environment. This includes,
for example, damage to buildings, productive assets, loss of crops
and livestock, loss of human life, immediate and immediately
foreseeable and calculable health impacts to the population in the
flooded area and ecological harm.
Double mass test A statistical test that is used to check the consistency of many kinds
of hydrologic data by comparing date for a single station with that
of a pattern composed of the data from several other stations in the
area.
Exposure A measure of the people, assets and activities threatened by a flood

hazard.
Flashiness Term reflects the frequency and rapidity of short term changes in
streamflow, especially during runoff events.
Guidelines Flood Risk Assessment ‐ ix ‐ May 2010

Flood control A structural intervention to reduce a flood hazard.
Flood damage Damage to people, property and the environment caused by a

flood. It includes direct as well as indirect damages.
Flood damage map Map of the predicted flood damage.
Flood damage risk The combination or mathematical product of the probability of the
(= Flood risk) flood hazard and the possible damage that it may cause. This risk
can also be expressed as the average annual possible damage or
expected damage. If it is expressed in a financial measure, it is the
expected net present value using economic valuations with
different measurement assumptions.
Flood hazard A flood that may potentially result in damage. A hazard does not
necessarily lead to damage.
Flood hazard map Map of the predicted or documented extent/depth/velocity of

flooding with an indication of the flood probability.
Flood proofing A process for preventing or reducing flood damages to

infrastructure, buildings and/or the contents of buildings located in
flood hazard areas.
Flood risk management Comprehensive activity involving risk analysis, and both
identification and implementation of risk mitigation measures.
Flood risk management Actions that are taken to reduce the probability of flooding or the
measures possible damages due to flooding or both.
Flood risk map Map of the predicted extent of different levels/classes of average
annual possible damage due to flooding.
Hydrograph A time series of water levels at a fixed location. For floods, it shows
the rise and fall of flood waters and the peak water level height of
the flood.
Hydrological hazard A hydrological event (discharge) that may result in flooding.
Hydraulic roughness Is the measure of the amount of frictional resistance water

experiences when passing over land and channel features.
Hydrodynamic load The pressures that result from water flowing against and around a
rigid structural element or system. The pressures are functions of
velocity, direction of flow relative to the object, object geometry,
and object surface roughness characteristics.
Hydrostatic load The pressure at any water depth due to hydrostatic pressures.
Hyetograph A graphical representation of the distribution of rainfall over time.
Indian 1954 A geodetic datum first defined in 1954 and used by MRC for the
Guidelines Flood Risk Assessment ‐x‐ May 2010

Mekong River Basin.
Indirect damage All non‐direct damage which relates to the disruption of economic
activity and services due to flooding including impacts on the non‐
flooded areas.
Integrated flood risk The approach to flood risk management that focuses on the full
management chain of a meteorological hazard leading to flood damages and
considers combinations of structural and non‐structural solutions to
reduce that damage.
Meteorological hazard A meteorological event (storm) that may result in a hydrological

hazard and, eventually, in flooding.
Monte Carlo analysis A computer sampling technique often used for simulating long term
events based on statistical probabilities, testing a series of multiple
random events by drawing random samples from a data set.
Non‐staggered grid A grid for which vector variables and scalar variables are stored at
the same locations.
Orographic effect Occurs when an air mass approaches a mountain range and is
rapidly forced upward, causing any moisture to cool and create
precipitation in the form of rain or snow.
Unit hydrograph (UH) The hypothetical unit response of a watershed (in terms of runoff
volume and timing) to a unit input of rainfall. It can be defined as
the direct runoff hydrograph resulting from one unit (e.g., one cm or
one inch) of effective (net) rainfall occurring uniformly over that
watershed at a uniform rate over a unit period of time.
Raster Defines values for pixels occupying a particular rectangular area of

the plane, not necessarily including (0, 0).
Rating curve A rating table or curve is a relationship between stage (water level)
and discharge at a cross section of a river.
Reach Any length of a stream between any two points.
Resilience The ability of a human social system (at the level of community or
society) to cope with the damaging effect of floods.
Return period Recurrence time, average time interval between subsequent events
in which conditions are exceeded. When designing a structure, the
return period is usually larger than the projected lifetime, because,
for instance, if both would equal 50 years, the structure would have
a 64% probability of failure during its lifetime. In statistical analysis
an event with a return period of N years is likely, on average, to be
exceeded only once every N years.
Routing A technique used to predict the changes in shape of water as it

moves through a river channel or a reservoir.
Guidelines Flood Risk Assessment ‐ xi ‐ May 2010

Saint Venant equations The shallow water equations in its unidimensional form are a set of
hyperbolic partial differential equations that describe the flow
below a pressure surface in a fluid (sometimes, but not necessarily,
a free surface).
Scour Washing away of the bed/bank material under the action of current
and wave.
Serial and cross The existence of serial correlation in a time series will affect the
correlation ability of the test to assess the site significance of a trend; and the
presence of cross‐correlation among sites in a network will
influence the ability of the test to evaluate the field significance of
trends over the network.
Spatial data Data or information that identifies the geographic location of

features and boundaries on Earth, such as natural or constructed
features, oceans, and more. Spatial data is usually stored as
coordinates and topology, and is data that can be mapped. Spatial
data is often accessed, manipulated or analyzed through
Geographic Information Systems.
Spot height The height of the highest point in a given area expressed in feet or
meters above sea level, as marked on topographical charts. These
are indicated by black dots with adjacent numerals.
Susceptibility The opposite of resilience: the inability of a human social system (at
the level of community or society) to cope with the damaging effect
of floods.
(N‐year) synthetic series Simulated long term event‐related data series.
Vulnerability The potential damage that flooding may cause to people, property
and the environment.
Guidelines Flood Risk Assessment ‐ xii ‐ May 2010

MRC Flood Maanagement and Mitigation Proggramme Compon
nent 2: Structural Measures andd Flood Proofingg
REFEREN
NCE SYMB
BOLS FOUN
ND IN THEE TEXT
The Flood Management
M and Mitigatiion Programmme Component 2 (FMMP‐‐C2) guideline es contain
symbols in the
t left marggins for quick reference. The symbols are of two typpes. They indicate:
A. Type of
o text/conten nt; and
B. Project stages within the five coonsecutive prroject phases.
The goal off the symbols is to help readers to scans the textt to look for theory, examples, or
applicationss or to try to find informattion needed for work at specific projecct stages.
A) Text/C
Content Symb
bols: The repoort texts are categorised into four grouups, as follow
ws:
i) Project backgrround/Reportrt informationn

FMMP‐projectt informationn and backgro ound,
or explanation
n of the reporrt structure or
o content.
ii) Theory
The theory behind the pro posed/applie
ed methods and guideliness.
iii) Example
Examples of th
he proposed//applied methods and guidelines.
iv) Applications and Guideline s

M
Methodology and theory aadapted/applied to the Lower Mekongg
R
River Basin (LM
MB), includinng guidelines. The guidelines are applieed
in
n one of the five
f project sttages describbed below (B).
B) Project Stage Symb bols: A projecct usually con

nsists of five phases
p (see SSection 1.6). FMMP‐C2
F
coverss only the second phaase: Plannin ng/Development/Design. This phase can be
subdivvided into five
e stages:
a) Preliminary/pre‐feasibility study stage
b) dy and overaall planning sttage

Feasibility stud
c) Preliminary de
esign stage
d) D
Detailed desiggn and detaileed planning stage
s
e) C
Construction/bid documennts stage
Any part of a guideline falling

f outsidee the scope
of the five stages
s is marked with a w hite cross:
Sometimes more than one

o symbol m
may apply to a section.
ood Risk Assessm

Guidelines Flo ment ‐ xiii ‐ May 2010
CHA
APTEER 1
INTRO
ODUCTTION
1 INTRODUCTION
1.1 Guide to the reporting structure of the Flood Management and Mitigation
Programme ‐ Component 2, Structural Measures and Flood Proofing
The Mekong River Commission (MRC) implemented Component 2 on Structural Measures and
Flood Proofing of the Flood Management and Mitigation Programme (FMMP) between
September 2007 and January 2010 under a consultancy services contract between the Mekong
River Commission Secretariat (MRCS) and Royal Haskoning, working in association with Deltares
and UNESCO‐IHE. The work comprised three periods, an Inception period and two
Implementation periods. During each period, the consultants delivered a series of outputs and
discussed them with the MRC, the National Mekong Committees, and line agencies of the four
MRC Member Countries. A portion of Component 2 ‐ on 'Roads and Floods' ‐ was implemented
by the Delft Cluster under a separate contract with the MRC. Component 2 included five
Demonstration Projects which are presented separately.
The consultancy services contract for Component 2 requests, in general terms, four main
products in addition to a Final Report. The reports produced as of the completion of Component
2 (FMMP‐C2) are structured as follows:
Volume 1 Final Report
Volume 2 Characteristics of Flooding in the Lower Mekong Basin

Volume 2A Hydrological and Flood Hazards in the Lower Mekong Basin
Volume 2B Hydrological and Flood Hazards in Focal Areas
Volume 2C Flood Damages, Benefits and Flood Risk in Focal Areas
Volume 2D Strategic Directions for Integrated Flood Risk Management in Focal Areas
Volume 3 Best Practice Guidelines for Integrated Flood Risk Management
Volume 3A Best Practice Guidelines for Flood Risk Assessment
Volume 3B Best Practice Guidelines for Integrated Flood Risk Management Planning and
Impact Evaluation
Volume 3C Best Practice Guidelines for Structural Measures and Flood Proofing
Volume 3D Best Practice Guidelines for Integrated Flood Risk Management in Basin
Development Planning
Volume 3E Best Practice Guidelines for the Integrated Planning and Design of Economically
Sound and Environmentally Friendly Roads in the Mekong Floodplains of
Cambodia and Viet Nam1
Volume 4 Project development and Implementation Plan
Volume 5 Capacity Building and Training Plan
Demonstration Projects
Volume 6A Flood Risk Assessment in the Nam Mae Kok Basin, Thailand
Volume 6B Integrated Flood Risk Management Plan for the Lower Xe Bang Fai Basin, Lao
PDR
Volume 6C Integrated Flood Risk Management Plan for the West Bassac Area, Cambodia
Volume 6D Flood Protection Criteria for the Mekong Delta, Viet Nam
Volume 6E Flood Risk Management in the Border Zone between Cambodia and Viet Nam
This report is Volume 3A in the above series.
1
Developed by the Delft Cluster
Guidelines Flood Risk Assessment ‐3‐ May 2010

MRC Flood
d Management and
P Com
The FMM MP Componeent 2, Structtural Measurres and Flood Proofing, was developped in three
steps: th
he Inception Phase and Stages
S 1 andd 2 of the Im
mplementatioon Phase. Thhe Inception
Phase beegan at the ende of Septe ember 2007 and concluded in accord dance with thhe Terms of
Referencce with a Reggional Worksshop in Ho CChi Minh Cityy at the end of January 22008, only 4
months aafter project initiation. The original TO
OR envisaged the Stage 1 Implementatiion Phase to
be carrieed out in a period of 6 months, leavving 12 mon nths for the Stage 2 Imp lementation
Phase. SSee for refereence Final Repport, Volumee 1.
1.2 Best Practiice Guideline

es for Integr ated Flood Risk
R Manage
ement
The MRCC Secretariatt has develop

ped Best Praactice Guidelines (BPG) fo
or Integratedd Flood Risk
Managemment under FMMP‐C2 to o provide pollicy‐makers, managers an
nd FMM proffessionals in
the MRC
C and Membeer Country line agencies wwith a common standard to o apply in:
 Policcy formulation;
 Strattegy and plan
nning for policy implemenntation;
 Projeect design an
nd evaluation;
for flood risk manageement in the Lower

L Mekonng Basin (LMB).
Each Meember Countrry of the MRC already haas its own policy and set of o legal fram eworks that
guide or regulate thee planning, evvaluation andd implementtation of floo
od risk managgement that
complemment the BPGs.
The BPGGs do not attempt to summarise or replace thesse national guidelines,

g nnor are they
d as a recipe for carrying out planningg or project design for flo
intended ood risk mannagement in
the LMB. Rather the BPGs provid de an inform ation resourcce/tool that can be adappted to each
country aand project context.
c
The basiss of flood rissk manageme ent in the LM

MB is proper assessment of o the managgeable risks.
The Bestt Practice Guiidelines for flood risk asseessment (BPG
G for FRA) in the LMB offeer one set of
guidelinees.
1.3 Concepts of
o Flood Risk
k Assessmen t in the Bestt Practice Gu
uidelines
1.3.1 Flood risk and flood risk managemennt
In generaal, the word ‘risk’ refers to

t the proba bility of loss or harm. In the context oof flood risk
managem ment it denotes the co ombination oof the probaability of a flood with its adverse
consequeences. The deefinition adopted by the EEuropean Com mmission is the following::
‘"Flood rrisk" means the combination of the pprobability off a flood eve
ent and of thhe potential
adverse consequencees for human health, thhe environmeent, cultural heritage annd economic
activity a
associated wiith a flood eve
ent.’
In this co
ontext, the teerms hazard and
a vulnerab ility are often n used. Hazarrd refers to thhe source of
danger, ii.e. the (probability of) flooding. Vulneerability relate
es to potentia
al adverse coonsequences
in case oof an event.
(Integratted) Flood rissk manageme ent is an appproach to ide

entify, analysse, evaluate, control and
manage flood risks in n a given rive
er system. Figgure 1‐1 pressents a general scheme foor flood risk
managem ment consistiing of four co
omponents.
Guidelines Flood Risk Asseessment ‐4‐ May 2010

 Definitiion of the flood system, thhe hazards annd the scale and
a scope of the analysis.
 Quantittative analyssis of proba bilities and adverse con nsequences ccombined into a risk
measurre, presented d graphically oor displayed on a flood rissk map.
 Evaluattion (assessm ment) of risks acceptable or
o unacceptab ble; and
 Identification of measures for risk reductio on and contrrol, includingg structural and non‐
structural measure es as well aas managem ment and co ontrol optio ns (e.g., mo
onitoring,
inspecttion or mainte enance).
S
System definition
zards
Scope, haz
Quantitative ana
alysis and
flood risk mapping
Structural and
non-strructural
-
meas sures
Risk Evalua
ation
R
Risk reductio
on and
control
Figure 1‐1 General sch

heme for flood risk manageme
ent.
This generaal scheme foccuses on miniimisation of flood

f risks to an acceptabble level. The approach
could also be used to o assess thee overall hyydrological performance
p of the systtem (e.g.
minimisatioon of drought, maximisattion of waterr quality and ecological qquality). In su uch cases,
the approaach can focu us on multip le objectivess: not only to t minimise the risk, bu ut also to
maximise thhe performan nce of the waater system.
The conceppt of flood risk assessmentt generally re efers to the second step inn this general scheme;
the quantitative analysiss of the level of flood risk in an area orr basin. This rreport focuse
es on that
step. Otherr guidelines//reports offeer insight into possible measures
m andd approache es for the
evaluation step
s (see Secction 1.2 for aan overview of
o other guide elines).
1.3.2 Flood risk asse

essment
The identiffication and mapping of fflood risks reequires several types infformation collected in
several step
ps. Figure 1‐2 me for flood risk assessm ent, consistin
2 presents a ggeneral schem ng of four
componentts:
1) Flood System
S defin
nition and coollection of basic
b data (e.g., basic datta on elevation of the
terrain and water flo
ows).
2) Flood hazard analyysis of the ooccurrence of o flooding. This step inccludes an an nalysis of
meteorrological events that mayy eventually lead to floo oding combinned with datta on the
hydroloogical hazard
d characterisstics of the respective watershed
w (peeak flood diischarges,
volumees). The resultts can be eveentually displaayed by means of flood (hhazard) mapss.
3) Vulneraability and damage
d asseessment in thet areas prrone to floodding based on socio‐
econommic data and a vulnerabilitty/damage model.
m
ood Risk Assessm

Guidelines Flo ment ‐5‐ May 2010
4) Risk determination and flood risk mapping combining the results of the flood hazard
analysis, (the probability of a certain hazard) with the results of the damage assessment.
These results can be displayed in risk maps, graphically, or numerically in ways that offer
insight into the expected annual damage.
These steps are presented in separate sections in this report.
System definition and

collection of basic data
Flood hazard Vulnerability and

analysis damage assessment
Flood hazard Flood damage

maps maps
Risk determination and
flood risk mapping
Flood risk
maps
Figure 1‐2 General scheme for flood risk assessment.
The FMMP‐C2 project uses a variation on this scheme that is shown in Figure 1‐3. The figure
indicates how meteorological information is combined with watershed and river information to
determine hydrological and flood hazards. Damages can be determined by combining data for
the possibility of floods with data on the vulnerability to damage of the affected area. When the
return period data for different flood events are included, the probability of a certain type of
damage ‐‐ i.e., the risk – can be determined. The figure also indicates how crisis management
can reduce the risk at the final stage.
precipitation peak discharges floods damage

volumes
hazard risk
meteorological
hazard
hydrological
hazard
flood hazard
watershed
damage risk
river channel
damage
vulnerability
crisis
management
Figure 1‐3 Scheme for flood risk assessment used in the FMMP‐C2.

1.4 Basis of the Flood Risk Assessment Guidelines
The Best Practice Guidelines in the area of Flood Risk Assessment on the basis of:
1. a review of other guidelines or ‘best practice’ documents in the field of flood risk
assessments (see Appendix 1 of this BPG);
2. a review of existing guidelines in the LMB countries regarding damage assessments; and
3. experience in flood risk assessments in the LMB focal areas during Stage 1 of the FMMP‐C2.
During Stage 1 of the FMMP‐C2, a flood damage data collection and processing methodology
was designed for the LMB focal areas that takes into account the availability of secondary data
as well as time and budget constraints. Survey teams collecting secondary and primary data
offered feedback on the feasibility of the methodology along with consultants. Both concluded
that the methodology was feasible and suitable for use in the MRC Member Countries and that
the damage data collected could be used for estimating flood risks. (For details on the FMMP‐C2
damage data collection, see Annex 2 to the Stage 1 Evaluation Report.)
The details and results of these methodologies applied for the assessment of the flood hazards
in the LMB focal areas were presented in Annex 1 of the Stage 1 Evaluation Report and were
updated in Stage 2.
The guidelines for flood risk assessment have been tested and amended during Stage 2 of
FMMP‐C2, especially through the demonstration projects. The methodology was also discussed
in regional training courses.
1.5 Use of the Flood Risk Assessment Guidelines
Who should use the “Best Practice” guidelines and for what purpose?
The four target groups for these guidelines are:
1. Policy makers who will participate in the development of the flood risk management
strategies at national and/or regional levels.
2. Regional and/or basin planners who should be aware of the impact of development
scenarios on flood risks and who should apply flood risk management as an essential
component of Integrated Water Resources Management. The IFRM BPG for Basin
Development Planning (BDP) is also relevant to this category of planners.
3. Flood risk management related project designers, for structural or non‐structural projects.
They can use these guidelines to estimate field level flood benefits and costs to input in
project analysis as described in the Best Practice Guidelines for IFRM Planning and Impact
Assessment.
4. Participants in multilateral discussions on transboundary flood risk related impacts.
The BPG for FRA provides a set of methodologies for assessing:
1. flood hazards;
2. flood damages; and
3. flood risks.

MRC Flood
d Management and
P Com
These guuidelines com

mbine theory and practicee with speciffic emphasis on flooding in the LMB.
This doccument follows the gene eral scheme for flood riisk mapping (Figure 1‐2)). Section 2
describess the processs of analysingg flood hazarrds. Section 3 presents the
e tools for floood damage
and vulnerability asseessment. Secttion 4 showss how flood risk can be de etermined annd displayed.
Each secttion is comprrised of four sub‐sections.
s . These prese
ent:
 nciples, conceepts and term

Prin minology;
 Seleected modelss (for definingg and quantify
fying hazards,, consequencces or risks);
 Proposed approaaches for the e LMB; and
 Examples of resu ults from Stagge 1 of the M
Mekong FMMP‐C2.
These gu uidelines cann be adopted d as part of a consistent and uniform m approach tto flood risk
assessmeent in the LM MB. These guidelines do nnot fully pressent all of thee details of tthe methods
and mod dels in risk asssessment, bu
ut focus, insteead on the prrinciples and offer an oveerview of the
steps thaat need to bee followed forr a complete risk assessmeent.
Precise eelaboration and applicatio

on of risk asseessment depends on man ny factors: obbjectives and
scope off the assessmment, availab ble data and local condittions (e.g., hydrology, lannd use). For
example,, there are seeveral types of
o flooding inn the LMB: tributary floodss, main strea m floods, or
floods in
n the floodplaains (see Secction 2.2). Eaach type of flooding
f is asssociated witth particular
ood duration or flow spe ed. These, among other factors, affe ct the flood
processees such as flo
hazard analysis.
1.6 The Best Practice Guidelines and pproject phase

es/stages
In order to facilitate managementt of an enginneering proje

ect, project managers
m norrmally divide
projects into phases such
s as the fo
ollowing five phases:
1. Initiiation
2. Plannning/Develoopment/Desiggn
3. Prod duction/Impllementation
4. Mon nitoring/Conttrol
5. Clossure
The Besst Practice Guidelines area almost exclusively applicable to Phase 22: Planning/
Developmment/Design. This phase, its stages annd the assocciated symbols used in th e guidelines
are elabo
orated in App
pendix 5 and are presenteed briefly, abo
ove, in the report introducction.
Guidelines Flood Risk Asseessment ‐8‐ May 2010

CHA
APTEER 2
FLO
OOD HAZARD ASSSESSMENT
2 FLOOD HAZA
ARD ASSESSSMENT
2.1 In
ntroduction
There are a variety of measures aavailable to prevent or reduce

r damaaging effects of flood
hazards. Whether
W or not these meeasures are effective
e stro
ongly dependds on the syystem (of
available measures)
m in place. In ordder to choose the approppriate protecctive measures, policy
makers must have a clear understannding of the processes
p of flooding
f invoolved. Some of
o the key
o start with are:
questions to a
 What types
t of floodds occur?
 From where
w does the water com me and to wh
here does it fllow?
 What quantities
q of water are invvolved?
 Which factors influence water fflows?
 What is the frequency of floodinng?
This type off analysis is re

eferred to ass “flood hazarrd assessmen nt”. This chappter presents the basic
guidelines in flood hazard assessmennt for the Me ekong River Basin
B (MRB). SSection 2.2 presents
p a
on list for the different typpes of flood events
classificatio e in the MRB. This claassification iss relevant
because th he different types requiire differentt guidelines. Sections 2..3 ‐ 2.6 pre esent the
guidelines that are app plicable to esstimating flo ood hazard for each of tthe different types of
flooding thaat are listed in Section 2.2 2 along with examples of their use for selected “foccal areas”
in the MRB.
Section 2.7 describes flo

ood hazard m
mapping, following the asse
essment proccedures.
2.2 C
Classification
n of type of fflood
A classificattion of floodss is used as aapplied by the

e MRC in the Lower Mekoong River Bassin (LMB),
including:
 Tributaary floods,
 Main stream
s floodss,
 Combiined floods,
 Floodss in Cambodia an Floodplainn, and
 Floodss in the Meko ong Delta.
The LMB is divided in sub‐basins nuumbered from m 1 to 10 (seee Figure 2‐1 ). Figure 2‐2 offers an
overview off the location
n where diffe rent flood typ
pes occur in the
t LMB.
The flood tyypes are desccribed in detaail as follows..
Tributary flloods occur in the steep ssloped upperr reaches of the basins. Thhey are generally flash
floods, caussed by intensse rainfall afteer a long rain
ny period andd forcing the ccatchment to
o respond
quickly. Flaash floods arre short livedd, rise and fall
f rapidly and have highh flow velocities. The
effects of flash
f floods, when accom mpanied by laandslides, are e equivalent to dam brea ak waves.
Further dow wnstream, in n the middlee sections off tributaries the flashinesss reduces, but flood
levels are not affected by
b backwater from the Me ekong, as in the Nam Maee Kok at Chian ng Rai.
Mainstream
m floods are floods alongg the Mekongg caused by high water leevels on the Mekong.
The hydrological hazard is determ ined by the flood peak water levelss responsible e for the
maximum water
w levels. Flood volum es are imporrtant for measuring flood duration. Ma
ainstream
ood Risk Assessm

Guidelines Flo ment ‐ 11 ‐ May 2010
floods occur at a number of locations along the Lower Mekong River all the way from Chiang
Saen to Kratie.
A Northern highlands
1 Northern Laos
2 Chiang Rai
B Central plateau and highlands
1V 3 Nong Khai / Songkhram
4 Central Laos
5 Mun / Chi
1L C Southeast highlands
6 Soutthern Laos
7 SeSan / Sre Pok / Se Kong
2T D Southern region
8 Kratie
9 Tonle Sap
10 Delta
4L
3L
3T
5T 6L
7L
7V
9T 9C 6C 7C
8C
8V
10C
Areas outside the

Lower Mekong Basin
but affected by
10V Mekong floods
Figure 2‐1 The sub‐basins or sub‐areas in the Lower Mekong River Basin. Letters refer to the countries in
which these sub‐basins are found (M= Myanmar, T = Thailand, L = Lao PDR, V = Viet Nam, C =
Cambodia).
Guidelines Flood Risk Assessment ‐ 12 ‐ May 2010

Figure 2‐2 Overview of

o flood types inn the Mekong River.
R
Combined floods
f are flo
oods in the ddownstream sections of tributaries, w here the floo od level is
determined d by a combination of tribbutary flow anda water levvels in the M ekong River that back
up the tribu
utary levels and
a impede ddrainage. Wh hen water levvels in the M
Mekong River are high,
backwater may flow into the tributaaries. These floods
f are not short term like flash flo
oods; they
may continue for weekss. In the shalloow areas along the Mekong River dow wnstream of Vientiane,
V
wer reaches off tributaries fface this type
several low e of flooding.
Floods in th
he Cambodia an floodplainn occur alongg the Mekong
g downstrea m of Kratie to
t Phnom
Penh and innclude flooding around t he Tonle Sapp Lake and th
he Tonle Sapp River that flows
f into
and out from the Tonle Sap Lake.
ood Risk Assessm

Floods in the Mekong Delta occur in the Mekong and Bassac Rivers and their floodplains,
including the areas of colmatage canals that divert and control the flow from and to the rivers.
In the Mekong Delta water levels rise slowly due to storage in the Tonle Sap Lake and in the
Mekong River floodplains. Flooding here is recognised as essential for soil fertility, biodiversity
and aquaculture. At the same time, it hampers use of agricultural land. The flood levels in the
Mekong Delta downstream are essentially the result of upstream and lateral inflow and
downstream water levels of the sea.
Table 2‐1 offers an overview of flood hazards of these different types of floods in the sub‐areas
of the LMB.
Table 2‐1 Overview of flood types per sub‐area (sub‐basin)

Assessment of hydrological
Sub‐area Type of flood Issue Assessment of flood hazard
hazard
SA1 Tributary yes From rainfall statistics, Rainfall‐runoff + hydraulic
Northern discharge statistics or regional model/satellite imagery
Laos approach
Mainstream yes Annual Flood Report Hydraulic model/satellite imagery
Combined no ‐ ‐
Northern discharge statistics or regional model/satellite imagery
Thailand approach
Combined yes HYMOS/other databases As above
SA3 Nong Tributary yes For some basins via statistics Models available for Nam Loei, Huai
Khai/ rest via regional approach Mong and Nam Songkhram/
Songkhram satellite imagery
SA4 Central Tributary yes From rainfall statistics, Models available for Nam Ngum, Xe
Laos discharge statistics or regional Bang Fai and Se Bang Hieng/
approach satellite imagery
SA5 Mun‐ Tributary yes For 12 via statistics, rest via Hydraulic models/satellite imagery
Chi RID data or regional approach
HYMOS/RID/other databases
Combined no ‐ ‐
Southern discharge statistics or regional model/satellite imagery
Laos approach
SA7 Se Tributary yes From rainfall or discharge Rainfall‐runoff + hydraulic
San/Sre statistics or regional approach model/satellite imagery
Pok/Se Mainstream no ‐ ‐
Kong Combined yes HYMOS/other databases Hydraulic model/satellite imagery
SA8 Kratie Tributary yes From rainfall statistics Rainfall‐runoff + hydraulic
model/satellite imagery
Mainstream yes Annual Flood Report Satellite imagery u/s Kratie + model d/s
Kratie
Combined yes No data ‐
SA9 Tonle Tributary yes MRC database Rainfall‐runoff + hydraulic
Sap model/satellite imagery
Mainstream yes HYMOS/other databases Hydraulic delta model/satellite imagery
Combined yes MRC database As above
SA10 Tributary yes HYMOS/other databases rainfall‐runoff+ hydraulic
Mekong model/satellite imagery
Delta Mekong Delta yes HYMOS/other databases Hydraulic delta model/satellite imagery

2.3 Tributary flood hazard assessment
2.3.1 General
Flood hazard assessment procedures rely on hydrological monitoring infrastructure in order to

provide measurement data, yet such monitoring capacity on the tributaries of the Mekong is
currently limited. Only a limited number of basins are equipped with rainfall and water level
recorders. In most locations only daily rainfall data are available. The water level gauging
network and the monitoring intervals are usually inadequate to properly describe flash floods.
Often, only daily water levels are measured and this is unsuitable for flash flood analysis. Flow
discharge measurements usually cover only the lower stages and occasionally include flood
data. Dependent on data availability one of the following three procedures is feasible for
estimating flood hazards:
1. Flood hazard derived from rainfall extremes.

This method is recommended for small basins when little or no historic data is available on
flows. It starts with the specification of the meteorological hazard is terms of design
rainstorms of desired return periods. The design rainstorms data are adjusted to net
rainstorm data and plotted on design discharge hydrographs for the required location using
appropriate unit hydrograph and hydrologic routing techniques to arrive at the hydrological
hazard measure. Finally, the flood hazard is determined by transforming the peak level data
of the discharge hydrographs into water levels by means of flow simulation techniques
using a hydraulic model.
2. Flood hazard derived from observed flows.
When a homogeneous water discharge data series (of >15 years) is available for a specific
river reach, the hydrological hazard can be derived from a statistical analysis of annual flood
peaks and flood volumes. A hydraulic model of river and floodplain can then be used to
convert the hydrological hazard into a flood hazard. Data from representative flood
hydrographs, covering the full spectrum of flood peak and flood volumes, can be converted
into water level data using this hydraulic model. Water level and flood volume data can be
used in a Monte Carlo procedure to derive the frequency of different water levels as a
function of time and space. A Digital Elevation Model (DEM) can then be used to generate
flood hazard maps.
3. Flood hazard determined from regional flood statistics.
The distribution of annual flood peak can also be obtained from regional flood statistics,
when available, if the average annual flood peak is known. For streams without water level
recorders and discharge measurement equipment, this value may be estimated from basin
characteristics.
2.3.2 Flood hazard derived from rainfall extremes
The method of using rainfall extremes for the assessment of the flood hazard includes the
following steps:
1. Collection of relevant information;
2. Development of a database and conduct of field visits;
3. Determination of the design rainstorms for different return periods;
4. Plotting of design rainstorm information on design hydrographs; and
5. Use of design hydrographs to determine design levels.
These steps are elaborated below.

Step 1 Collection of relevant information

Data required for flood hazard assessment includes:
1. Digital Elevation Model (DEM) of the area and topographical maps;
2. Soil maps, geological maps and land use maps (past and present);
3. Data on historical floods;
4. Layout of the hydro‐meteorological network in the river basin, its operation and
maintenance;
5. Time series data on rainfall, climate, and, when available, water levels, stage‐discharge
measurements, discharges and sediment of the river basin;
6. Hydraulic infrastructure specifications, dimensions and operation in times of flooding;
7. Survey data of the river (in cross‐sections) and floodplain (DEM) at the reach and upstream
and downstream of the reach under investigation and of relevant hydraulic parameters
including river bed composition, vegetation, embankment elevation, etc.;
8. Data on the morphological development of the river and its environs; and
9. Planned land use and infrastructural construction.
Step 2 Database development and field visits

The collected information is next organized in a database consisting of GIS‐related and time
series data. In the database it is important to make a clear distinction between original/raw
data, data under validation, and fully validated data.
There are a number of procedures for preparing data for use in models. For example, rainfall
data in weekends is often suspect since there may be no monitoring staff available. Thorough
validation, correction and completion procedures should be applied to improve the reliability of
the data. Since the flood hazard assessment relies heavily on quality rainfall data, available
records have to be thoroughly reviewed to assure their quality. Graphical and tabular
comparisons are often useful in preparing the data. In some cases, data from the nearest
neighbour (i.e., closest monitoring stations) can be used for cross checking purposes and to fill
in missing data. A double mass analysis can also be used. Standard textbooks offer detailed
explanations of these procedures.
In this data collection and validation process one or more field visits is essential to validate
collected information and to assess its quality. Experts should visit key rainfall monitoring
stations to review their monitoring practices and. Interviewing residents and officials on the
extent of historic floods, land use patterns and construction. The field visits can also be used for
collection of additional data or establishment of a (temporary) monitoring program to check on
the validity of the data.
Step 3 Determination of the design rainstorms for different return periods

The determination of design storms involves the following steps:
1. Assess the time of concentration (Tc) of the basin up to the location for flood hazard
assessment. Time concentration is defined as the travel time required for a drop of water to
travel from the uppermost part of the catchment to the outlet. Tc is a measure that
expresses the time delay between rainfall and flood runoff.
2. Determine the basin rainfall from point rainfall data for the smallest available interval if
applicable.
3. Develop intensity (or depth)‐duration‐frequency curves for specific rainfall return periods of
2 to 100 years, for durations of 5 minutes to 24 hours. If no short interval rainfall data is
available, short duration rainfall intensities can be derived from daily rainfall data. In such
cases, Extreme Value Type 1 (EV1) or Generalized Extreme Value (GEV)‐distributions can be

fitted to the observed distribution of annual maximum daily rainfall, and the rainfall for
selected return periods can be estimated. These values should be adjusted to 24 hr values
by applying a correction factor of 1.12. Then regionally acceptable values for the
coefficients b and n in the following intensity duration relation are to be obtained:
a
I
(b  D)n (2.1)
where: I= rainfall intensity (mm/hour)
D= rainfall duration (hours)
a,b,n = constants
Values for coefficients for Phnom Penh e.g. are b = 0.23 and n = 0.86 (Watkins et al., 1984).
4. Apply the Rainfall Ratio Method to derive the rainfall for durations shorter than 24 hrs. The
Rainfall Ratio method assumes that the constants in equation (2.1) apply for all durations
less than 24 hours and all return periods. Thus, the rainfall of duration D for any return
period is derived from the 24 hour rainfall extreme via equation (2.1) as follows:
D  b  24 
n
RRD   
24  b  D  (2.2)
where: RRD = rainfall ratio for conversion of 24 hr rainfall into D hr rainfall
D= rainfall duration (hours)
5. In case of single point rainfall, the point rainfall data needs to be adjusted for the variability
of rainfall over the area (spatial distribution). This can be done through the application of an
Areal Reduction Factor (ARF). The ARF is a function of the basin area (A) and rainfall
duration (D) and can be derived through statistical processing of single point rainfall data.
For example in the case of convective rainfall in Cambodia for durations less than or more
than 8 hours and where A is measured in km2,, the ARF is given as follows:
for D  8hrs : ARF  1  0.04 D 0.33 A0.50

for D  8hrs : ARF  1  0.02 A0.50 (2.3)
6. The design rainstorms for selected return periods can be obtained by the incremental
intensity of the rainfall totals of the calculated durations up to the time of concentration:
RA, D j  RA, D j1  mm 
Ij    for : D j  D j 1
D j  D j 1  hr 
(2.4)
where: Ij = jth incremental intensity (mm/hr)
RA,D = areal rainfall of duration D
7. The incremental intensities can then be arranged around the highest intensity (peak value).
The graphical location of the peak is determined by the storm advancement coefficient r
which is the ratio of the time to peak relative to the total duration of the design storm Tc.
The advantage of this procedure is that up to Tc all design intensities are included in the
design storm. Another option to determine the design rainstorms is by using triangular
hyetographs, where the occurrence of the peak value is determined by the storm
advancement coefficient. When possible the arrangement is according to observed
hyetographs in the region. The availability of short duration rainfall data is then a necessity.
8. Repeating the above step for required return periods yields a set of hyetographs.

MRC Flood
d Management and
P Com
Step 4 Using the design storm to

o determine tthe design hyydrograph
The convversion of dessign rainstorm
m into a desiggn hydrograpph involves thhe following ssteps:
1. Deteermination off the net rain nfall by estim
mating rainfall losses, to be distributedd among the
ordin
nates of the hyetograph;
h
2. Deveelopment of a unit hydrograph of thhe basin up to the location for whicch the flood
hazaard assessment is undertaken, and
3. Convvoluting the net hyetograph to the design hydrograph usin ng the unit hydrograph
deveeloped underr the previouss activity.
Rainfall losses are connsidered not contributingg to the flood d hazard. Onee of the methhods used to
determin ne rainfall lo
osses is the Soil
S Conservaation Service e Curve Num mber methodd (SCS‐Curve
Number method or SCS‐CN),
S applied to wet aantecedent so oil moisture conditions. TThis method
requires information on soil types, land use aand condition n of the landd. The SCS‐CN N method is
extensiveely describedd in hydrological textbookss (e.g., Viessm
man, 1989).
To devellop a unit hyydrograph off the basin itt is recomme ended that use
u is made oof the Clark
method. Clark's meth hod requires estimation
e off three basin parameters for the derivaation of unit
hydrograaph and is baased on the concentration
c n time Tc, onn the routing
g of a time‐a rea diagram
with a time base equ ual to the co
oncentration time Tc, and d on a simula
ation of channnel storage
through a linear reserrvoir.
When th he unit hydrograph has been estabblished, the net design rainstorm ddata can be
convolutted to a design hydrogra aph, which ccomprises the hydrological hazard wiith a return
period eequal to thee return periiod of the ddesign rainsttorm. This method m of reeducing the
abstractiions that cause the difference
d beetween extrremities of storms andd discharge
hydrograaphs under drier
d initial conditions,
c iss valid in vie
ew of the wet
w antecede nt moisture
conditionns assumed in the comp putation. By repeating th his process for
f design raainstorms of
differentt return perio
od peak wate er levels, it iss possible to obtain data for dischargge values for
the locattions and to plot
p a frequen ncy distributi on.
The abovve proceduree is applicablee to relativelyy small catchm

ments, roughhly up to 1,0000 km2. With
some mo odification, this
t procedure can also bbe applied to larger area as. For large r areas, the
calculatioons should use a segmentted approachh. This will transform exce ess rainfall innto runoff to
allow sp patial variation in the excess
e rainfaall and runo off characterristics. In maaking these
calculatioons, the basin up to thee location off interest can n be divided into segme nts and the
rainfall‐runoff transfoormation cann be applied to each seggment. The hydraulic
h outtput of each
segmentt is then addeed to the are ea of interesst. This routinng can be doone by simplee hydrologic
routing mmethods or with
w the hydra aulic model eextended to thet outlet of the
t uppermoost segment.
Step 5 Using design

n hydrographs to determinne design levvels
By meanns of hydraulic model simu ulation, the ppeak dischargges are transfformed into w
water levels
of requiired return periods. Th he hydraulic model is schematised according to detailed
bathymeetric survey data
d on the river and flooodplain, and the
t estimated or calibrateed hydraulic
roughnesss. The calcuulations shou uld take into account exissting flood protection
p meeasures and
include p
planned prottection workss for investiggation of alte ernatives, sin
nce these meeasures may
cause futture water levels to rise in
n areas furtheer upstream.
2.3.3 Example of flood hazard

d derived from
m rainfall extremes
To demo onstrate how w the above procedure w works, a hypo

othetical basin on the Uppper Se San
River neaar Pleiku, Vieet Nam can be b used as a n example. For
F the calcu ulations, we aassume that
only daily rainfall serries are available. For thee development of a desig
gn storm calcculation, the
followingg parameterss have also be een assumed :
Guidelines Flood Risk Asseessment ‐ 18 ‐ May 2010

 Basin area of 50 km2;

 Rainfall intensity parameters for disaggregation of daily rainfall (equation 2.1) b = 0.23 and
n = 0.86 apply;
 Storm advancement coefficient r = 0.3;
 ARF for convective storms in tropical countries apply; and
 Rainfall conditions are similar to those at Pleiku with an orographic adjustment factor of
1.25.
Daily rainfall series data for Pleiku (monitoring station 140703) are available for the years 1927‐
2006, though with some gaps. After validation and homogeneity tests, EV1 and GEV
distributions have been fitted to the annual maximum rainfall of the monitoring station. The
results are shown in Table 2‐2 and in Figure 2‐3.
Table 2‐2 Parameters of EV1 and GEV distributions fitted to annual maximum daily rainfall at Pleiku and
rainfall values (mm) for selected return periods.
Parameters/
Pleiku
Return period (years)
EV1 GEV
k ‐ 0.0048
α 30.12 30.26
u 98.08 98.11
2 109 109
5 143 143
10 166 166
25 194 194
50 216 215
100 237 236
500 285 283
1000 306 304
Annual maximum daily rainfall at Pleiku, 1927-2006

350
Observations
300 EV1 distribution
GEV distribution
Annual maximum rainfall (mm)
250
200
150
100
50
0
-2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00
Reduced EV1 variate
Figure 2‐3 EV1 and GEV‐fit to annual maximum daily rainfall at Pleiku for the years 1927‐2006.

The depth‐duration‐frequency curves have been derived for intervals of 5 min to 24 hrs starting
from the daily rainfall extremes for return periods (T) of T = 2 to 100 years, using equation (2.2).
The results are presented in Table 2‐3 and shown in Figure 2‐4.
The calculation for the T = 50 year design storm is shown in Table 2‐4. The result is presented in
Figure 2‐5. It is derived from a re‐ordering of the derived incremental intensities. Here, a basic
interval of 15 minutes has been assumed. The sensitivity of the design hydrograph for the
choice of this interval should be tested, once the basin Unit Hydrograph is derived. This
procedure is to be repeated for other selected return periods that will be used in the flood
hazard assessment.
Table 2‐3 Short duration rainfall (mm) at Pleiku derived from daily rainfall extremes.
Duration Rainfall ratio Return Period in years

hrs RR 2 5 10 25 50 100
5 min 0.083 0.15 18 23 27 32 35 39
10 min 0.167 0.24 29 38 44 52 58 63
15 min 0.250 0.30 37 49 56 66 73 81
30 min 0.500 0.42 52 68 79 92 102 112
1 hr 1 0.54 66 87 101 117 131 144
2 hr 2 0.65 79 104 121 141 157 172
4 hr 4 0.75 91 120 139 162 181 198
8 hr 8 0.84 103 135 157 183 204 224
12 hr 12 0.90 110 144 167 196 218 239
18 hr 18 0.96 117 153 178 208 232 254
24 hr 24 1.00 122 160 186 217 242 265
Day Day 109 143 166 194 216 237
Depth-duration-frequency curves for Pleiku

300
T = 2 year
T = 5 year
250 T = 10 year
T = 25 year
T = 50 year
T = 100 year
200
Rainfall (mm)
150
100
50
0
0.01 0.10 1.00 10.00 100.00
Duration in hrs
Figure 2‐4 Depth‐duration‐frequency curves for Pleiku.

Table 2‐4 Computation of design storm ordinates for 15 min intervals.
Time (hrs) 0.25 0.50 1.00 2.00

Point Rainfall (mm) 73 102 131 157
Correction for orography 92 128 164 196
ARF 0.55 0.64 0.72 0.78
Areal rainfall (mm) 51 82 117 152
Incremental intensity (mm/hr) 202 127 70 35
No of 15 min increments 1 1 2 4
2 hr-design rainstorm for Pleiku, T = 50 years

250
200
Rainfall (mm/hr)
150
100
50
0
0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Duration (hrs)
Figure 2‐5 A 2‐hour‐design rainstorm diagram for the Upper Se San River derived from measures used for
Pleiku.
In the development of the flood hazard, the next step is the transformation of the design storm
into the design hydrograph. This involves, first, the assessment of the excess rainfall. For this,
the Curve Number Method is useful. Next, the Unit Hydrograph of the basin is determined for
which the Clark Method can be used. This method is based on the routing of a time‐area
diagram (with a time base equal to the time of concentration Tc) through a linear reservoir,
simulating channel storage. In the diagram of the hypothetical basin, isochrones representing
points of equal travel time to the segment outlet (downstream basin limit) can also be
constructed. (See Figure 2‐6.)
The time area diagram fully represents the physical runoff characteristics from (mountainous)
catchments. This diagram can be thought of as the outflow from the catchment if only
translation and no deformation takes place of an instantaneous unit supply of rain. Flow paths,
slopes and hydraulic roughness determine the speed of water outflow and these characteristics
are easy to obtain from a Digital Elevation Model (DEM) or a topographic map. In addition to
the concentration time, plotting of the time area diagram requires estimating a water recession
coefficient, k, for routing. Note that the output from the reservoir is representative for the
instantaneous unit hydrograph (IUH). In order to develop the Unit Hydrograph, the IUH data
must be transformed into a Unit Hydrograph (UH) of the required interval by multiplying with

observed rainfall depths. The two parameters, Tc and k, can also be obtained from observed
rainfall and discharge hydrographs.
segment
5
6 4
7
2
3
1
isochrones
1-hour
Unit
Time-area diagram
UH(t)
Hydrograph
area
time
7 6 5 4 3 2 1 Linear
reservoir
time
Tc
k
IUH UH
Figure 2‐6 Principles of the Clark Unit Hydrograph method, shown in various diagrams that are related to
each other.
The time of concentration is equal to the time interval between cessation of rainfall and the
time the hydrograph has receded to its inflection point. Note that this method is only applicable
when drainage conditions and infrastructure are constant and are not expected to change. If
conditions are not constant, the time of concentration can be determined from physical
features of the catchment segment such as length and slope. The recession coefficient, k, can be
calculated from the surface runoff hydrograph by measuring the slope of the curve in the area
where it descends (the area of flood water recession). The Unit Hydrograph can be used to
generate the excess hyetograph and the data can then be used to calculate the design flood for
the selected return periods. Using these results, it is possible to generate a frequency
distribution of the floods resulting from the selected rainfall duration (equal to the time of
concentration).
It is to be verified whether storms of shorter or longer duration lead to higher peak discharges.
The height of the peak discharge relative to storm duration is dependent on the intensity
distribution in the rainstorm and the way the losses are distributed over the rainfall ordinates.
The data for frequency distribution of flood peak can be transformed to design flood levels
through calculations using measurements of backwater. If upstream water storages are
significant, then the full flood hydrograph should take the entire hydraulic network into
account, using realistic initial values for the storage capacities.

2.3.4 Flood hazard determined from observed flows
When a water discharge data series of sufficient length (≥ 15 years) is available, the hydrological
hazard can be determined by statistical analysis of these data, provided that the time series
data are reliable and homogeneous. This procedure involves the following steps:
1. Collection of relevant information;
2. Database development and field visits;
3. Assessment of the hydrological hazard in terms of flood peak water levels and flood
volumes;
4. Conversion of selected flood hydrograph data to flood level data using a hydraulic
model;
5. Flood hazard assessment by means of a Monte Carlo sampling procedure, interpolating
between the simulated water levels based on samples from the flood peak water levels
and volume distributions; and
6. Preparation of flood maps by selection of flood levels, durations etc. of the same return
period.
Step 1 Collection of relevant information

The required data for this type of flood hazard assessment does not differ much from the one
above. (See Step 1 in the procedure described in Section 2.3.2.) Here, emphasis is on several
factors in the hydrometric (water measurement) network layout and operations. These include
sampling and measurement procedures, monitoring station and equipment maintenance, gauge
ranges and re‐settings, available water level data series, stage‐discharge data and assumed
discharge ratings with extrapolation procedures and discharge series data.
Step 2 Database development and field visits

(See also Step 2 in the procedure described in Section 2.3.2.)
Thorough validation, correction and completion procedures (filling in missing data) should be
applied to improve the reliability of the water discharge data. This requires inspection of all
hydrometric monitoring stations relevant to assessment of the hydrological hazard and
validation of the data from key stations. Flood marks should be levelled and discussed with local
residents. Sometimes there is lateral water inflow. In such cases, when rainfall data are used to
determine the lateral inflow, the rainfall network should be inspected as part of the validation
procedure. The hydraulic infrastructure and flood protection works should also be inspected on
field visits, to ascertain their capacity and operations in times of floods and to determine their
impact on the shape of the hydrograph curves as a result of water storage and/or backwater.
Construction in the basin, which can affect the hydrograph (specifically the peak value), should
be quantified.
Screening of historical water levels involves inspection of the historical patterns of hydrographs,
comparison with hydrographs at nearby monitoring stations, stage relation curve analysis to
identify outliers and shifts in gauge locations, and comparison of hydrographs with hyetographs
to detect anomalies in the rainfall – runoff process. Special attention should be paid to
differences between instantaneous peak water levels and maximum daily average water levels.
The latter information is generally only available in the databases. The peak values of the
hydrographs should later be adjusted in case of differences between instantaneous peak water
levels and maximum daily average water levels.
Stage relation curves describing the relation between water levels and discharges at gauging
stations may be applied to fill in gaps in the data of a station (with appropriate time shifts to
account for flood wave celerity). With respect to the stage‐discharge data the stage data is first

compared with the stage‐hydrograph. The computation of the discharge from field data is
checked and its accuracy is assessed. Possible backwater and unsteady flow effects are to be
identified by means of a first order backwater calculation and assessment of the Jones
correction, respectively. The establishment of the discharge ratings is to be inspected, with
special emphasis on the extrapolation. The latter should be physically based taking into account
flood protection works in and downstream of the river section.
The next step is to apply the discharge rating to the water level data series in order to compute
the discharge data series. A comparison is made with the series in the database. Discharge
hydrographs at neighbouring monitoring stations can be compared graphically and the
consistency of the data can be verified by means of water balances and double mass analysis.
Similar comparisons with data at neighbouring monitoring stations can be used to validate peak
values. Finally, peak values should be examined, graphically, statistically and using documents
on drainage conditions to spot possible trends indicating problems in the data and other
anomalies. Data should be corrected to eliminate these trends and anomalies.
Special attention should be paid to the variability of the discharge ratings at locations where
there have been morphological changes in the river bed. The variation in water levels for
particularly high discharges over the years should be calculated and then applied to assessing
the additional uncertainty in water levels for the specific discharge rating used in the model.
Step 3 Assessment of the hydrological hazard

The features of the hydrograph which are of importance for flood hazard assessment include
the peak value considering maximum water levels and the flood volume. The flood volume
determines the duration of flooding. The following tasks are required in the assessment
process:
1. For flood peak water levels:

 Creation of a homogeneous series of annual maximum flows of sufficient length;
 Application of extreme value analysis to the annual maximum flow series to derive
frequency distributions for the flow extremes;
 For locations upstream of Vientiane it is advised using comparisons of the obtained
distribution with the regional flood frequency curve described in Section 2.3.5.
2. For flood volumes:

 Selection of data from 20 to 25 of the largest recorded floods;
 Derivation of a regression line that shows the relationship between flood peak
discharge (Q) and flood volumes (V) as V = f(Q) and its scatter (the deviation from the
fitted line), or “standard error”, (Se). The regression line gives the “average” or
“expected” value of the flood volume, for given values of the flood peak. The expected
flood volume increases in relation to the increasing flood peak. Note that the
relationship is not perfectly described by the fitted line. There is some scatter (volumes
above and volumes below the expected values of the fitted line). Figure 2‐7 offers an
example of such a regression line and the deviation from the line for a single point. The
value, Se, is the standard deviation of all points with respect to the regression line. It
can be used (in step 3) to take the effect of the scatter into account;
 Derivation of representative dimensionless flood hydrographs from the selected
hydrographs. The base of the hydrograph should be long enough to represent the full
duration of characteristic floods (from M days before until N days after the peak water
level). For each flood assign the time of Qmax (the peak discharge of the flood) as the
initial time (t) such that t=0. Then, determine for the range of times (t = ‐M, …, t=0, ...,
t=N) the water discharge for each time relative to Qmax (=Q/Qmax). Repeat the
procedure for all other selected peak discharges (for the different return period floods).

Then, apply a frequency analysis on Q/Qmax for each time measurement (t = ‐M, …,
t=0, ..., t=N). The result will be a distribution function of Q/Qmax for each time at which
measurements are taken. Thus, for any give probability, p, of (non‐)exceedance the
value of Q/Qmax will be known for all time at which measurements are taken.
Connection of these values yields the (dimensionless) hydrograph for p. In this manner,
for a variety of p‐values, hydrographs can be generated for use in the next step.
3. For creation of a flood peak water level‐flood volume matrix for use in hydraulic
simulations:
 Selection of flood peaks of return period 2, 5, 10, 25, 50 and 100 years;
 Determination of a range of flood volumes for the selected flood peak values, based on
the scatter plot data around the flood volume (V) ‐ flood peak discharge (Q) regression
line relationship, V = f(Q). For each selected peak value (the different return period
floods), we can generate a very low (R‐1.96 Se), a low (R‐Se), medium (R), high (R+Se)
and very high flood volumes (R+1.96 Se) – R being the mean of the flood volumes ‐ by
scaling of the representative flood hydrographs to the flood peak value of selected
return periods. This results in 30 (6 return periods x 5 flood volumes = 30) hydrographs;
 Adjustment of flood hydrographs for differences between daily average flow maxima
and instantaneous peak flows;
 Lateral water inflows, assumed to be relatively small, can be defined as a percentage of
the selected flood hydrograph measurement.
2500
regression flood peak/flood volume
2000
flood volume
deviation from
regression line
1500
1000 regression line
500
0
0 20 40 60 80 100 120 140 160 180
flood peak
Figure 2‐7 Hypothetical scatter plot of data points and fitted regression line showing the relationship
between flood volumes and flood peak water levels. The red arrow in the Figure shows the
deviation from the regression line for a single observation.
Step 4 Conversion of flood hydrographs to flood levels using a hydraulic model

There are a number of options of hydraulic models for use in converting flood water data into
flood levels. An accurate hydraulic model is needed for the river and floodplain being analysed.
The model’s boundaries should be chosen in accordance with the hydrometric network
equipment and the boundaries of the area of analysis. It is suggested using a one‐dimensional
(1D) – two‐dimensional (2D) hydraulic model, where rivers are represented as 1D and the
floodplain as 2D, both derived from a DEM. This approach has clear advantages over a quasi‐2D
approach, which is a 1D model with storage cells for the floodplain, where the need to make

qualitative assumptions or estimates on the connection between the river and floodplain
introduces subjectivity. The 1D‐2D model has also advantages over a full 2D model, since
narrow rivers cannot be properly represented in the 2D model. Any selected model should be
calibrated for river and floodplain conditions representing the base case. Special attention
should also be given to the position of the river bed in comparison with possible river bed shifts
due to natural or human induced morphological developments.
The 30 selected hydrographs are subsequently routed in the model through the river and
floodplain system, and the water levels are computed as functions of time and space which are
placed in a database for use in a Monte Carlo simulation procedure for flood hazard
assessment. These simulations can be repeated to test the results of construction plans that are
under consideration, after adjustment of the hydraulic model for the different scenarios.
Step 5 Flood hazard assessment

The probability of exceedance of a water level at any location can be determined using Monte
Carlo simulations. Monte Carlo simulations of water level exceedance use a large number (N) of
paired samples of the peak discharge and flow volume. In this procedure, the derived frequency
distribution of peak water level discharges is used to sample the peak discharges. Subsequently
the flood water volume can be derived from the peak discharge, taking the sum of:
1. the regression line to derive the “expected” volume, given the value of the peak water level
discharge; and
2. a sample from the normal distribution with mean 0 and standard deviation Se (see step 2)
to account for deviation from the expected volume.
Application of this procedure ensures that sampled volumes are correlated with the sampled
peak discharges to reflect the relationship in the raw data. For each sampled pair of water
volumes and peak discharges, the water levels at the areas under study are derived from 2‐
dimensional interpolations of the grids of the 6 x 5 simulated hydrographs described above. The
result is an N‐year synthetic series of water levels (h1, …, hN,) at each location. From this series
the T‐year return period flood water levels can be easily derived with an equation such as
Gringorten’s formula:
ri  0.44
pi 
N  0.12 (2.5)
where: pi = probability of exceedance of synthetic water levels of height hi for I,

the particular area of interest under study;
N= total number of Monte Carlo samples;
(1 = highest, N = lowest), and
ri = rank number of the maximum synthetic water levels hi
Step 6 Preparation of flood maps

The mapping of the hydraulic modelling results is described for mainstream flows in Section 2.7.
For tributary floods the mapping principles are the same as those for mainstream flows, but the
data interpolation is more critical because of the hilly terrain.
2.3.5 Flood hazard derived from regional flood statistics
Adamson (2007) proposed the use of a regional approach to flood hazard assessment. The
method has been adapted for use in the upper part of the LMB for 16 drainage basins with

areas ranging from 200 to 6,000 km2 and for 1 basin of 19,700 km2. The technique is applicable
provided the regional curve for the upper part of the LMB is valid for the particular basin. Its use
involves the following steps:
1. Creation of a regional sample of annual maximum flood peak water levels by pooling
together the individual annual maximum values, scaled to their individual mean annual
flood value. Discordance and homogeneity tests should be applied to assure regional
homogeneity of data from the selected monitoring stations (see Hosking and Wallis, 1997).
2. Fitting of an extreme value distribution to the scaled ranked annual extremes. For the LMB
basins considered by Adamson (2007), the Two Component Extreme Value (TCEV)
distribution fits the observed frequency distribution of these pooled values. The regional
curve for the upper part of the LMB is shown in Figure 2‐8. The TCEV‐distribution was
chosen to account for different phenomena creating the water discharge extremes (i.e.
monsoon and typhoons). Typhoons create extremes far beyond the monsoon levels.
3. Applying the TCEV‐values to un‐gauged sites. A method is applied using regional data for
the mean annual maximum floodwater discharge that describes the relationships between
floodwater discharge and one or more climatic and/or basin characteristic(s). This
relationship allows for floodwater discharge rates to be determined anywhere in the region
if those other characteristics are measured. For the upper part of the LMB, the following
relationship between floodwater levels and drainage area was identified:
Qave  A 0.75  err . (2.6)
where: Qave = mean annual maximum floodwater discharge rate (m3/s)

A= drainage surface area (km2)
The regression error (dispersion of the data around the fitted line) is considerable (Se is
about 50% of Qave), suggesting that additional descriptive variables characterising the area
are needed to fully describe the floodwater discharge rate.
Estimated TCEV distribution of the ratio (%) of the T-year event to the mean annual flood for
pooled regional sample (n=530) (Source: Adamson (2007))
700
Ratio (%) of T-year flood to mean annual flood
600 Lower 95% CL

TCEV-values
Upper 95% CL
500
400
300
200
100
0
1 10 100 1000
Return Period (years)
Figure 2‐8 TCEV‐distribution of the ratio of the T‐year return period event to the mean annual flood for the
pooled regional sample (Upper part of LMB, from Adamson, 2007).

In using the regional approach, the following remarks are made:

 In principle the method is very attractive and easy to use, provided that the base data can
be properly validated and a clear relationship can be established between mean annual
floodwater discharges and meteorological and river basin characteristics.
 The data set used in Adamson (2007) has been taken from the MRC database. Inspection of
the data reveals inconsistencies in the data for flow extremes across different monitoring
stations.
 Using Adamson’s data set, Qave appears to correlate well with the mean annual floodwater
flow. This, in turn should be directly related to annual rainfall which are generally available
for un‐gauged streams in view of the better distribution of rain gauges. Furthermore, other
data can be used in other equations that predict maximum floodwater discharge rate or
rainfall (such as stream length, basin slopes and land use data from available DEM), to
improve the estimate of Qave values.
 An adjustment is required for instantaneous peak values relative to daily average maximum
values in the above approach. In the case of flash floods a considerable adjustment is
expected as the flood may last only for a number of hours.
2.4 Mainstream flood hazard assessment
2.4.1 General
The hydrological hazard along the Mekong River is determined by the discharge peaks on the
river that generate the maximum water levels and by the flood volumes determining the
duration of flooding. Conversion of hydrological hazard data into flood hazard data requires
using a hydraulic model of the river and floodplain to convert water flow data into water level
data. Data from selected hydrographs covering the full spectrum of possible combinations of
flood peaks and of flood volumes can be used in the model. Special attention should be paid to
variations of the river bed due to natural or human induced morphological changes, which
affect the water level discharge ratings along the river. The Monte Carlo simulation sampling
technique can be used to derive exceedance probabilities of water levels and floodwater
damage. The simulation generates results for a large number of combinations of peak
discharges, flood volumes and river bed conditions and this data can be used to estimate
floodwater damage.
The flood hazard assessment for mainstream floods includes the following steps:
1. Data collection;
2. Database development and conduct of field visits;
3. Data validation and processing;
4. Hydrological hazard assessment in terms of flood peaks and flood volumes;
5. Flood hazard assessment using a hydraulic model that transforms selected flood
hydrograph data to flood level data and applies Monte Carlo sampling to determine flood
levels and durations for distinct return periods;
6. Preparation of flood maps through selection of flood levels, durations, and other features of
the same return periods. (See Section 2.7).
The steps are elaborated in the following sub‐sections.
2.4.2 Data collection
For assessment of the hazard of mainstream floods, the following data and information are
required:

1. Identification of key hydrological monitoring stations (Qh‐stations) along the Mekong River
in the area of study;
2. Survey data of the Mekong River (full river bathymetry) and of the floodplain in the study
area and beyond and relevant hydraulic parameters including river bed material,
vegetation, embankment elevation, etc.;
3. Updated Digital Elevation Model (DEM) of the floodplain in the area under study;
4. Soil maps of the floodplain;
5. Land use maps of the floodplain, including past, present and expected future conditions;
6. Data on hydraulic infrastructure in and downstream of the study area, including its
dimensions and operation in times of flood and planned future developments;
7. Historical flood maps (including flood levels, extent, depth and duration);
8. Layout, operation and maintenance information for the hydro‐meteorological network
around the study area including information on the nearest upstream and downstream
monitoring stations along the Mekong River and monitoring stations on the tributaries
where relevant;
9. Time series data of water levels, stage‐discharge measurements, discharge ratings and
discharges series of the relevant monitoring stations as well as series data of rainfall and
climatic variables in the region;
10. Changes that have occurred in runoff characteristics of basins upstream of the study area
that affect the homogeneity of the historical water discharge data series (including
construction of reservoirs and land use changes).
Below, is a list of the agencies in the MRC Member Countries that are responsible for collecting
and providing this data.
Cambodia
Ministry of Water Resources and Meteorology (MOWRAM)

 Department of Hydrology and River Works (DHRW), for water level, discharge and rainfall
time series;
 Department of Meteorology (DOM), for rainfall, climatic data (air temperature, relative
humidity, wind direction, wind velocity, duration of sunshine, radiation)
Ministry of Public Works and Transport (MPWT)

 Department of Waterways River and Lakes for bathymetric data, hydrographic atlases;
 Department of Road and Transport
Ministry of Agriculture, Forestry and Fishery (MAFF)

 Department of Agronomy and Agricultural Land Improvement for soil maps and land Use
maps)
Phnom Penh Municipality:

 Department of Public Works and Transport of Phnom Penh Municipality for urban flood
protection, dykes/drainage in and around Phnom Penh
Ministry of Land Urban Planning Land Use and Construction (MUPLC)

 Department of Land Use Planning, Commune Land Use Planning (CLUP)
Lao PDR
Water Resources and Environment Administration (WREA)

The WREA is a new government agency that combines what were previously the
Department of Meteorology and Hydrology (DMH) of the Ministry of Agriculture and the
Waterways Administration Division (WAD) of the Ministry of Communication, Transport,

Post and Communication. WREA is responsible for collecting and providing all related water
data in Lao PDR as well as for the establishment and management of the country’s
hydrological and meteorological network.
The Hydrographic Atlas data of the Mekong River mainstream has also been transferred to
the WREA.
Thailand
Ministry of Agriculture and Cooperatives (MOAC)

 Royal Department of Irrigation (RID), for hydro‐meteorological data in most of the Mekong
River tributaries, especially the Nam Mae Kok, Nam Mae Ing, Nam Chi and Nam Mum
Rivers.
Ministry of Natural Resources and Environment (MNRE)

 Department of Water Resources for management of hydro‐meteorological networks
(previously under the Department of Energy Development and Promotion), including all
the Thai Mekong River mainstream monitoring stations. Climatic and water quality data
collection previously supported by the Mekong Committee were also collected and
provided by this department including bathymetric survey data and hydrographic atlases.
Ministry of Information and Communication Technology (MICT)

 Department of Meteorology (TMD), for meteorological data, climatic data, and weather
forecasting.
Ministry of Interior (MOI)

 Department of Disaster Prevention and Mitigation, Department of, Local Administration,
Department of Public Works, for urban and rural planning and flood mapping.
Viet Nam
Ministry of Natural Resources and Environment (MoNRE)

 Department of Water Resources Management
 Hydro‐meteorological Services of Viet Nam, Southern Region Hydro‐meteorological Center
for hydrological, meteorological, water quality, and salinity data collection network
management, data collection and dissemination, and flood, salinity and weather
forecasting data.
 Sub National Institute for Agricultural Planning and Production (NIAPP) for flood maps in
the Mekong Delta and soil suitability maps.
 Southern Institute of Water Resources Research (SIWRR), for hydraulic works (sluices,
hydropower, reservoirs, sea, dykes, embankments), river training works, coastal zone
management (river morphology, strategies and work for river bank protection), irrigation,
drainage, soil reclamation, water supply and drain information. The SIWRR also operates
and maintains meteorological and hydrological data, and cross‐section data for the
Mekong Delta and Dong Nai River basin.
2.4.3 Database development and conduct of field visits
The collected information is to be organised in a database consisting of GIS‐related and time

oriented data. In the database a clear distinction is to be made between original/raw data, data
that is being validated, and fully validated data.
During the field visit the data on flood protection and hydraulic infrastructure in the study area
should be checked, including maintenance and operation in times of flood. All key hydrometric
monitoring stations should be visited and monitoring and discharge measurement practices

should be reviewed. Gauge stations and gauge histories should be inspected, including gauge
reach, staff gauge connections, bench marks, gauge zero controls and shifts, gauge damages
and repairs as well as records from comparisons of manual and automatic gauges. Flow
measurement equipment calibration and functioning should be tested. Residents and officials
should be interviewed on the extent of historic floods, flood protection developments and land
use changes. Flood marks should be levelled and discussed with local residents. In connection
with the field visit and where required, additional data can be collected or a (temporary)
monitoring program can be established and implemented.
2.4.4 Data validation and processing
Thorough validation, correction and completion procedures (filling in missing data) should be
applied to improve the reliability of the water discharge data (See also Section 2.3.2). The
process should include three components:
 Screening of historical water levels involves inspection of the historical patterns of
hydrographs, comparison of hydrographs at nearby monitoring stations, stage relation
curve analysis to identify outliers and gauge shifts. Stage relation curves (with appropriate
time shifts to account for flood wave celerity) may be applied to fill in gaps in the data.
 Stages of discharge measurement data should first be compared with the stage
hydrograph. The computation of the discharge from field data is checked and its accuracy
is assessed. Possible backwater and unsteady flow effects are to be identified by means of
a first order backwater calculation and assessment of the Jones correction, respectively.
The establishment of the discharge ratings is to be inspected, with special emphasis on the
extrapolation. The latter should be physically based, taking into account flood protection
works in and downstream of the section. The variability of the discharge ratings at
locations as a result of natural or human induced morphological changes in the river bed
can be assessed, by determining the variation of the water level for distinct high water
discharges through the years. Particularly, this change is assessed for the natural bank full
floodwater discharge.
 The next step is to apply the discharge rating to the water level data series in order to
compute the water discharge data series. A comparison is made with the series in the
database. Discharge hydrographs at neighbouring monitoring stations can be compared
graphically and the consistency of the data at each hydrograph can be checked by means
of water balances and double mass analysis. Comparisons with data at neighbouring
monitoring stations can be used to validate peak water level measurements. Finally, peak
values should be examined, graphically, statistically and using documents on drainage and
storage conditions to spot possible trends indicating problems in the data and other
anomalies. Data should be corrected to eliminate these trends and anomalies.
2.4.5 Hydrological hazard assessment in terms of flood peak water levels and flood volumes
Hydrological hazard is determined by peak flow and flood volume and their interaction. A matrix
of flood hydrographs can be generated covering the full spectrum of possible hydrographs with
peak water level values for return periods of more than or equal to two years (T ≥ 2 years). The
development of this matrix involves the following tasks:
1. Creation of homogeneous water discharge series data for the key measurement station at
the upper boundary of the river reach under study;
2. Estimation of the marginal distributions of annual flood peak water levels and flood
volumes (between June 1st and November 30th) at this upper boundary. The observed
distributions of flood peak water levels and flood volumes are generally represented
effectively by a GEV distribution;

3. Development of a regression equation that describes the relationship between flood

volumes (V) and flood peak water levels (Qp) V = f(Qp) and estimate its standard error SeV, in
a process similar to step 3 in Section 2.3.4;
4. Specification of the matrix of flood hydrographs for distinct discharge peak water levels,
similar to step 3 in Section 2.3.4.
Changes in the stage‐discharge rating due to river bed variations can be taken into account by
changing the hydraulic roughness used in the hydraulic model commensurate with the change
in the water level. For this purpose, one should start by assessing the annual changes in the
water level for natural bank full flow. The effect of roughness can be determined from
comparisons of applied discharge ratings or plots of discharge data series versus water level
data series, (provided that shifts in gauge zeros have been eliminated). The differences are
given relative to the levels from the discharge rating used in the calibration of the hydraulic
model (see Section 2.3.6) and are determined for all key monitoring stations in the modelled
river reach. According to Manning’s equation for a wide cross‐section, if the water depth at
bank full discharge changes from the original depth, h1, to a second depth, h2 ,with h2 = h1 + Δh,
there is a simple equation for the hydraulic roughness, n, over the same cross sections. The new
hydraulic roughness value is described by the relationship:
5/3
h 
n2  n1  2  (2.7)
 h1 
Note that the stage‐discharge relation at the downstream boundary has to be adjusted to avoid
unwanted backwater effects.
2.4.6 Flood hazard assessment
The hydrological hazard in terms of peak flow statistics can be converted into water level data
by means of a hydraulic model of the concerned Mekong River reach. An accurate hydraulic
model is needed for the river and floodplain being analysed. The model’s boundaries should be
chosen in accordance with the hydrometric network equipment and the limits of the area of
analysis. It is strongly advised using a 1D‐2D hydraulic model, where the Mekong River is
represented as 1D and the floodplain as 2D, both readily derived from a DEM. This approach has
clear advantages over a quasi‐2D approach with storage cells, where qualitative judgments on
the connection between river and floodplain introduce subjectivity. The chosen model should
be calibrated for river and floodplain conditions representing the base case. Special attention
should also be given to the position of the river bed (particularly in case of possible river bed
shifts due to natural or human induced morphological changes).
Data from the 30 selected hydrographs can then be input using the model for the particular
river and floodplain system. The water levels that are computed as functions of time and space
can be placed in a database. The procedure can be repeated at least twice with adjusted
hydraulic roughness and downstream stage‐floodwater discharge relations to cover the full
range of water level changes for bank full floodwater discharge. These results can be added to
the database before using a Monte Carlo simulation procedure for flood hazard assessment.
These simulations can be repeated to test the results of construction plans that are under
consideration, after adjustment of the hydraulic model for the different scenarios.
The final step is the execution of a Monte Carlo simulation as described in step 5 of Section
2.3.4.

2.4.7 Example: Mekkong near Na khon Phanom

m (Thailand)
Preparation
ns for Monte
e Carlo analyssis
Figure 2‐9 shows the GEV‐function
G fitted to th
he series of observed
o an nual maximu um water
discharges of the Meko ong River at Nakhon Phaanom, Thailand. This funnction can be e used to
derive the floodwater discharges
d coorresponding to return pe eriods of 2, 55, 10, 25, 50
0 and 100
years (Tablee 2‐5). Subseequently, the relation betw ween peak flows and floood volumes (ttotal flow
volumes fro om June 1st until
u Novembber 30th) can be derived frrom a regresssion analysis (with the
fitted regreession equation shown ass the full line e in Figure 2‐10). The spreead around this
t fitted
regression line
l is also deepicted in Figgure 2‐10 by the
t 4 dotted lines at distannces of ‐1.966 Se, ‐Se, 0,
+Se, +1.96 Se from the regression lin e (with Se the standard deviation from m the regresssion line).
Then, for each
e selected
d peak dischaarge‐flood vo olume in Tab ble 2‐5, the 5 accompanyying flood
volumes caan be specified accordiing to the spread in the flood voolume‐peak discharge
relationshipp. Figure 2‐111 shows the resulting 30 (65) grid points in com bination with h the five
fitted lines from Figure 2‐10.
2 (This reepeats some of
o the processes as shownn in Section 2.3.3.)
For each off the 30 grid points a moddel simulation with ISIS ca an be run to derive waterr levels at
various locaations for a range
r of hydrraulic conditiions. The sim
mulations servve as the bassis for the
Monte Carrlo analysis. For these ssimulations, ISIS requiress 30 hydroggraphs of daily water
discharge as
a input serie es data. Thesse (synthetic)) hydrographs need to haave the same e features
(i.e. peak discharge and flow volumee) as the 30 grid
g points. In n order to obttain realistic synthetic
hydrograph hs, observed hydrographhs can be se elected and (slightly) addapted. The observed
hydrograph hs that most closely resem mble the feattures of the grid
g points shhould be selected. This
selection procedure is depicted
d by tthe green linees in Figure 2‐12).
2 The obbserved hydrrograph is
then “scaled” in such a way
w that its ppeak dischargge and flow volume
v are exxactly equal to that of
the associaated grid poinnt. Figure 2‐ 13 shows an n example of such scalingg. In this exa ample the
peak dischaarge was slightly altered, but the rem mainder of the e hydrographh was slightlyy lowered
to decrease the flow volume. Thee resulting hydrograph is a realisticc model tha at closely
resembles the
t observed hydrograph..
Figure 2‐9 Fit of the GEV function to a nnual maximum

m daily discharg
ge of the Mekoong River at Nakkhon
Phanom.
ood Risk Assessm

MRC Flood
d Management and
P Com
Table 2‐5 Parammeters of the GEV

G distributionn fitted to annuual maximum da aily discharge oof the Mekong
River at Nakhon
N Phanom m and dischargees (m3/s) for selected return periods.
paarameter value
k 0.309
α 4,685
u 24,475
return p
period (years)) arge (m3/s)
discha
2 26,098
5 30,097
10 32,070
25 33,989
50 35,090
100 35,970
Figure 2‐10 Peak water level discharrge‐Flood volum

me relation at Nakhon
N Phanom
m.

station: Nhakon Phanom; selected grid

340
V=f1(Q)+1.96Se
320
V=f1(Q)+1.00Se
300
V=f1(Q)
280
V=f1(Q)-1.00Se
volume [BCM]
260 V=f1(Q)-1.96Se
240
220
200
180
160
140
25 30 35 40 45
3
peak discharge [1000m /s]
Figure 2‐11 Defined 2D grid of discharges (with return periods 2, 5, 10, 25, 50 and 100 years) and volumes
(based on the regression line and lines of deviation from the regression line).
station: Nhakon Phanom; selected grid and observed combinations

350
observed
grid
300
250
volume [BCM]
200
150
100
10 15 20 25 30 35 40
3
peak discharge [1000m /s]
Figure 2‐12 Observed combinations of peak discharge and flow volume (red dots) and selected 2D grid (blue
dots). The green lines show the observed combinations that most closely reflect the grid
combinations.

(scaled) hydrograph at station Nhakon Phanom; year: 1972

observed peak: 25500 [m3/s]; scaled peak: 26098 [m3/s]
observed volume: 172 [BCM]; scaled volume: 165 [BCM]
30
observed
scaled
25
20
discharge [1000m3/s]
15
10
0
14/05 03/06 23/06 13/07 02/08 22/08 11/09 01/10 21/10 10/11 30/11
date
Figure 2‐13 Example of scaling of an observed hydrograph to make it suitable for serving as one of the grid
points in Figure 2‐12.
Monte Carlo analysis and results

For this example, a Monte Carlo analysis was executed with N=100,000 samples. For the
resulting N water levels the probability of exceedance was estimated through application of the
formula (2.5) presented in Section 2.3.4. For a floodplain just downstream of Nakhon Phanom,
the resulting frequency curve for the water level is presented in Figure 2‐14. It starts with a
basic level (measured in meters over Mean Sea Level (MSL) of just over 130 m+MSL. This is the
output of the ISIS‐model if the floodplain location does not flood during the year. The
probability that this happens is approximately 0.5 per year. Naturally, for increasing water levels
the probabilities of exceedance are lower. The once every 100‐year floodwater level is
approximately equal to 134.8 m+MSL (where the graph crosses the line of x = 10‐2 per year).
Similarly, the once every 100‐year flood water level can be determined for other locations.
Based on the results for all locations a spatial map can be produced using the GIS‐procedures,
described in Section 2.7.

Figure 2‐14 Derived frequency curve foor maximum annual water leve
els at a locationn in the Mekongg River
floodplain ju
ust downstream
m of Nakhon Phanom.
2.5 C
Combined flo
ood hazard aassessment
2.5.1 G
General
Combined floods
f are flo
oods that occuur in the dow
wnstream sections of tribuutaries. The flood level
is determin
ned by the co ombination oof tributary water
w flow and the waterr level in thee Mekong
River, backiing up into th
he tributary aand impedingg drainage.
The hydrolo ogical hazard of tributary runoff waterrs with high water
w levels aat the conflue
ence with
the Mekongg River can beb simulated with a hydraaulic model. The levels off river tributa ary runoff
water can beb converted into water leevels in riverr and floodpla
ain in the bacckwater affeccted zone
near the rivver mouth using this modeel. The mode el uses data across
a the ful l spectrum of possible
combinatioons of upstre eam floodwaater inflow and a downstrream water levels. The hydraulic
model shou uld encompass the full arrea including the tributary and its flooodplains starrting from
an area with no backwatter flows andd extending to o the area wh
here the rive rs meet, as well
w as the
area of thee Mekong River around this conflue ence where there
t shouldd be key hyd drometric
monitoring stations. Th he model appplies discharge hydrogra aph data uppstream and a stage‐
discharge reelation downnstream.
The simulation results derived from m this modell can be used in a Montte Carlo proccedure to
derive the exceedance probabilitiess of water leevels and flooodwater dam mages. The procedure
p
usually includes the following three or four rando
om variables, representinng the main causes
c for
high water levels in thee downstreamm part of thee tributary th
hat is backedd up with ma
ainstream
floodwaterss:
ood Risk Assessm

1. maximum water discharge volumes in the Mekong River at the confluence of the Mekong
and a tributary;
2. total volume of the floodwater in the Mekong at the confluence;
3. total volume of the tributary floodwaters; and
4. maximum water discharge volumes in the tributary.
Note that it may not always be necessary to explicitly take the tributary peak flow into account,
particularly when floodplain inundations and associated damages are considered. In such cases
the tributary flood volume, rather than the peak flow plays a more important role. Tributary
peak flow is mainly a concern in the design of flood protection works along the tributary, in
which case tributary flood volumes are not the relevant concern. In practice, generally only one
of the two is considered and rarely both.
Special attention should be paid to variations of the river bed as a result of natural or human
induced morphological changes. These can affect the conveyance of the rivers and hence the
resulting stages. The river conveyance may be an additional variable that is needed in the
Monte Carlo simulation procedure.
Finally, combining the water levels of equal probability and comparing the results with a DEM,
flood maps of levels, depth and duration can be produced.
The flood hazard assessment for combined floods includes the following steps:
1. Data collection;
2. Database development and conduct of field visits;
3. Data validation and processing;
4. Hydrological hazard assessment in terms of flood peak water levels and flood volumes on
the Mekong River and the specific tributary.
5. Flood hazard assessment using a hydraulic model that transforms flood hydrograph data to
flood level data and interpolates from simulated water levels by means of a Monte Carlo
procedure to arrive at flood levels and durations for distinct return periods.
6. Preparation of flood maps (see Section 2.7).
The steps are elaborated in the following sub‐sections.
2.5.2 Data collection
For assessment of the hazard of combined floods, the following data and information are
required:
1. Identification of key hydrological monitoring stations (Qh‐stations) on the Mekong River
tributary extending beyond the backwater reach of the mainstream and the along the
Mekong River in the area of study;
2. Survey data of the tributary and Mekong River (full river bathymetry) and of floodplains in
the study area and beyond, and data of relevant hydraulic parameters, including river bed
material, vegetation, embankment elevation, etc.;
3. Updated Digital Elevation Model (DEM) of the floodplain in the area under study;
4. Soil maps of the floodplains;
5. Land use maps of the floodplain, including past, present and expected future land use;
6. Data on hydraulic infrastructure in and downstream of the study area, including its
dimensions and operation in times of flood and planned future developments;
7. Historical flood maps (including flood levels, extent, depth and duration);
8. Layout, operation and maintenance information for the hydro‐meteorological network
around the study area including information on the nearest upstream (and for the Mekong

River, downstream) monitoring stations along the tributaries and Mekong River where
relevant;
9. Time series data of water levels, stage‐discharge measurements, discharge ratings and
discharges series of the relevant monitoring stations as well as series data of rainfall and
climatic variables in the region, particularly in cases when lateral water inflow downstream
of the key monitoring station on the tributary is substantial;
10. Changes that have occurred in runoff characteristics of basins upstream of the study area
that affect the homogeneity of the historical water discharge data series (including
construction of reservoirs and land use changes).
2.5.3 Database development and conduct of field visits
The same database development strategy as outlined for mainstream floods (Section 2.3.3)
applies here. The same holds for the field visits, with the following addition. When lateral
floodwater inflow is substantial and can only be estimated from rainfall records, it is important
to visit and inspect the rainfall/climatic monitoring stations in and around the contributing area.
To judge the suitability of a monitoring station for use in rainfall‐runoff modelling, the visit
should include inspections and review of the monitoring station layout, equipment exposure,
operating practices and maintenance.
2.5.4 Data validation and processing
Data validation and processing should follow the same steps as presented in Section 2.3.4. If
lateral floodwater inflow measurements need to be derived using rainfall‐runoff modelling
(SWAT), rainfall and climatic data of the area of the lateral inflows needs to be validated and
processed. This involves that the historical series are first screened graphically by comparison
with nearby monitoring station records. Erroneous date entries can be identified by tabular
comparisons. Monitoring practices during weekends and holidays should be inspected and data
should be adjusted where necessary. Next, the data should be compared with data collected by
the nearest neighbouring monitoring stations and also subject to double mass tests. Series data
should be adjusted when necessary and completed where data is missing, using estimates
based on data from reliable measurement stations. The resulting data series can then be
converted to areal rainfall for sub‐basins used in the SWAT rainfall‐runoff model. The data series
can then be compared with the series used earlier for the modelling. Where there are
differences, it may be necessary to recalibrate the SWAT model for the areas where it is to be
applied. Note that in several areas, rainfall‐runoff modelling may not be needed to estimate the
lateral floodwater inflow, such as in the Xe Bang Fai River example discussed in Section 2.4.7.
2.5.5 Hydrological hazard assessment
The hydrological hazard is determined by peak flow and floodwater volume on the Mekong
River. These can be used to estimate the floodwater levels at the confluence, and the tributary
flood volumes and peak flow on the tributary, as well as their interrelation. A matrix of flood
hydrographs can be plotted to cover the full spectrum of possible hydrographs starting with the
peak flow on the Mekong River for floods with return periods of T ≥ 2 years. The development
of this matrix requires the following tasks:
1. Creation of homogeneous floodwater discharge series for the key monitoring stations at the
upstream boundaries of the hydraulic model on the tributary and the Mekong River.
2. Estimation of the marginal distributions of annual flood peak water levels and flood
volumes (between June 1st and November 30th) at these upper boundaries (on the Mekong
River and tributary). The observed distributions of flood peak water levels and flood
volumes are generally well represented by a GEV distribution.

3. Generation of the following regression equations between peak water level flows and flood
volumes with their standard errors:
 Floodwater volume on the Mekong River (VM) as a function flood peak discharge on the
Mekong River (QpM):
VM = f1(QpM);
 Floodwater volume on the tributary (VT) as a function of floodwater volume of the
Mekong River (VM):
VT=f2(VM); and
 Flood peak discharges on the tributary (QpT) as a function of floodwater volume on the
tributary (VT):
QpT=f3(VT);
Note that these equations are not required when tributary peak flows do not play an
important role. If, on the other hand, the tributary flood volume is not critical, it may be
replaced by the tributary flood peak; see Section 2.4.1;
 Plotting of the regression equations with year‐numbers (return periods) to the grid
points and additions of lines, parallel to the regression line at distances of +/‐ 1x Se and
+/‐1.96 x Se.
4. Specification of the matrix of flood hydrographs for distinct flood peak discharge values.
This process requires:
 Selection of flood peak discharge values and volumes (65=30) and associated
hydrographs for the Mekong River in the same manner as in Section 2.4.7;
 Determination of range of tributary flood volumes for each of 30 selected Mekong River
flood volumes from the scatter plot about the regression line for the tributary flood
volume – Mekong River flood volume relationship, generating, using the standard
errors: for each selected Mekong River flood volume a low (R‐1.96 SeVT), medium
(designated as Rf2) and a high tributary flood volume (R+1.96 SeVT). This results into 65
x 3 = 90 combinations of Mekong River flood peak, flood volumes and tributary flood
volumes (producing a 3D‐grid, as shown on the next page); and
 Derivation of 90 synthetic hydrographs, by scaling observed hydrographs in the same
manner as in Section 2.4.7.
The steps in the development of the matrix up to a level of 3 variables are depicted graphically
in Figure 2‐15.
As for mainstream floods, changes in the stage‐discharge ratings (mainstream and/or tributary)
due to river bed variations should be taken into account by repeating the model simulations
with different values for hydraulic roughness. The roughness change will be commensurate with
the maximum observed changes in the water level for bank full discharge relative to the model
values used for the calibration.

Step 1
Selection of 6 peak flows

from annual maximum
discharge distribution GEV
T= 2, 5,10, 25, 50, 100 years
Step 2
T = 100
T = 50
T = 25
s
el
T = 10 l ev
w
flo
ak
Pe
T=5
T=2
medium
very high
very low
Selection of
high
low
Flood Volumes
R = regression equation V=f(Qp)
R-1.96 S e R-Se R R+Se R+1.96 S e
Se = standard error
Step 3
high volume tributary flood
medium volume triburaty flood
low volume tributary flood
For each selected main stream flood select 3 tributary

floods on volume (high, medium and low)
Figure 2‐15 Development of the flood simulation matrix for combined floods.
2.5.6 Flood hazard assessment using a hydraulic model
The hydrological hazard in terms of statistics and relations between:

 flood peak flow and flood volume on the Mekong River, and
 flood volume and/or flood peak flow on the tributary
is transformed into water levels in the rivers and floodplains in a hydraulic model of the
tributary confluence with the Mekong River. The model requirements are specified in Section
2.3.6.
Data from the 30 selected Mekong River hydrographs and 90 tributary floods can then be input
using the model for the particular river and floodplains system. The water levels that are
computed as functions of time and space can be placed in the database. If relevant, in view of
river bed changes, the simulations can be repeated at least twice with different specifications
for hydraulic roughness and downstream stage‐discharge relations to accommodate the
conveyance changes. These should be added to the simulation database before using a Monte
Carlo simulation procedure. Simulations are also necessary to analyse any planned construction.
The final step is the execution of a Monte Carlo simulation, as described in step 5 of Section
2.3.4. The only difference here is that in this case a 3D‐interpolation is required instead of a 2D
interpolation.

MRC Flood
d Management and
P Com
2.5.7 Example: co
ombined floo
ods in the low
wer Xe Bang Fai
F River Focal Area (Lao PPDR)
Combineed flooding taakes place in the floodplaains along the e lower Xe Bang Fai Riverr, where the
water levvels are affeccted by backw
water from t he Mekong River.
R The Xe Bang Fai Rivver drains an
2
area of 110,240 km and discharge es its waters into the Mekkong River att That Phanoom, Thailand
which is located betw ween the key Mekong Riveer monitoringg stations: Na akhon Phano m upstream
and Mukkhdahan, Thaailand downstream. The fflow in the Xe e Bang Fai Riiver is measuured, among
others att Mahaxai an nd Ban Xe Ba ang Fai at Higghway Bridge e 13. (See Fig gure 2‐16.) TThe Mahaxai
2
monitoring station su urveys an are
ea of 4,520 km m , whereas the drainage e area at Ban Xe Bang Fai
amountss to 8,560 km m2. These acccount, respeectively, for 44% and 84 4 % of the eentire basin.
Between n Mahaxai and Ban Xe Ban ng Fai, the Xee Bang Fai Rivver is joined by the Nam Oula and Se
Noy Riveers. The mon nthly averagee flows in thee Xe Bang Faai River at Mahaxai
M and tthe Mekong
River at N
Nakhon Phan nom are prese ented in Tablle 2‐6.
Figure 2‐16 Hydrological map of Low

wer Xe Bang Faii River basin.
Table 2‐6 Averaage monthly an

nd annual flow ((MCM) of the Xe
X Bang Fai Rive
er at Mahaxai aand Mekong
River at Nakhon
N Phanomm.
Jan Feb M
Mar Ap
pr May Jun Jul
Mahaxai 60.0 41.5 337.2 344.7 90.8 565.6 1601.4
Nakhon P
Phanom 6,277 4,469 44,130 4,0
000 6,492
2 17,722 35,993
Aug Sep Oct Noov Dec Year

Mahaxai 2,370.7 1,687.9 6635.9 2400.5 105.8
8 7,504.4
Nakhon P
Phanom 54,457 51,055 299,808 14,8
860 9,111
1 238,376
Floods in
n the Xe Ban ng Fai River may occur f rom June to early Octob ber, coincidinng with high
water levvels on the Mekong
M Riverr. The water levels in the Xe Bang Fai River at Highhway Bridge

13 at Ban Xe Bang Fai are seriously affected by backwater from the Mekong River, while
Mahaxai is backwater free.
Hydrological hazard assessment

The hydrological hazard in the lower Xe Bang Fai River is determined by the floodwater
discharge of the river at Mahaxai and the lateral floodwater inflow further downstream as well
as by the water levels on the Mekong River at the confluence near That Phanom. The following
data is available to describe the hazard:
 The Xe Bang Fai River floodwater discharge records for Mahaxai for the period 1988‐2006.
 River flow measurements at Ban Xe Bang Fai. The Ban Xe Bang Fai data establishes the
relationship between tributary water inflow along the river and at a specific point, given
the backwater effects on the stage‐discharge relation at Ban Xe Bang Fai. Tributary water
inflow between Mahaxai and Ban Xe Bang Fai can then be estimated as a function of the
flow at Mahaxai, without having to rely on questionable inflow estimates from a very
limited number of rainfall monitoring stations.
 The Mekong River floodwater discharge at Nakhon Phanom and the flow from the Xe Bang
Fai River. The water levels at That Phanom can be determined by adding these two data
sources and considering the conveyance capacity downstream of the confluence,
represented by the stage‐discharge relation of Mukhdahan. The Nakhon Phanom
floodwater discharge series data cover the period 1924‐2005.
The discharge series data have been thoroughly validated using similar approaches to those
explained in Section 2.3.4. Where necessary, the data has been adjusted. (See the Xe Bang Fai
River Flood Hazard Assessment (FHA) Report for details.) Using this data, the hydrological
hazard in terms of marginal distributions of peak water flows and flood volumes and their
interrelations have been quantified. Statistics for Nakhon Phanom are presented in Section
2.4.7. Statistics for Mahaxai are presented in Table 2‐7, Figure 2‐19 and Figure 2‐20.
The matrix of flood hydrographs for distinct discharge peak water levels can then be specified,
starting with the flood peak water levels at Nakhon Phanom. This is described above in Section
2.4.7. The result consists of 30 (65) combinations of peak water level discharge and flow
volumes and 30 associated water discharge hydrographs. For each of these 30 Mekong River
flood volumes, 3 Xe Bang Fai River flood volumes can be selected at distances ‐1.96 SeVXBF,
0,+1.96 SeVXBF from the Xe Bang Fai River flood volume‐Mekong River flood volume regression
equation using the standard error measure, SeVXBF. (See Figure 2‐20.) The result is a 3‐
dimensional grid for peak discharge and volume in the Mekong River (at Nakhon Phanom) and
flood volume in the Xe Bang Fai River (at Mahaxai) as depicted in Figure 2‐17. The grid contains
90 points (653), meaning 90 discharge hydrographs for Mahaxai are required for the Monte
Carlo analysis. This can be achieved through rescaling of observed hydrographs, as described in
Section 2.4.7. Figure 2‐18 shows an example of a scaled hydrograph.

Table 2‐7 GEV‐parameters, flood peak discharges and flood volumes (June‐November) for distinct
return periods in the Xe Bang Fai River at Mahaxai.
Parameter Flood peak Flood Volume

discharge (MCM)
(m3/s)
k 0.341 0.221
α 498 2,304
u 1,614 6,105
T (years)
2 1,757 6,916
5 2,177 9,045
10 2,398 10,188
25 2,626 11,386
50 2,765 12,126
100 2,881 12,755
3D grid of volumes and peak discharges
14
12
Volume Se Bang Fai [BCM]
10
2
350
300 36
34
250 32
30
200
28
Volume Mekong [BCM]

peak discharge Mekong [1000m3/s]
Figure 2‐17 Impression of the 3‐D grid for peak discharge and flood volume in the Mekong River (at Nakhon
Phanom) and flood volume in the Xe Bang Fai River (at Mahaxai). The blue dots only show the
‘outer side’ of the grid; the full grid contains 90 such dots.

(scaled) hydrograph at station Mahaxai; year: 1988

observed peak: 1440 [m3/s]; scaled peak: 1255 [m3/s]
observed volume: 4.0 [BCM]; scaled volume: 3.5 [BCM]
1500
observed
scaled
1000
discharge [m3/s]
500
0
14/05 03/06 23/06 13/07 02/08 22/08 11/09 01/10 21/10 10/11 30/11
date
Figure 2‐18 Scaling of an observed hydrograph for the Xe Bang Fai River at Mahaxai.

GEV-fit to annual maximum discharge in Se Bang Fai at Mahaxai

3,500
Annual maximum discharge

Annual maximum discharge at Mahaxai (m /s)
3,000
GEV-distribution
3
2,500
2,000
1,500
1,000
500
-2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00
EV1 reduced variate
GEV-fit to annual flood volume in Se Bang Fai at Mahaxai

14,000
Annual Flood Volume

12,000
GEV-distribution
Annual flood volume at Mahaxai (MCM)
10,000
8,000
6,000
4,000
2,000
0
-2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00
EV1 reduced variate
Figure 2‐19 GEV‐fit to marginal distributions of annual maximum discharge and flood volume at Mahaxai
(Xe Bang Fai River).

Correlation of flood volume Se Bang Fai (Mahaxai) and Mekong (Nakhon Phanom)
14,000
12,000
Fvol,Mahaxai = 0.0335Fvol,NPhanom + 52.8
2
R = 0.58
Flood volume Mahaxai (MCM)
10,000
8,000
6,000
4,000
2,000
0
50,000 100,000 150,000 200,000 250,000 300,000 350,000
Flood volume Nakhon Phanom (MCM)
Correlation of flood volume Se Bang Fai (Mahaxai) and Mekong (Nakhon Phanom)
14,000
VM+1.96 Se
12,000
Fvol,Mahaxai = 0.0335Fvol,NPhanom + 52.8
2
R = 0.58
VM=f(VNP)
Flood volume Mahaxai (MCM)
10,000
8,000
VM-1.96 Se
6,000
4,000
2,000
0
50,000 100,000 150,000 200,000 250,000 300,000 350,000
Flood volume Nakhon Phanom (MCM)
Figure 2‐20 Flood volume – peak discharge relation for Nakhon Phanom and flood volumes relation between
Mahaxai and Nakhon Phanom, without and with Se‐related lines.
Flood hazard assessment

The 90 flood hydrographs of the Mekong River and Xe Bang Fai River selected above can be
input to the hydraulic model of the Xe Bang Fai River delta model for conversion of water flow
data into water level data. The model layout and boundary conditions are schematically
displayed in Figure 2‐21.

Nakhon Phanom Q
Mekong Mahaxai
Q35 i
Fa Q81
ng
Ba Q38
Se
That
Phanom
Q
Se Noy
Mukhdahan
QH
Figure 2‐21 Xe Bang Fai River hydraulic model with boundary conditions.
The morphological development of the river bed, particularly of the Mekong, is taken into
consideration by adjusting the hydraulic roughness of the river in accordance with the level
changes for bank full discharge relative to the calibrated situation. The equation that can be
used is Manning’s equation, as described in Section 2.4.5 and is slightly rewritten here as:
h + ∆h
Nnew = ncal (‐‐‐‐‐‐‐‐‐‐‐‐‐)5/3 (2.8)
h
where: h = water depth at Qbankfull [measured in meters, m]
∆h = change [m]
In addition, the Mukhdahan rating should be adjusted to eliminate unwanted backwater effects
on the floodwater level data for That Phanom. The hydraulic model can be run with the same
hydraulic boundaries while adding the adjusted Manning‐roughness values.
Monte Carlo sampling of the model boundaries interpolation can be run for computed water
levels for each grid cell in the hydraulic model to obtain a frequency distribution for the water
levels at each cell. Figure 2‐22 shows an example of the frequency distribution at a location in
the floodplain of the Xe Bang Fai River, derived using the Monte Carlo random sampling
method.

Figure 2‐22 Derived frequency curve foor maximum annual water leve
els at a locationn in the floodpla
ain of the
Xe Bang Fai River.
R
Flood hazarrd mapping
The proceddure for floodd hazard mappping is desccribed in Section 2.7. As an example the flood
depth and extent map of the Loweer Xe Bang Faai River is sh hown in Figu re 2‐23 , forr a return
period of T = 100 years.
ood Risk Assessm

MRC Flood
d Management and
P Com
Figure 2‐23 Flood dep

pth and extent map of the Low
wer Xe Bang Fai River, for a retturn period of T = 100 years.

2.6 Flood Hazard Assessment in the Mekong Delta
2.6.1 General
The floods in the Cambodian floodplains and Mekong Delta (see Figure 2‐24) are classified as a
special type in the LMB due to their specific external and internal boundary conditions and the
delta’s unique hydraulic infrastructure. Whereas the flood levels near Kratie are mainly
dependent on the Mekong River floodwaters and a small amount of backwater from the Tonle
Sap Lake, additional factors play a role further downstream. In the Mekong Delta’s downstream
areas, flood levels are dependent on upstream and lateral inflows, local rainfall and
downstream water levels where the river empties into the sea. Floodwaters in the Mekong
Delta are conveyed via the Mekong and Bassac Rivers and their floodplains, including the
colmatage canal system, which diverts and regulates the flow of floodwaters from and into the
River. In the Mekong Delta the river regime is modified by seasonal water storage in the Tonle
Sap Lake and in the Mekong River floodplains, creating slowly rising and falling water levels.
Essentially, the hydrological hazard in the delta is the result of upstream and lateral inflows,
local rainfall and downstream water levels at sea. The flood hazard in the Mekong Delta is
determined by the frequency of flood water levels and flow velocities in the delta. These can be
derived from the hydrodynamic delta model based on the ISIS‐software system. This model
covers the Mekong Delta downstream of Kratie to the Gulf of Thailand and East Sea. It has been
developed to compute water levels, flow velocities and river discharges in the delta as a
function of the hydrological hazard components.
Assessment of the flood hazard in the Mekong Delta downstream of Kratie benefits from the
existence of a relatively long historical water discharge data series for Stung Treng, just
upstream of Kratie. Furthermore, for the tributary inflow further downstream and to the Tonle
Sap multi‐year data series have been created preserving the serial and cross‐correlation with
the Mekong River flow. The Stung Treng time series data, which cover the period 1910‐2006,
have been used as boundary conditions for a hydrodynamic model (based on the ISIS‐modelling
package) to derive a 97‐year series of water levels in flood‐prone areas. The ISIS model also
adds local rainfall, evaporation, water use and tidal conditions at the Gulf of Thailand and the
East Sea for the year 2000. Probabilities of flooding and related damages for return periods
from 2 to 100 years can be derived directly from the series of water levels and depths computed
by the model.
The steps required for the flood hazard assessment are similar to those in other hazard
assessments and include data collection and conduct of field visits, database development,
hydrodynamic model development, and then flood hazard assessment followed by the
preparation of flood maps.
2.6.2 Data collection and conduct of field visits
The hydraulic model is based on conditions in the Mekong Delta in the year 2000. To assess the
flood hazard for development scenarios relative to the base case, it is necessary that the
hydraulic infrastructure pertinent to the scenarios is reflected in the model schematization with
different potential constructions and for those that have occurred since 2000. Hence, detailed
information is required on layout, dimensions, capacities and operation of the hydraulic
infrastructure and flood protection works relative to the model conditions. Field visits can be
used to compare the currently existing infrastructure with its schematisation in the model. The
model needs to be updated to reflect key changes.

MRC Flood
d Management and
P Com
As for th
he hydrologiccal boundary conditions, tthe 97‐year time
t series data
d of upstre
ream, lateral
and dow wnstream boundary conditions are a realistic rep presentation of the variabbility of the
hydrologgic conditionss in and aroun
nd the Mekonng Delta.
Figure 2‐24 Hydraulicc infrastructure of the Mekongg Delta.

Figure 2‐25 The schemattisation of the M Mekong Delta using

u the hydro
odynamic delta model (ISIS) involves a
large numbe er of hydraulic nnodes downstre
eam of Kratie, next
n to floodplaain ‘reservoirs and
a river
cross‐section
ns.
2.6.3 D
Database deve
elopment
A database has been de eveloped for flood hazard assessment in the Meko ng Delta thatt includes
the followin
ng types of in
ncluded databbases and datta:
1. A modeel database, including schhematisation n of the hydraulic infrasttructure, its structure
layout and operation ns, and hydroological bounddary conditio
ons.
2. A result database, in ncluding the ssimulated waater levels annd flow veloccities of the base
b case
and poteential infrastructure consttruction scennarios as inputs for flood hhazard assesssments.
3. A digitall elevation model (DEM) oof the Mekon ng Delta.
4. Data forr water levelss, flood depthh and flood duration for distinct floodinng return perriods.
5. Flood maps.
m
ood Risk Assessm

2.6.4 Hydrodynamic model development
The hydrodynamic model of the Mekong Delta is based on the ISIS modelling system for the
simulation of unsteady water flows in river channel networks. It provides an implicit numerical
solver for the de Saint Venant equations for 1D flow. The model computes water levels and
floodwater discharges at select intervals on a non‐staggered grid. Floodplains are included as
storage cells connected to the stream channels via two‐way weirs. The system was introduced
to the MRC under the Water Utilisation Programme (WUP‐A) and now serves as part of the
Decision Support Framework (DSF). The model covers the MRB from Kratie, Cambodia to the
East Sea, including the Tonle Sap Lake and its floodplain, the Cambodian floodplains, and the
Vietnamese sections of the Mekong Delta. The model used in FMMP‐C2 is based on Delta
Model Version‐2009, with adaptations and recalibrations of the original model.
The schematisations of the Cambodian and Vietnamese sections of the Mekong Delta that are
used in the model are presented in Figure 2‐25.
The boundary conditions of the model are based derived from the following data, with each
category then described in turn, below:
 upstream discharge at Kratie;
 tributary inflows to the Tonle Sap and Tonle Sap Lake;
 tributary inflows to the Mekong River d/s of Kratie and the Tonle Sap;
 rainfall;
 evaporation;
 water use; and
 downstream boundary characteristics at the Gulf of Thailand and East Sea.
Upstream discharge at Kratie

A 97‐year discharge series has been established for Kratie, based on the record of Stung Treng
which is available since 1910. For several reasons preference has been given to the series of
Stung Treng, including length of record, absence of gauge shifts and backwater effects, stable
and predictable hydraulic control for high flows and availability of discharge records. The
discharge measurements of Stung Treng, transferred to Kratie reveals excellent agreement with
the current meter measurements at the latter location in the past. A serious problem and
source of confusion, however, is the difference between stage‐discharge ratings according to
conventional discharge measurements and those based on Acoustic Doppler Current Profiler
(ADCP) records in use at Kratie since 2002. The ADCP records, due to erroneous referencing of
the measurements, lead to lower discharges for the same water levels than the ratings based on
conventional discharge measurements, the differences increase for the higher stages.
Therefore, unless concurrent current meter and ADCP‐measurements are taken at Kratie and
Stung Treng, and discharge rating curves are updated, it is advised using only the Stung Treng
records, based on conventional measurements. The Stung Treng records are consistent with
measurements of the water flow at Pakse on the Mekong River. Since the hydrodynamic model
has been calibrated on recent discharge series at Kratie with flood volumes that are about 8%
less than those at Stung Treng, the historical Stung Treng data used in the model can be
corrected to reflect these differences, using a reduction of 8%. (Note that this 8% reduction is
not applied to the data and statistics presented below for Stung Treng, itself.)
The annual maximum discharge series and flood volume of the Mekong River at Stung Treng are
shown in Figure 2‐26 and Figure 2‐27.

Annual maximum discharge in the Mekong at Stung Treng, period 1910-2006

90,000
80,000
70,000
Peak discharge (m3/s)
60,000
50,000
40,000
30,000
20,000
1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Figure 2‐26 Annual maximum water discharges at Stung Treng, 1910‐2006.
Annual flood volume in the Mekong at Stung Treng, period 1910-2006

500,000
450,000
400,000
350,000
Flood volume (MCM)
300,000
250,000
200,000
150,000
100,000
50,000
0
1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Figure 2‐27 Annual flood volumes at Stung Treng, period 1910‐2006.
Tributary inflow to the Tonle Sap Lake

The area draining into the Tonle Sap Lake, bounded by Cambodia’s National Highways 5 and 6
that surround the lake, encompasses 68,830 km2 and includes the drainage areas of 13 Stungs
(rivers). Water inflow data series of daily discharges are available for the years 1997‐2004. A
model has been developed to extend the time series data to 1910‐2006, using contemporary
data to determine the relationship between flows at Stung Treng and then applying data from
Stung Treng to generate the missing data at other sites. The estimates rely on a multiple

regression equation for monthly tributary flows. The flows have been regressed on the tributary
inflows of the previous of month to reflect the serial correlation of the sites, and on the Mekong
River flow at Stung Treng in the same month to preserve the cross‐correlation. Following
standard statistical practices, a normally distributed random number was added to preserve the
random variance that would normally exist between the two data sets. The monthly flows were
then disaggregated to daily flow values based on their degree of agreement with data from the
years in which there were existing measurements. These daily flow measurements from years
with actual data were then scaled per month to the required generated value. The resulting
seasonal inflow to the Tonle Sap Lake is shown in Figure 2‐28. The average monthly inflows as a
percentage of the annual totals in comparison to the Mekong River discharges at Stung Treng
are presented graphically in Figure 2‐29.
Seasonal inflow (May to December) to Tonle Sap Lake

45000
40000
Seasonal inflow (May-December) (MCM)
35000
30000
25000
20000
15000
10000
5000
0
1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Figure 2‐28 Seasonal hydraulic inflow to the Tonle Sap Lake, estimated from 1910‐2006, with recorded data
for 1997‐2004.
Comparison of runoff regime of Mekong at Stung Treng and Tonle Sap Lake inflow
30
Monthly flow volume as percentage of annual total (%)
Mekong at Stung Treng

25
Tonle Sap Lake inflow
20
15
10
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 2‐29 Average monthly flow regime of the Mekong River at Stung Treng and of the Tonle Sap Lake as
percentage of the annual totals.

Tributary inflows to the Mekong River

Apart from the inflows to the Tonle Sap Lake, the Delta Model also requires inflow series data
for the Prek Te, Prek Chhlong, Prek Thnot, East Vaico River, and West Vaico River tributaries.
The daily water flow series for the period 1985‐2006 is available from the DSF files created by
the SWAT model. Since the SWAT‐series data show no correlation with the flows in the Mekong
River, a pragmatic approach was used to extend the data series; applying a repetition of the
time series 1985‐2006 also for the years 1910‐1984 to create a 97‐year series.
Rainfall
Daily rainfall series of 9 locations in Cambodia and of 5 locations in Viet Nam are required as
inputs to the hydraulic model for the Mekong Delta. Data is available for these locations for
1985‐2006. The annual maximum daily values generally are in the order of 100 to 150 mm,
occasionally with larger values up to 400 mm, such as in Can Tho in 1985. Analysis of this data
shows weak or no correlation between that seasonal rainfall at the selected locations with
water inflows to the Tonle Sap Lake, and no correlation at all with the flows in the Mekong
River. As with tributary inflows, a repetition of the series 1985‐2006 was applied to create a
data set covering the years 1910‐1984.
Evaporation
The model requires evaporation data from the same locations as the rainfall data. For the
locations in Viet Nam, a daily data series is available from 1985with the exception of 2002. The
data series available for Cambodia are shorter. In view of the limited variability of potential
evaporation in a particular month from year to year, long term monthly averages have been
applied where daily evaporation data was unavailable.
Water use
There are 128 different sites included in the Mekong Delta model where water is used for
agriculture, domestic and industrial purposes. The variation in water use varies from about
1,400 m3/s in January down to almost 0 m3/s at the end of September. The total annual water
usage amounts to 16.5 Billion Cubic Meters (BCM). During the flood season, the demand for
water is about 4.4 BCM in total, i.e., less than 300 m3/s.
Sea boundary characteristics

There are 19 different water level boundaries defined in the Delta Model. These boundaries are
taken from hourly observations made at the 6 monitoring stations listed in Table 2‐8. The hourly
observations used in the Delta Model are records taken from the year 2000. A comparison with
historical records demonstrates that the use of this series does not introduce a bias in peak
water levels.
Table 2‐8 Overview of water level monitoring stations at coastal boundaries in Viet Nam.
Monitoring
River Province Remark
station
Rach Gia Cai Lon Kien Giang Waters drain into the Gulf of Thailand
Song Doc Song Ong Doc Ca Mau As above
Ganh Hao Ganh Hao Ca Mau/Bac Lieu Waters drain into the East Sea
My Thanh Bassac Soc Trang/Tra Vinh Waters drain into the East Sea
Ben Trai Cua Cung Hau Tra Vinh/Ben Tre Southern Mekong River outlet, with waters
draining into the East Sea
Vam Kinh Cua Dai Ben Tre/Tien Giang Northern Mekong River outlet, with waters
draining into the East Sea

MRC Flood
d Management and
P Com
2.6.5 Flood hazarrd assessmen

nt
The ISIS model is usedu to simu ulate data sseries througghout the MekongM Deltaa based on
regressio
on equations and the use of the 97 yeaar water discharge data se eries at Stungg Treng. The
main outtputs of the model
m are esttimated wateer levels at alll river sections and floodpplains in the
model arrea. For each h location, thhe series of 997 annual maximum water levels cann be derived
using thee model outp puts. The esttimates of th e probabilitiees of exceedance of the w water levels
can then
n be generateed using the Gringortens
G fformula (see equation (2.5 5) in Section 2.3.4). Since
the data series of annnual flood maxima coverss close to 100 0 years, the estimated
e 1000‐year cycle
of water levels is, by definition, approximatelyy the same ass the maximu um observedd water level
over the 97 year periood of actual data.
d
2.7 Flood hazaard mapping
2.7.1 General
The MRC C uses the ISSIS hydraulic model (and sometimes the t VRSAP model) to calcculate water
levels (+MSL) in partss of the Mekkong River Baasin (see pre evious section
ns). The wateer levels are
calculateed for (amongg others) ISIS river cross‐ssection nodess and reservooir nodes. A ccross‐section
is an imaaginary line perpendicular
p r to the riverr in an area where
w the river can flow m more or less
unobstru ucted. A reseervoir node represents a ‘reservoir’ orr ‘cell’ (polyggon), which iss an area of
the river (mostly emb banked) with an even watter level. A GIIS can be used to identify areas of the
river wheere water levvels are evenn, through a pprocess of interpolation ofo the node w water levels.
(See Figuure 2‐30.) Aftter subtractin
ng the grounnd levels (+M MSL, stored inn a Digital Teerrain Model
(DTM) o or Digital Elevation Mode el (DEM) thee water depths, i.e. the flooding levvels, can be
determin ned. (See Figgure 2‐31 and d Figure 2‐322.) Flood maps can also be producedd for various
flooding frequencies (return perriods or proobabilities), for f annual‐m maximum levvels and for
seasonal maximum levelsl up to the first of August. Floo od damage anda flood rissk maps for
ure, housing and
agricultu a infrastruccture are bassed on these water depth maps and onn economic/
damage data.
Figure 2‐30 Water levvel map, based on Figure 2‐31 Water depth h map, generateed from a
interpolattion of ISIS nod
des. water level map
m subtractingg the DTM
(ground) levels.

Figure 2‐32 Depiction of how water or fflood depth is obtained

o by sub
btracting the grround level (DT
TM) from
the water levvel (hydraulic s imulation + GISS interpolation).
The map layers in a GIS need all to hhave the samme datum and d projection tto be able to
o combine
them. The standard dattum at the M MRC for the whole
w MRB is Indian 19554, while the standard
MRC projecction is UTM (northern zo nes). As the data
d sets from
m the hydrauulic model use e another
datum, theyy need to be converted too Indian 54.
The mappinng of the hydraulic modellling results caan be divided d in three stepps:
A: Transferr of modellingg results (watter levels) intto a ‘geo‐refe
erenceable’ taable;
B: Creation
n of spatial da
ata; and
n of flooding maps.
C: Creation
The three main

m steps (see
( Sectionss 2.7.2, 2.7.3
3 and 2.7.4) and correspoonding data sets and
tasks are scchematised and
a numbereed in the flow chart in Appendix 5. TThe descriptio on below
follows the
e numbering in the flow chhart.
2.7.2 Transfer of mo
odelling resullts (A)
The transfeer of modelliing results (w

water levels) into a ‘geo‐referenceabble’ table invvolves the
following stteps:
A‐1, A‐2:
Thee hydraulic model,
m ISIS, ccan export th
he (unique) node identifieer ‘ID’s’ (nam
mes), their
x/yy‐coordinatess, and the callculated wate er level per node,
n into a comma‐sepa arated file
(CSSV).
A‐3, A‐4, A‐‐5:
Somme of the ISSIS output ccan be proce essed with MatLab
M softw
ware, which provides
suitable tools tot determinee e.g. the maximum
m watter levels asssociated witth certain
return periods. Otherwise, t he data can be b presented d on a spreaddsheet and transferred
to a database, like Microsoftt Corporation n’s Access.
A‐6, A‐7:
In Access,
A separate data setts can easily be combined into one taable by usingg Queries
(baased on node ID’s). Placingg all of the daata in one tab
ble reduces tthe work to be
b done in
thee GIS, explained in the foll owing steps.
ood Risk Assessm

MRC Flood
d Management and
P Com
Figure 2‐33 Example ofo ISIS data impported into Acccess, with node information an nd water levels for different
scenarioss, various return
n periods (probbabilities), and different
d period
ds of the year.
2.7.3 Creation of spatial data (B)
ISIS calcu
ulations are mainly
m basedd on node annd water leve el informationn in a data‐fiile (DAT‐file,
ASCII forrmat). The crooss‐section nodes
n have x//y‐coordinatees specified. The reservoirr nodes also
have theeir central (po
oint) coordina
ates in the DA
AT‐file, but no
ot their outlin
ne coordinatees.
ISIS has also a GIS‐ccomponent containing

c thhe river crosss‐sections (lines) and thhe reservoir
T cross‐secction and cenntral reservoir node coord
outlines (polygons). The dinates in thee GIS should
correspoond with thee coordinatess in the DATT‐file, but th
he GIS files are
a not alwaays updated
simultaneously with the
t DAT‐file update.
u
Due to th he lack of coo

ordination with the DAT‐ffile, the GIS‐ffiles from ISIS
S need to be uused for the
reservoirr outlines. Foor the cross‐ssections, eithher the DAT‐‐file or the (n
not always u pdated) ISIS
GIS‐files can be used. Extracting thhe relevant crross‐section information from
f a DAT‐fiile for use in
a GIS is time‐consumiing (the DAT‐‐file for the LM MB contains more than 17 70,000 lines)..
The methodology desscribed below w uses the ISSIS GIS‐data and a not the DAT‐file.
D It shhows how to
h the inaccurracies in the GIS‐files. If t he ISIS GIS‐files have bee
deal with en synchronissed with the
DAT‐file, the methodoology is essenntially the sa me, but some steps can be b skipped (e .g. steps B‐9
to B‐12; ssee Appendix 6), depending on the infformation supplied along with the GIS files.
B‐1: TThe data sto ored in the Access

A table can be impo orted into ArrcGIS and geooreferenced
uusing the x/yy‐coordinates.
B‐2, B‐3, B‐4:
IISIS calculatees with (amonng other dataa) reservoir nodes
n and (rivver) cross‐secction nodes.
((See Figure 2‐34.)
2 For thee creation off maps, thesee two types of nodes ne ed different
pprocessing. ToT accommod date for this, the Access data‐table can
n be split in twwo with one
ttable contain ning the reseervoir nodes, and anotherr containing the cross‐secction nodes.
TThese two taables may be split again foor examining different are eas inside thee basin (e.g.,
ffocal area in Cambodia an nd focal area in Viet Nam).
B‐5, B‐6, B‐7, B‐8:
A spatial (GIIS) compone ent of ISIS coontains cross‐sections (ppolylines) andd reservoirs
((polygons). However,
H the
e relationshipp between the
t calculatio on nodes andd the lines/
ppolygons is not
n always one‐to‐one (inn case the GIS is not upd dated). The ccross‐section
llayer and resservoir layer can be addeed to ArcGISS. Duplicate lines or polyggons can be
rremoved, if necessary.
n

Figure 2‐34 ISIS schemattization. ISIS callculates with (among other data) reservoir noodes (+) and crooss‐section
nodes (dots)). A spatial com ponent of ISIS contains cross‐sections (blue llines) and reserrvoirs
(purple polyggons).
Creating a rivver cross‐section point data seet for interpolattion
To create a data set of water levels or water dep pths that covvers every loccation within
n the area
under studdy, the node e informationn from ISIS needs to be e prepared ffor interpola ation and
conversion to a ‘raster’ using
u both thhe nodes of th
he reservoirs and river crooss‐sections.
The cross‐ssection polylines each reepresent a horizontal

h waater level. If the GIS‐filess are not
updated an nd do not con
ntain the nodde names, on nly part of the polylines ccan be associated with
cross‐sectio
on nodes for which ISIS w
water levels have been calculated. In suuch case the following
steps need to be conduccted for the ccross‐sectionss:
B‐9: In ArcGIS,
A a bufffer of 150 m should be drawn around the cross‐secction polyline es.
B‐10: If only
o one ISIS cross‐sectionn node falls inside a buffe er, the (attribbute) data of the node
shoould be copie ed to the bufffer. If more than
t one node falls inside the buffer, the
t buffer
shoould be deleted.
B‐12, B‐13:
Thee polylines innside the (remmaining) buffers should get
g the samee data as the buffer (=
datta of the crosss‐section nodde) attached.
B‐15: Thee polylines with
w attached data should then be con nverted into a series of po oints (e.g.
15 per polyline)) representinng the line, suuch that all points have thhe data of the original
ISISS cross‐sectio
on node. The layer with thhe buffers cann then be rem moved (delete ed).
ood Risk Assessm

 Creating a reservoir point data set for interpolation
Like the cross‐sections, reservoirs represent horizontal water levels. Two of the same steps
apply here as for the cross‐sections.
B‐11, B‐14:
The reservoir polygons with only one ISIS reservoir node inside use the data of that
node. If more than one node falls inside, the reservoir polygon should be deleted (in
this case, the GIS‐files have apparently not been ‘synchronised’ with the DAT‐file). The
resulting layer contains reservoir polygons with the information of one reservoir nodes
attached.
There are two options to obtain the required water levels: 1) the reservoir information is
merged with the cross‐section information for joint interpolation to obtain the associated water
level (in step C‐2), or 2) only the cross‐sections are used for interpolation. The choice depends
on the number of available cross‐section points in the area, the flood mitigation scenario
considered, the flood type in the area, and the height of the reservoir embankments in the area.
Steps B‐16 and B‐17 can be skipped if option 2 is selected.
B‐16: Each polygon should be converted into a series of points on the polygon outline (e.g.,
15 points are sufficient). These nodes have the same data as the polygon and also the
original ISIS reservoir node.
 Merging the cross‐section point data with the reservoir point data for interpolation
B‐17: All original cross‐section nodes (B‐3), the derived cross‐section points (from the
polylines; B‐15), all original reservoir nodes (B‐4), and the derived reservoir points (from
the polygons; B‐16) should be put into one layer (C‐1). This layer (which may be split
into, for example, a Cambodian and a Vietnamese layer) should be the basis for water
level point interpolation (See Section 2.7.4).
2.7.4 Creation of flooding maps (C)
Water level (flooding) maps can be created in two steps: 1) the point data, with water levels (B‐
15 or B‐17, depending on the chosen option), can be interpolated to create a raster that will
cover the area(s) under study, and 2) the reservoir polygons can be converted into a raster and
used to replace the raster resulting from step 1.
There are many methods to create a continuous surface based on spatial points with values.
The type of data, the spatial distribution of the points, the range in the values etc. determine
the appropriate method. A few methods have been considered, including: Inverse Distance
Weighting (IDW), Spline, Kriging, and Natural Neighbours (NN). Although no in‐depth analysis
has been done, the Natural Neighbours (NN) method and the Inverse Distance Weighting (IDW)
method appear to be suitable for the LMB. Both the NN and IDW methods allow for the option
of including ‘barriers’ or ‘linear discontinuities, such as embankments or canals in the
interpolation. Details on the interpolation methods can be found on the Internet.
Flooding maps can then be created by following the steps presented in Appendix 5.
C‐1, C‐2, C‐3:

The node/point water level values (e.g. year‐maximum water levels with a return
period of 10 years) may be interpolated with the NN method (the raster size is 50 m as

in the
t DTM), resulting in a rrectangular raaster covering the area w with the pointts used as
inpputs.
C‐4, C‐5, C‐66:
A reservoir
r rastter file can bbe created byy converting the polygonss to rasters, using the
reqquired water level values ffrom the polyygon attribute table as rasster cell value es.
C‐7: Thee polygon‐bassed raster (rreservoirs; C‐‐6) can replace2 the poinnt‐based raster (cross‐
secttion nodes, re eservoir nodees; C‐3) to assure that the e reservoir areeas do, indeeed, have a
horizontal waterr level.
C‐8, C‐9, C‐10:
Thee raster can then
t be ‘cut oout’ more precisely using a “mask” (poolygon) with the exact
dellineation of the
t area of innterest (or ussing several masks
m to deleete various areas
a that
aree under studyy). The resultt is a raster of
o the area(s)) under studyy with each raster
r cell
(500 m x 50 m) having
h a wateer level value
e. The raster values (e.g., year‐maximum water
levvels with a retturn period oof 10 years) caan then be classified usingg suitable class limits.
C‐11, C‐12, C‐13:
If water
w depth maps are reequired, the DTM with the t measuree of ground elevation
shoould be subtrracted from t he water leve el raster, and classified.
Figure 2‐35 Example of a raster, with thhe reservoir rasster (angular are
eas) superimpoosed on the point based
raster.
us steps resullt in a flood (w

The previou water) depth
h map; an exa
ample is depiccted in Figure
e 2‐36.
2
Using the ArrcGIS Map Calculator with thee ‘con’ statemen
nt (e.g. con(isnu
ull([reserv]),[xseec],[reserv]))
ood Risk Assessm

MRC Flood
d Management and
P Com
Figure 2‐36 Example of

o a flood depth map, here in the Mekong De
elta for a flood with a return pperiod of 25
years.
2.7.5 Data sourcees
nerated by the ISIS model (with x/y‐cooordinates) inccludes:

Data gen
a) Tho ose polygons that represennt reservoirs (i.e., areas with
w horizonta al water levell);
b) Tho ose polylines that
t represen
nt river cross ‐sections (i.e., lines with horizontal
h waater level);

c) Model calculation nodes for the reservoirs and cross‐sections. These nodes (from the DAT‐
file) may not coincide one‐to‐one with the polygons and polylines mentioned above if the
ISIS GIS‐files are not updated. (See Section 2.7.3); and
d) Calculated maximum water levels (referenced to MSL) for various return periods, for
various periods of the year (e.g., ‘early floods’), for various areas (‘focal areas’), and/or for
various scenarios (flood risk reduction measures).
Data from the MRC/FMMP GIS includes only one category:

e) DTM or DEM data. In 2009 two DTM’s were being used at the MRC; b‐dtm50upd and
dem0603. Both are based on interpolation of 10 m contours (on some map sheets, 20 m
contours), spot heights and rivers/streams (using topographical maps with ratios of
1:50,000 and 1:100,000, depending on the area). b‐dtm50upd is a 50 x 50 m raster, while
dem0603 is a 100 x 100 m raster. Both have significant areas with errors. In b‐dtm50upd,
the areas with errors are two large areas near Kampong Chhnang and Kampong Cham. In
dem0603 many smaller areas across the LMB have errors. As dem0603 is used in ISIS and
its erroneous areas are smaller, the MRC prefers use of this DEM.

CHA
APTEER 3
FLOO
OD DA
AMAGE ASSSESSMENT
3 FLOOD DAM
MAGE ASSESSSMENT
3.1 In
ntroduction
This chapteer focuses on

o flood dam mage assessment. It outtlines the cllassifications of flood
damages and of benefits (Section 33.2) and approaches for damage
d asseessment (Secction 3.3).
Existing dam
mage modelss (Section 3.44) and economic and financial issues inn damage asssessment
are presented (Section n 3.5). Meth ods for the assessment of other tyypes of dam mages are
summarised d in Section 3.6. Section 3.7 describes how the re
esults of dam
mage estimate es can be
presented. In order to develop dam mage models, data collection is alsoo critical. This topic is
treated in Section
S 3.8. Finally,
F exam
mples of damaage assessme
ents for the LMB are presented in
Section 3.9.
The next Chapter (4) explains how the results of

o damage assessment c an be combined with
o estimate thee risk.
flood hazard analyses to
3.2 C
Classification
n of flood dam
mages and benefits
b
3.2.1 Flood damage

es
The conseq quences of a flood encom mpass multiple types of da amage. All thhese differentt types of
consequencces can be ob bserved afterr large flood disasters,
d inclluding the floooding of New
w Orleans
due to hurrricane Katrina
a in 2005. Thee term “floodd damage“ refers to all varrieties of harm caused
by flooding. It encompasses a wide rrange of harm mful impacts on humans, their health and their
belongings,, on public infrastructuure, cultural systems, cultural herittage, social systems,
ecological systems,
s induustrial producction and the competitive strength of tthe affected economy.
e
However, flooding
f mayy also have bbenefits and this topic is presented sseparately in the next
Section (3.2
2.2).
An overview w of different types of daamages/consequences is presented inn Table 3‐1. Damage

D is
divided intoo tangible annd intangiblee damage, deepending on whether or not the losse es can be
easily assesssed physically. Tangible ddamages can be directly priced
p in monnetary terms, because
a market value
v exists for
f the damaaged goods or services. Other types of damagess, such as
casualties, health effectts or damagees to ecologiccal goods andd to all kind of goods and d services
are not tradded in a market and requuire different methods of assessment iif monetary terms t are
to be used asa a standardd for compariison. They are
e indicated on the table a s ‘intangibless’.
Table 3‐1 Classifica

ation of flood ddamages.
Tangible Intangibble
Physical damage to: ‐ Health
h effects;
‐ Loss of life; ‐ Environmental dam mages;
‐ Housing, strructure and asssets; ‐ Cultural heritage vaaluation.
Direct
‐ Infrastructure and public utilities;
‐ Agriculture;
‐ Cultural herritage.
‐ Temporary relocation; ‐ Societa
al and culturaal disruption;
‐ Cleaning and sanitation; ‐ Increased vulnerabiility of survivo
ors.
Indirect ‐ Loss of income;
‐ Industrial prroduction lossses;
‐ Others.
ood Risk Assessm

Another distinction can made between the direct damage, caused by physical contact with
floodwaters, and damage indirectly following from a flood. Direct damage includes, for
example, damage to buildings, economic assets, loss of standing crops and livestock in
agriculture, loss of human life, immediate health impacts, and loss of ecological goods. Indirect
damages are damages caused by disruption of physical and economic linkages of the social and
cultural systems and the economy3, and the extra costs of emergency and other actions taken
to prevent flood damage and other losses (see, e.g., Parker et al., 1987). This category includes,
for example, the loss of production of companies affected by the flooding, induced production
losses of their suppliers and customers, the costs of traffic disruption or the costs of emergency
services.
The main damage categories that have been used in the past by neo‐classical economists are:
1. Direct and tangible damages
 Damages to properties and infrastructure

In the targeted areas, property and infrastructure damage can include damage to houses,
schools, offices, commercial and industrial buildings and installations, hydraulic works,
energy, transportation and other public infrastructure, and supplies/assets of individuals
and businesses. The extent of these damages is usually proportional to the magnitude and
duration of the flood. Damages to property are usually calculated as the costs of
replacement or restoration of an affected structure and inventories.
 Agricultural damages
Crop damages depend on the depth, timing and duration of a flood. The bigger the flood,
the larger the submerged cultivated area that sustains losses. However, if floods occur after
harvesting, there may be no crop damage at all. Damage to livestock and agricultural
equipment also depends on the magnitude and duration of the flood. These damages are
measured by the costs of replacing livestock or the loss of income due to a distress sale. The
damages to agricultural equipment/tools are the costs of replacement or restoration of
these assets.
2. Indirect and tangible damages
 Costs of temporary relocation and rescue

The costs of relocation and rescue are incurred by local government agencies and non‐
governmental organisations (NGOs) before, during and after floods to support affected
people living during the flood and recovering after the flood.
 Costs of cleaning and sanitation

Cleaning and sanitation expenses incurred after a flood are mainly labour costs borne by
individuals, households, enterprises and institutions to clean their houses and buildings.
 Income losses and higher costs of living

Losses of income due to disruption of economic activities and/or services by floods effect:
(i) individuals, landless labourers, families and enterprises; and (ii) large commercial and
industry enterprises that have been partly or fully closed for some period of time during
high flood water levels. Higher costs of living due to temporary relocation, the purchase of
food and other items that otherwise would be available from a family farm of household,
3
Alternative definitions of indirect damage exist. For example, Merz et al. define indirect damage as damage
that occurs outside the flooded area (Merz et al., 2004). E.g. companies can lose supply and demand from the
flooded area.

lack of basic supplie es as safe waater and elecctricity, and additional

a coosts for transsportation
and daiily activities all
a fall into th is category.
 Costs of
o illness
Importaant indirect damages caategories are e the loss off life and injjuries and th he health
effects.. It is noted that
t part of tthese health effects could be monetizzed by estim mating the
(increased) cost of treatment
t annd medical caare. Health co
osts due to fllooding affect humans
and livestock are re elated to woorsened environmental conditions annd increases in water‐
borne diseases.
d The
ese costs incl ude additional healthcaree costs for fammilies and individuals,
and additional expe enditures by local governm ment for diseease preventtion and treatment, as
well as for sanitatioon control duuring a flood.. Methods foor the estima tion of loss ofo life are
briefly discussed in Section 3.6 aas an example e of measurement of one of these cate egories of
other costs.
c
 Other costs:
c Other categories aare elaborate ed by specialists in other fields. Some
e of these
are becoming incre easingly impportant in asssuring that the measurres meet concerns in
internaational laws and treaties and particu ularly for susstainable devvelopment and rights
protecttions during disasters li ke flooding. Other important indireect damagess include
environnmental losse es, e.g., due to pollution, and societaal disruption in general. The
T latter
includee loss in social stability, pproblems in income distrribution and loss of conffidence in
government, negatiive impacts oof flooding on o tourism re evenue, loss oof income orr revenue
for peo ople or enteerprises outsside the targget areas, and other foorward and backward
econom mic linkages. These
T are ommitted from the study.
3.2.2 Flood benefitss
Flooding in the MRB cau uses not onlyy damage, bu ut also benefits. Since theese benefits can
c offset
the damagees, it is imporrtant to take these into account in damage assessm ment. Flooding has an
important and
a beneficia al role in maiintaining natural fisheries, wetlands, aand soil fertility. Flood
control measures need to take thesse into accou unt. Benefits can be meassured in the following
areas:
 Fisheryy sector
Flood water
w providees habitat fo r fish breedin
ng, nursing and growth. N Natural fish stocks
s are
an impportant resou urce for residdents of the LMB; particuularly in Cammbodia and Viet
V Nam.
Low floood levels re educe naturaal fish stocks. Though benefits depennd on the magnitude,
timing and duration n of floodingg, there is ge
enerally a direct relationsship between n flooding
and nattural fish stoccks.
 Agricultural production
Flood waters
w have a number oof agriculturral benefits. They: (i) briings sedimen nt to the
floodplain especiallyy boosting thhe fertility off acid sulphate soils; (ii) fflush toxic su
ubstances
from cultivated soil; (iii) kill thee rats and otther pests; (iv) speed upp the processs of plant
waste decompositio
d on; (v) impro ve soil structture; and (vi) promote sooil fertility. Th his impact
can be measured byy comparing hharvests in ye ears of differe
ent flood leveels.
 Coastal land expanssion

Flood water
w brings sediments tto river moutths in Viet Nam,
N slowly eexpanding th
he Delta’s
land area into the se
ea. This can eeasily be meaasured and va
alued.
 ersity conservvation
Biodive
Floodin
ng plays an im
mportant rolle in maintaining biodiversity in the M
MRB, especia
ally in the
wetlandds in the LM
MD. There a re a number of studies on these beeneficial imp pacts, the
monetaary values callculated in thhese studies can
c be applied.
ood Risk Assessm

MRC Flood
d Management and
P Com
In general these benefits are associated with relatively smmall floods and processess that occur
relativelyy frequently (e.g., annua
al or seasonnal floods). Extreme
E events that aree infrequent
(higher return periods) generally bring
b fewer bbenefits.
3.3 General ap
pproach to economic dam
mage assesssment
3.3.1 Scope of the analysis and level of dettail
Before sttarting a dammage assessment it is impoortant to decide on the scope and leveel of detail of
the damaage evaluatio on. This depe
ends on a) thhe geographicc area of the study, b) thee objectives,
c) the avvailability of resources for conductingg the study, and
a d) the avvailability of data. These
factors are outlined below.
b
a) The ggeographic area

a
The size of thee area unde er investigatiion can varyy from area as of nationnal or even
interrnational scalle to local are
eas, hence raanging from basin,
b countryy, district, villlage or even
indivvidual structu
ure level. Duee to the fact tthat very preccise methodss require morre effort (i.e.
timee and money)) than less de etailed approoaches and th hat the resouurces are usuually limited,
the most precisee methods are often resttricted to sm mall areas unnder investiggation, while
studies with a reesearch area of regional oor even natio onal size mostly have to rely on less
detailed methodss.
b) The objectives
The method and the accuracyy of the resullts of a speciffic study musst match the objective of
the sstudy. If the goal
g of the sttudy is strateegy developm
ment and plannning, a less pprecise (and
conssequently less costly and time‐consum ming) methodd might be suufficient. If, oon the other
handd single flood d protection measures hhave to be assessed, (e.g g., for a priooritisation of
invesstment), morre precise me easures are nneeded. Befo
ore investing in a more prrecise study,
one should consiider whetherr such detail information really effectss the results or decisions
to bee made.
c) The availability of
o resources
Budgget or time reestrictions usually limit th e precision th
hat can be acchieved in a sstudy. If only
a sm
mall budget and/or
a little time
t are ava ilable, quick and easy me ethods and aapproximate
measures are neeeded.
d) The availability of
o data
The effort or costs of damagge evaluationn can be siggnificantly reduced if neccessary data
alreaady exists. The
T availability of pre‐exxisting data should determine the chhoice of an
apprropriate meth hod. If, for exxample, adeqquate land‐usse data or a thoroughly deeveloped set
of ddamage funcctions alread dy exists, thhis may faccilitate the application of detailed
apprroaches on a regional orr even nationnal scale. Fu urthermore, the t type of ddata is pre‐
existting and availlable in a cou
untry or regioon can also haave a significa
ant impact onn the choice
of a method.
n of damage categories
Selection
A key quuestion in dam

mage evaluattion is which categories ofo damages sh hould be incl uded. While
it would be desirablee to take all of the typess of damagess listed in tab ble 3.1 into aaccount in a
damage assessment, this could lea ad to extensiive and costlyy studies. It iss often moree practical to
come to an agreement accepted by all stakehholders on wh hich damage categories aare the most
importannt, seek conssensus on measures of eestimation if precise measures are unnavailable or
costly, and to focus on them to o keep the aanalysis man nageable. It is importantt that these

selected caategories con nstitute the largest partt of the tottal damage. For example, in the
standardiseed damage accounting m method that is used in the Netherlandds the follow wing main
damage cattegories are prioritised: laand use, infraastructure, households, coompanies an nd ‘other’.
Within each of the ma ain categorie s are sub‐caategories (e.g g., agriculturaal use, urban use, or
recreationaal use within the
t main landd‐use category).
Less importtant categorie es can also bee included byy using estimation measurres, such as applying
a a
typical dam
mage multiplier to accou nt for them.. For “intang gible” damagge categoriess that are
inappropriaate to standardise in term ms of market values, other measures ccan be chosen such as
the numbeer of lives lo ost. (See, foor example Section
S 3.6 for a discusssion of metthods for
estimating loss of life).
3.3.2 D
Direct damage
e assessmentt
Methods fo or the estima ation of direcct economic damage to physical

p objeccts, such as structures
s
and housess are well established, andd the use of ’damage functions’ to esttimate these damages
is widespreead. (See bellow for furthher explanatiion and also Penning‐Row wsell and Ch hatterton,
1977; Duttaa et al., 2003;; Kok et al., 2 005; and Jonkman et al., 2008
2 for add itional background.)
Flood damaage estimatess require infoormation on the:

t
a) intensity of flooding;
b) numbeer and type of o land uses a ffected by flo
oods, and the
eir value; andd
c) suscep ptibility of la
and uses to flooding (re epresented by floodwatter damage function/
equatiions).
These facto
ors are presen
nted schemattically in Figure 3‐1 and de
etailed below
w.
Figure 3‐1 General approach to damagge estimation, with

w the relatio
onship betweenn land use and flood
f
depths used to calculate daamages.
ood Risk Assessm

MRC Flood
d Management and
P Com
a Intensity of flooding
f
DData on the intensity of a flood, (i.ee., the affecte
ed area, the flood waterr levels, and
iinundation characteristicss) are essenttial to damagge estimates. The area annd depth of
iinundation are
a the mostt important ffactors in me easuring inte
ensity, but nnon‐intensity
ffactors such as the durattion, time off occurrence,, and water velocity can also have a
ssignificant influence on damage
d depeending on the types of floood. Informaation on the
iintensity of flooding
f and its measuremment is preseented in the flood
f hazard assessment
((Section 2).
b Number and type of land d uses affecteed land uses, and their value
DDifferent typpes of land usses that correespond with damage cate egories are a key part of
ddamage valu uation. For example,
e for property daamage, inform mation on thhe location,
nnumber and type as well as the elevattion of prope erties potentially affected by a certain
fflood is a parrt of damage estimation. SSuch informaation is normally availablee in the form
oof primary data
d from fie
eld surveys oor secondary data, such as a public maaps. General
iinformation ono data collection and lannd use surveyys is presented in Section 33.8.
c SSusceptibilityy of land uses to floodingg

SSusceptibilityy relates eco onomic damaage to flood d conditions. Often, (stagge) damage
ffunctions ussing flood depth
d measuures are the e standard tools, but oother flood
ccharacteristiccs, such as flo
ow velocity o r flood durattion could also be relevantt and can be
aadded as variables in the modelling eqquations. Furrther details on o damage fuunctions are
ppresented in Section 3.3.
It is possible to descriibe how these elements inn the damage e model can be
b combinedd to estimate
the total damage in a flooded area a in an equattion. It is simp
ple to describ
be in words foor those not
accustom med to matheematical sym mbols. The tootal damages are a summation of the damages at
all of thee specific poin
nts in the flood area and the damage in all of the categories
c reelated to the
intensityy and characteeristics of thee flood:
m n
D   i ( h r ) D max, i n i , r
i r (3.1)
where: Dmax,x,i =
maximum m damage am mount for an object or lan nd use categoory i
i= damage or land use ccategory
r= location in flooded arrea
m= number of damage caategories
n= number of locations iin flooded are ea
hr = hydraulicc/flood charracteristics (e.g., intensity) of the flood at a
particula
ar location
αi(hr) = stage damage functioon that expresses the fracttion of maxim mum
damage for category i as a functio on of flood characteristics at a
particula
ar location r (00 ≤ αi(hr) ≤ 1))
ni,r = number of objects of damage cate egory i at loca
ation r
For each damage category a specific stage dam mage functio

on can be derrived and theen applied in
the damaage assessmeent (e.g., corrrelating histoorical damage
e data with flo
oodwater deppth).
3.3.3 Damage fun

nctions
Damage functions relate the leve el of damage to flooding conditions. AnA example oof a damage
function is given in Figgure 3‐2 for damage
d to hoomes and theeir contents in the Netherrlands based
on the 19953 flood (Seee Kok et al. (2005).) The ffigure shows that flood da
amage is totaal (100% or a
factor off 1.0 in the chart) when water
w depth exceeds 4.5 meters. Therre is a large vvariability in

damage functions, often different by country. For example, in the United States the Federal
Emergency Management Agency (FEMA) has developed the HAZUS risk assessment software
model that includes damage functions for different damage categories4.
1,1
1,0
0,9
0,8
damage factor
0,7
0,6 total damage

damage building
0,5 damage content
0,4
0,3
0,2
0,1
0,0
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 5,5 6
Flood depth (m)
Figure 3‐2 Example of a stage damage function for houses in the Netherlands (Kok et al., 2005). Note that
the curves for building damage and contents show these as percentages of the total damage.
Influence of flood characteristics on the damage function

Most stage damage functions include water depth as the main determinant of direct damage.
However, other flood characteristics, such as flow velocity or flood duration can also influence
damages. Thieken et al. (2005) investigated the influence of these and other factors, including
flood duration, contamination and flood damage preparations, based on data for the 2002
floods in Germany. Other factors, including psychological ones like risk perception of residents
in flood prone areas and their preparedness can also influence susceptibility to risk and, hence,
damage. Such factors have yet to be included in damage functions − at least explicitly.
In the application of damage functions to estimate the damage for agricultural land use (e.g., for
crops or rice paddies), other factors besides flood depth are important in assessing damage.
Duration of flooding, for example, can extend damage and/or prevent harvesting. There is also a
seasonal effect, since damage depends on the stage at which floods occur in the growth and
harvest cycle. Damage functions need to account for these damage processes (moment during
the year, duration) and characteristics of the land use (such as crop characteristics).
Relationship between structural damage and economic damage

An important source of economic damage is the structural damage to buildings, especially
homes. Figure 3‐3 shows how economic damage is connected to structural damage. The level of
structural damage caused by a flood is determined by the physical forces (or loads) and the
building resistance (or strength) (Kelman and Spence 2004). The degree of structural damage
depends on the intensity and magnitude of the flood (i.e., hydrostatic and hydrodynamic
forces), and on the building’s resistance to flooding. To convert the structural damage to
economic estimates requires insight in the building’s pre‐disaster market value and the
replacement cost. Apart from damage to the structure, the damage estimate in monetary terms
should also include damage to the inventory of the structure (for instance items in the building).
4
see http://www.fema.gov/plan/prevent/hazus/ for more details

MRC Flood
d Management and
P Com
In the deescription of damage functions, abovee, the economic damage estimate is ddescribed as
directly rrelated to flood characterristics. Howevver, it is someetimes imporrtant to havee insight into
the degrree of structtural damage e and the nuumber of daamaged or destroyed
d buuildings. The
degree o of structural damage ca an be measuured in the number of damaged o r destroyed
buildingss. In Appendixx 2 existing methods
m for tthe analysis of
o structural damage
d are ssummarised.
For strucctural damagee both the de epth and flow w velocity will be importannt. The level oof structural
damage is also imporrtant for hum man safety annd loss of life as buildinggs are importtant shelters
during floood events.
Figure 3‐3 Schema of

o structural flood damage anaalysis.
3.3.4 Relative and

d absolute da
amage assesssments
There arre two basic approaches to damage assessment using damage functionss; one using
‘absolutee damage’ fu unctions (a specific
s finanncial measurre) and one using ‘relatiive damage’
functionss (percentagee loss of the total
t potentiaal loss). The approaches
a differ
d in the w
way in which
the inforrmation sourcces on land use, values of assets and damage functions are com mbined:
1. The use of absoluute damage functions

f forr damage assessment
he absolute damage appro
In th oach, the dam mage functions describe the
t relationshhip between
floodd depth or flood
f water levels and tthe direct monetary
m dam
mage value. Surveys on
damage during historical
h floo
ods combineed with data on inundation depth annd land use
inforrmation, can be used to generate eqquations for damages in specific asseet categories
(agriculture, houssing, infrastru
ucture, etc.).
2. The use of relativve damage fuunctions for ddamage asse

essment
he relative daamage approach, the dam
In th mage functionn describes the relationshhip between
flood
d depth and d flood dam mage for eacch specific laand use cate egory as a fraction (or
percentage) of th he maximum potential daamage (100% % losses). The
e total value of assets at
risk is calculated by combining land use information and asset values. Then, this sum is
compared to thee maximum potential daamage to express the da amages as a percentage
ratheer than in moonetary terms.
Figure 3‐‐4 depicts hoow the differe

ent elementss are combine
ed in the two
o approachess. The figure
also showws the differeent formats of
o the damagge functions for
f both case es. The main ddifference is
that in th
he relative daamage approach, the dam
mage function n produces a relative dam
mage fraction
(or perceentage) as a function off water dept h, whereas in the absolu ute damage approach a
damage value followss directly fromm the damag e function.

Approach Functtion
Damage ($)
Damage
Absolute approach
H
Hydraulic A
Absolute depth Land use
characteristics damage function
d information
Water de epth or
Susceptibility of level (m))
Value of
o assets
assetss
Damage (%)
Relative approach
Damage
100%
H
Hydraulic Relative depth Total value of
charracteristics dam
mage function assets at risk
Water de
epth (m)
Su
usceptibility of Land use
V
Value of assets
assets information
Figure 3‐4 Absolute and

d relative damaage approachess.
The two ap pproaches ha ave advantagges and disad

dvantages in applicability to other are
eas and in
adaptabilityy to local data
a.
Absolute daamage functions are speccific to the areas a where they have beeen derived. For each
new region n, data is req quired to derrive the speccific absolute e damage fu nctions for that
t area.
Relative damage functio ons are moree generally ap pplicable sincce they look aat damage immpacts on
land‐use caategories. If they have bbeen derived for a certaiin region andd event, the ey can be
applied to a wider region for whicch no historical flood damage data iss available. However,
relative dammage functio ons can only be extended d to other areas where ddamage proce esses and
susceptibilitties are similar (e.g., floodds to similar agricultural lands and struuctures). The
e absolute
damage ap pproach is th hus more suiitable when large region nal differencees exist in the above
factors.
Absolute damage functtions are alsso advantage eous where a comprehe nsive set of absolute
damage fun nctions alreaady exists. If on the other hand, detailed data on asset valuess is easily
available, th
he relative da
amage approoach may be more
m convenient.
Note that forf both app proaches, thee varying leve els of detail at different geographic levels can
distort the analysis and hide errors. If data at the commune level is combbined with da ata at the
district leveel and the ressults are pressented at the
e commune le evel, this kindd of ‘false’ acccuracy is
likely.
3.3.5 A
Assessment off indirect dam
mages
Indirect damages includ de such cateegories as daamages caussed by disrupption of phyysical and
economic liinkages of the
e economy, aand the extraa costs of eme
ergency and oother actionss taken to
prevent flo
ood damage and other llosses. The loss of production of coompanies afffected by
flooding, in
nduced prod duction lossees of their suppliers
s and
d customers,, the costs of traffic
disruption or
o the costs of
o emergencyy services are also included
d in indirect ccosts.
ood Risk Assessm

MRC Flood
d Management and
P Com
The lossees due to bussiness interruption and indduced losses in other partts of the econnomy can be
very sign
nificant, espeecially if impo
ortant econo mic hotspotss are flooded. If an imporrtant airport
or road nnetwork is afffected by floooding and haas to be closed for a long
g period or iff a cluster of
importannt companiess is affected, these lossees can be su ubstantial. Th
his effect cann estimated
using a mmodel that in ncludes the existing
e netw
work of linkagges within an
n economy. TThe method
relies on
n “input‐outp put” models. (See, for exxample, Van der Veen et al. (2003))). Yet, these
approach hes are bein ng researche ed and publ ished in acaademic litera ature but haave not yet
developeed for practiccal use.
Today, inn practice, lo

osses due to business inteerruption are
e generally estimated
e by assuming a
certain in
nterruption period
p (e.g. a number of months) and d a cost factor that inclu des the lost
economic activity (added value) an nd the numb er of affected
d employees..
Surveys ccan also be useful

u to determine the i ndirect damaage categorie es and the daamage costs
for indireect damages. Eventually, this offers ddata that can demonstrate a ratio bettween direct
and indirect damagees. With such h a ratio, it i s possible to
o estimate th he indirect ddamage as a
function of the directt damages. Further detaills on the app plication of th
his approach to estimate
indirect eeconomic damage to the LMB are founnd in Volume e 2C.
3.4 Damage models

m
For conssistent decisiion‐making itt is desirablee to have a standardised approach for damage
assessmeent on a wide level, such as for a riveer basin or at the country level. With tthis in mind,
governmments in seveeral countriess have deve loped standaardised meth hods and sofftware tools
that can be used acro oss countriess for the esti mation of daamage due to
o flooding ass well as the
risk and ccost‐benefit calculation.
c
Exampless are the HAZUS

H meth
hodology, deeveloped in the USA (FEMA, 20033), and the
standarddised damage model de eveloped in the Netherlands (Kok et al., 20055). In these
approach hes, the land
d use data an
nd damage fuunctions are included in a computer m model, such
that the damage can be easily esttimated if infformation forr a certain flo
ood scenario is available.
Based on n the approaach developeed in the Neetherlands, a more generric model/toool has been
developeed under thee name of DA ACA (Damagee and Casualties Assessm ment) and haas also been
tested in
n Thailand – see Textbox 1.
1
An autommated approaach is specificcally attractivve, when:

 A larrge number of o damage ca ategories andd/or a large amount of la and use dataa need to be
proccessed;
 The underlying daata on land useu and valuees of assets is available in GIS
G format;
 Therre may be a large numbe er of calculattions and/or different scenarios beingg compared
(e.g., situations with
w various alternative
a prrotection sche
emes and meeasures).
 Diffeerent areas co onsidered in the model uuse the same equation forr damage esttimates (i.e.,
the land uses and d housing typ
pes are compaarable).

Textboxx 1 DACA (Damage and CCasualties Asssessment) an

nd its applicaation to Thailland
DACA (DDamage and Casualties

C Asssessment), iss based on th
he Dutch dammage and casu ualties
assessm
ment model ‘H HIS‐SSM’. Thee assessmentt takes into account
a the sspatial distrib
bution
of land uses
u and their economic value, as well as the diffeerent charactteristics of flo
ooding
scenario
os. The tool calculates
c dirrect damage, production losses
l and inndirect damagges, in
ways that tie damagges to their oorigins and that are desig gned to aid ddecision makkers in
land usse planning, flood manaagement and d mitigation planning D DACA also pushes
p
governmmental agencies and depaartments tow wards more collaboration due to its re eliance
on a standard approaach and transsparency in methods
m and data.
The Thai Department of Water Reesources has tested DACA A in a pilot arrea in Thailan
nd; the
provincee of Chiang Rai, which inclludes several rivers.
Source: “Developmen nt and Use/D

Demonstration of DACA in de LMB” ((Anonymous, 2009;
Leenderss et al., 2009).
3.5 Economic and

d financial isssues in dam
mage assessm
ment
Within the context of damage

d asseessment the economic va aluation of d amages and land use
values are important. Some
S of thee main (and, yet, fully un
nresolved) isssues are sum mmarised
on is includedd in Appendixx 3 and provided in backgground literatture; (see,
below. Furtther discussio
for examplee, Messner ett al., 2007).
3.5.1 Financial vs. economic valuuation
There are different

d perrspectives onn damage ap ppraisal, startting with thee distinction between
financial an
nd economic valuation.
v Finnancial evaluaations consid
der damage frrom a perspe ective of a
single persoon or firm, ne
eglecting pubblic affairs and focusing onn the actual ffinancial burd
den on an
individual or
o firm. Econo omic evaluatiions are broader. They see ek to assess tthe impact onn national
or regional welfare, including impactts on intangible goods an nd services. TThis broader economic
e
perspectivee is the approopriate one tto apply if caalculations off flood damaage are to be e used for
public policcy.
3.5.2 Estimating values of assetss
e insight in the values off the objects and land

For damagee assessmentts it is imporrtant to have
uses that co ntially damagged. The quesstion is what economic vaalue to assign to them.
ould be poten
The basis foor valuation favoured

f by nneo‐classical economists is the markett price for an object or
unit. The economic evaluation of gooods and serrvices which are traded inn the markett is a task
that is relaatively easy to accomplissh. Howeverr, it is widely recognisedd that several effects
“distort” market prices and
a need to bbe taken into o account in damage
d mation5.
estim
5
Since these ideal condition ns are never fulffilled in a real‐w
world market, neoclassical
n ecoonomists try to calculate
what they calll the ideal ‘shadow price’ of thhe good under examination. TheyT translate mmarket prices in
nto
shadow pricees by trying to account for som me of the distortions that are easy
e to measuree such as taxes and
subsidies (‘traansfer paymentts’ to particularr individuals or groups), or by trying to accouunt for monopolistic
market pricess or influences of major buyerrs and sellers with other agend das into ‘compeetitive’ market prices that
assume an ‘equality’ of all potential buyerss and sellers.
ood Risk Assessm

MRC Flood
d Management and
P Com
The conccept of depreeciation shoulld be applied in order to reflect

r the value of a goodd at the time
when it is damaged by
b a flood. Replacement coosts, alone, canc sometime es overestimaate damage,
because replacemen nt usually involves improovements whilew depreciation accou nts for the
diminutio
on in value ofo somethingg over its prroject usable life. Old goo ods which arre damaged
during a flood are ussually substittuted by neww, more prod ductive and better
b perforrming goods
(Messner et al., 2007).
Similarly,, for a small group of o irreplacea ble assets, replacement cost accoounting and
depreciation are inapplicable and market vaalues may not apply; such as artisticc treasures,
religious or heritage structures
s with emotionall value or spiritual significance tied to tthe belief of
a people (such as anccient trees an
nd spirit shrinnes in the MR
RB), and forms of cultural oor biological
diversity that may be key to survivval of a large r system. Oth
her standardss of valuationn need to be
applied.
3.5.3 Time value of money
In an eco o time. Forr example, a

onomic analysis it is imporrtant to consiider the value of money over
building that is wortth USD 50,0 000 this yearr could have e a higher or lower pricee next year
dependin ng on expectations people e have aboutt the future and
a their willingness to coonsume now
or later, and on otherr factors.
Damage values that are

a used in a certain damaage model arre usually bassed on data ccollected for
n year in thee past. When one is esttimating dam
a certain mages for the current sittuation it is
importannt to take in
nto account changes in the values and amountts of damagges. Usually,
economic growth incrrease values, but values c an also decre
ease and dam
mage factors ccan increase
or decrease with changing conditions (like chaanging population densitie
es), while rellative values
can also change.
3.6 Loss of life and health effects
3.6.1 Introduction
Apart froom the estim mation of ecconomic dam mages, or conversion of damages intto economic
terms,, tthere are sevveral approacches to the eestimation off other dama age types andd/or for the
calculatio
on of damagees in different measures. LLoss of life an
nd general he
ealth impactss are an easy
example. In this Section some exissting approacches and literrature are briefly summariised.
Note thaat while thesee approachess and methodds are not ye

et part of standardised meethods used
by those currently em
mpowered to conduct dam mage assessm ments6, this shhould not refflect on their
importannce.
Note thaat when damaages like the loss of life arre expressed in terms of economic
e dammage (e.g., a
monetarry value or daamage sum per p fatality), tthere is no uniform appro oach given thhat different
societies, different decision‐make ers and diffeerent disciplin
nes apply diffferent valuees to human
uing the losss of life direcctly raises etthical and po
life. Valu olitical questtions that aree not easily
visible w
when econom mic ‘market’ valuations aare used for other dama age categoriees. To avoid
these disscussions, thoose currentlyy conducting damage asse essments pre efer to assesss the loss of
life separrately and ind
dependently from the ecoonomic damaages.
6
One exceeption is the ap
pproach to estim
mating loss of liife that is includ
ded in the stand
dard model forr flood
damage asssessment in thhe Netherlands.

3.6.2 Lo
oss of life
In recent years
y several methods haave been devveloped for the estimatioon for loss off life (see
Jonkman, 2007).
2 In gen
neral the folloowing steps are needed to estimate loss of life in a given
area:
1) mation regard
Inform ding flood chaaracteristics;
2) An anaalysis of the exposed
e popuulation and evacuation procedures;
3) An estimate of the mortality (peercentage of deaths) among the expossed populatio
on.
The applicaation of thesse steps and the level off detail of the analysis deepend on the type of
flooding stu
udied. For anyy applicationn, the output of hydrodyna amic flood sim
mulations can be used
to estimatee flood characcteristics (seee Section 3.3). For large‐scale floodingg associated with
w dyke
breaches (ssuch as in the e Netherlandds or New Orrleans), steps 2 and 3 cann be applied without
w a
high level of
o detail. Gen neral estimattes of the possibilities for evacuation of the population can
also be used. In this ana alysis the effeects of warning systems also need to bbe taken into o account.
An examplee of a mortality function iss presented in Figure 3‐5. This functionn is applicable e to areas
where the water
w is risingg rapidly, so tthat people have
h limited time
t to reachh a safe place.
Figure 3‐5 Example of a mortality funcction derived fo

or large‐scale flo
ooding in the N
Netherlands (Jonkman,
2007).
For smallerr‐scale floodss where veloccities are mo ore importantt, criteria havve been deve eloped to
assess hum man instabilityy (i.e., ‘toppliing’) in flood flows. In general, depth‐‐velocity multiples are
used to indicate the com mbination of depth (h – [m m]) and veloccity (v – [m/s]]) that would lead to a
person’s in nstability. Ovverall, the avvailable stud dies and exp periments shhow that people lose
stability in flows of relatively low deepth‐velocity multiples. Th he critical deepth‐velocity multiples
for maintaining vertical stability rannge from 0.6 m2/s to about 2 m2/s. Thhis means th hat a flow
with a 1m depth
d and a velocity
v of 1 m
m/s would most likely lead d most individduals to be swept into
the floodwaaters. Further details on hhuman instab bility are included in Appenndix 4.
3.6.3 H
Health effects and social annd environmental impactss
Important non‐lethal
n he
ealth impacts of floods are
e injuries and
d illnesses. Thhe occurrence
e of these
and their im
mportance is highly depenndent on (am
mong other fa actors), generral health and
d medical
ood Risk Assessm

MRC Flood
d Management and
P Com
infrastructure in the area (before e and immeddiately after the flood), climatic fact ors, type of
floodwatter (fresh waater or salt water) and duration of flooding, Th here is no sstandardised
approach h to “predictiing” these he
ealth effects. Hajat et al. (2004)
( and Ahern et al. (22005) offer a
compreh hensive overvview of the avvailable infor mation regarrding health impacts of flooods.
In addition, there arre often longger term psyychological he

ealth effects as a result of flooding.
There is also evidencce of a link between
b psycchological he
ealth effects and
a post floood mortality
(see, e.g.., Bennet, 197
70).
Flooding can also haave adverse effects on the environment and th he social sysstem. These
damagess are generallly analysed byb means off qualitative/d
descriptive methods
m and require the
hiring of appropriate specialists.
3.7 Presentatio
on of damagge assessme nts results
The resuults of flood damage assessment cann be presented in differe ent ways. Asssuming that
flood dam
mage is calcuulated for a giiven flood sceenario, the fo
ollowing presentations aree possible:
 The ddamage amo ount for the fllood scenarioo (in USD or another
a curre
ency);
 The absolute dam mage value byb category aand the distriibution of the total damaage over the
categgories;
 “’Un
nderlying’ datta (raw data that has alre ady been pre esented in monetary dam mages as part
of th
he overall dam
mage) such as numbers off flooded hou uses, or flooded hectares oof farmland;
 Dammage per hecttare as an indicator of “da mage densityy” in the area a.
As the efffects of a flo

ood are spatially distributted it can be
e relevant to present mapps that offer
informattion on the daamages. Geographical infoormation systems (GIS) facilitate this. ((See Section
2.7). Thee resolution of
o these map ps and the leevel of spatiaal detail depend on the ((geographic)
scope of the analysis and the data available.
Damage per hectare can be determined preecisely if the underlying data d (e.g., laand use and
values) aare available in a high leve
el of detail peer unit area. It is also possible to averaage damage
per hecttare over a laarger area when
w damagee functions area available for a larger area, e.g. a
district.
Examplee of a damage
e assessmentt
A damagge assessmen nt for a floodd scenario in the Netherlaands is offere

ed below as aan example.
This flood scenario deepicts the flooding of the densely popu ulated South‐‐Holland provvince due to
an extreeme coastal flood. This scenario
s desccribes the flooding of ann area with one million
inhabitannts where floood consequ uences can bbe serious. The total dam mage for thiss scenario is
estimateed with the Dutch standard dised damagee model at rooughly 23.7 billion
b Euros.
Table 3‐11 offers an overview

o of the
t distributi on of the total damage overo damagee categories.
The spattial damage distribution
d for
f a flood sccenario in the e Netherlandds is given byy Figure 3‐6,
showing the density of damage per p hectare. These resultts are used in flood risk studies and
discussio
ons on the revvision of the national policcy on flood protection
p in the
t Netherlannds.

Table 3‐2 Distributtion of damagee over damage categories

c for a flood scenarioo for South Hollland
t Netherlandss (Jonkman et al., 2008).
Province in the
Direct damage
Damage Category % of total damage
d
(mln
n. EUR)
Residences and inventoryy 16,000 69%
Commerciall and public bu
uildings 2,300 10%
Vehicles 40 0%
Infrastructure 300 1%
Airports 500 2%
Agriculture 1,900 8%
Direct damaages due to bu
usiness interruuption 2,700 11%
Total 23,740
2 100%
Figure 3‐6 Damage inte

ensity per hectaare of the flooding of South Ho
olland Provincee in the Netherlands.
3.8 D
Data collectio
on
If no damagge functions are

a available,, they need to be developped using dataa from historrical flood
events. Thee table beloww offers a geeneral overvie
ew of the daamage categoories, their definition,
d
the type of o data and sources thaat could be utilised to collect data,, and a sum mmary of
measurablee items that can be useed as data. The T table is based on thhe presented d general
classificatio
on of flood damages
d (seee Section 3.2.1). As partt of this datta collection,, data on
benefits of flooding shou
uld also be inncluded.
The table below

b can be
e seen as thhe model list of information for seve ral items. In practice,
however, data
d on manyy of these ittems are stilll often impossible to col lect given th
he limited
availability of secondary data, tim
me and budgget constrain nts. Section 3.9 shows how this
approach has been applied in Stage 1 of the FMMMP‐C2.
ood Risk Assessm

Category Definition Type Source of data Summary of data items that could be collected
Direct damages
Table 3‐3
Loss of life and Reported number of deaths, injuries and missing persons Secondary District/Provincial ‐ No. of deaths
injuries due to the flood data level direct flood ‐ No. of injuries
damage reports ‐ No. of missing persons
Damage to Estimated replacement or restoration cost of an affected ‐ No. of houses collapsed or damaged
properties and structure and inventories in the targeted areas such as ‐ No. of business collapsed and damaged
infrastructure. houses, schools, offices, commercial and industrial ‐ Km of roads
Guidelines Flood Risk Assessment

buildings and installations, hydraulic works, energy, ‐ No. of schools/classrooms
transportation, infrastructures and supplies/assets of ‐ No. of irrigation schemes
individuals and businesses ‐ Other items. based on the actual indicators that
are available in the Provincial and District damage
reports
Crop damages Agriculture land inundated or agriculture production lost, ‐ Ha of crops destroyed
Overview of Flood Damage Data
food and seed stocks damaged ‐ No. of fruit trees destroyed

‐ Ha of land damaged/covered in sand
‐ Tonnes of expected harvest lost
‐ Stocks of produce and seeds lost
‐ 84 ‐
Livestock Drowned livestock and loss of income due to distress ‐ No. of dead poultry, large and small livestock
damages sales
Damages to Damaged aquaculture and fishing equipment ‐ Fish/shrimp ponds damaged
fishery ‐ Fish cages, rafts and traps damaged/lost
‐ Boats lost
Damage to Inundation depth‐damage relationships and effect of Primary data Focus Group ‐ Hypothetical flood damage to paddy in relation to
paddy flooding duration and timing on net benefits Discussion inundation depth and timing of floods
Damage to Inundation depth‐damage relationship and house value Primary data Household/ ‐ Actual and hypothetical damage to houses as
houses business survey percentage of house value in relation to
inundation depth
May 2010
Category Definition Type Source of data Summary of data items that could be collected
Indirect damages
Cost of illness Costs associated with increased incidence of water borne Primary data District ‐ Amount spent by individuals on flood related
diseases compared to conditions of no flooding. Such authorities, line health care
costs include additional cost of health care that families agencies ‐ Loss of labour from flood‐related disease
and individuals spend on flood‐related illnesses; and ‐ Amount spent on drugs, campaigns, care for
additional expenditures from local government for disease prevention
disease prevention and treatment and for sanitation
control during the flood
Cost of Costs by local government and NGOs incurred before, Primary data District, NGOs ‐ Transport, temporary shelter costs, costs of food

temporary during and after a flood and drinking water, etc.
relocation and
rescue
Income losses Losses due to disruption of economic activities and/or Household survey ‐ Cost of repair/reconstruction of residential
services by the flood for (i) individuals, landless labourers, Business survey structures, including:
families and enterprises; (ii) large commercial and ‐ Hypothetical damages to the residential
industry enterprises that have been partly or fully closed structures based on flood levels
during high flood water levels ‐ Agricultural assets/incomes, crop losses
‐ Livestock losses, fish/shrimp losses
‐ Other HH income sources losses (e.g., home‐
based businesses, hired labour)
‐ Loss of work days
‐ 85 ‐
‐ Health costs
‐ Damage to business structures, equipment and
inventory
‐ Losses in revenues and employment
‐ Private expenditures for flood protection
Higher cost of Due to relocation water, food and fuel that otherwise ‐ Costs/purchases of food, water, fuel, etc.
living would be available on the farm must be purchased
Cost of Preventive measures ‐ Cost of sand bags, etc.
prevention
measure
Cost of cleaning Cleaning of houses, shops and other buildings ‐ Labour costs and materials
up after the
flood
Benefits
Flood benefits Positive aspects of flooding for agriculture, fisheries and Primary data Focus Group ‐ Household benefits from fishing the flood waters,
otherwise Discussion fertility and pest control savings. to fields.
May 2010
3.9 Flood damage assessment in the Mekong River Basin
3.9.1 Scope and selected areas
In Stage 1 of the FMMP, several focal areas were selected to investigate in detail the flood
damages and risks. These areas and their counties were:
Country Focal Areas

Thailand: Lower Nam Mae Kok
Lao PDR: Bokeo province and the Lower Xe Bang Fai River basin
Cambodia: Kratie province, Right Bank Bassac River and Left Bank Mekong River
Viet Nam: Upper Se San River basin, Plain of Reeds (PoR) and Long Xuyen Quadrangle
(LXQ)
The four delta focal areas in Cambodia and Viet Nam compose two transboundary areas, one
east of the Mekong River and the other west of the Bassac River.
The detail of the assessment is at district level, although part of the data is available at
commune level. (See Section 3.9.4 on data collection and surveys, below.)
Below, the general implementation of damage assessment is described and results are
illustrated for three districts in Cambodia that were analysed as focal areas in the Stage 1. These
are the districts of Koh Andet, Koh Thom and Kampong Trabek. More information on the
damage assessments for these districts can be found in the Volume 2C “Flood Damages,
Benefits and Flood Risk in Focal Areas”.
3.9.2 Approach and methodology
As described in Section 3.3.4, there are basically two approaches to flood damage/risk
assessment: absolute damage and relative damage. In the absolute damage approach, historical
damage data for an (administrative) area are used to assess the flood damage/risk in that area.
In the relative damage approach inundation‐damage relationships are developed on a per unit
(land area such as hectares or homes) basis, and the flood damage/risk is assessed by applying
the per unit risk to the number of units in the concerned area.
For the top‐down approach only a limited set of data is needed to determine flood damage (in
absolute damage curves) and it should be available at government agencies (see Section 3.3.4).
For the bottom‐up approach, various data sets and many processing steps are required to
generate relative damage curves (see Section 3.3.4).
In this study both approaches are applied for damages to housing and agriculture. For
infrastructure damage and relief costs, only the top‐down approach can be applied due to a lack
of adequate spatial information.
Hydrological hazard data (peak water discharges and volumes for different return periods) are
of sufficient accuracy to be used in risk assessment. The methodology can usually be applied
basin‐wide. However, use of the data is not easy. Hydrological hazards must be converted into
flood hazards (with measures of flooding depth and duration for different return periods).
Limited topographical information and the lack of available high quality models hamper
accurate flood damage/risk assessments. Flood damage/risk maps for LMB focal areas should,
therefore, be considered with caution.

Flood vulnerability assessments have been made for a number of sample districts in the focal
areas. These assessments allow for the estimate of indirect damages due to flooding in addition
to reported direct damages. Damage curves could be estimated providing the relation between
flood damages and inundation depth and time of flooding in certain areas.
There is a GIS‐based tool that allows for the calculation of flood damages/risks by combining
flood maps (see Chapter 2) and other available GIS data sets with their derived damage curves.
This tool is, in principle, applicable basin‐wide, provided that it is supplemented with flood maps
and that adequate GIS data sets are available.
3.9.3 Damage categories
In the LMB the most important damage categories that were analysed as part of Stage 1 of the
project are:
 Infrastructure and Relief

The damages combined in this category included those to facilities, equipment and
materials for education, health, irrigation, bank erosion, fisheries, transport,
communication, industry, construction, drinking water and sanitation, along with the costs
of rescue operations, support and relief.
 Housing
Housing damages measured the value of homes swept away or demolished by floodwaters,
partly damaged or submerged, and those with damaged roofs, along with other private
property damage such as damage to cultural and historical structures, offices, small
industrial units, markets and commercial centres and warehouses.
 Agriculture
Agricultural losses were measured in rice growing areas, flower and vegetable growing
areas, and areas for other annual crops and perennial crops, and also included losses to
large and small livestock and poultry, agro‐chemicals and erosion of farm land and housing
land.
The importance of these categories varies among the focal areas given their different
topographical characteristics and pre‐existing levels of flood protection. The rank share of each
category in terms of total flood damages for the areas in the four MRB countries are presented
in Table 3‐4.
Table 3‐4 Observed rank share of damages in the focal areas
Categories Laos Thailand Cambodia Viet Nam

Public infrastructure 3 2 1 1
Housing 2 3 3 2
Agriculture 1 1 2 3
Source: Direct flood damage inventory
Description of rankings: Number 1 is the most important share; Number 2 is moderate share, and Number 3 is a small share.
3.9.4 Data collection and surveys
In Stage 1 of the FMMP a data collection and processing methodology was developed, which
was expected to yield the required data for the focal areas, within the constraints of availability
of secondary data, time and budgetary limitations. This data collection and processing
methodology is summarised here.

Flood damage data is collected and flood level data is measured from various administrations at
provincial, district and commune levels. At the provincial and district levels, flood damage
records were collected for as many years they existed.
Households and businesses were surveyed in order to inventory of flood vulnerability

characteristics and direct and indirect flood damages.
Focus group discussions in the project areas also provided insight into a range of issues such as
the benefits of flooding, traditional coping mechanisms and community resilience.
One of the purposes of surveys was to collect and/or validate existing information on direct
damages for an agreed reference flood; and, to the extent possible, and to document and
quantify indirect damages associated with the reference flood. For this purpose the 2006 flood
was selected as this flood was considered the ‘reference flood’ given its characteristic as
‘average flood’. The year was appropriate since the survey was conducted while the memory of
the flood was relatively fresh.
Household and business interview respondents were asked to also estimate damages they
would suffer if water levels were to reach higher levels than in 2006.
The sources of data on direct and indirect damages included interviews with officials,
households, and businesses:
 Provincial and District officials

Review and validation of existing data on casualties and direct damages caused by historic
flood/erosion events. The data consisted of extensive listings of in‐kind and monetary
damages. Damage categories included educational facilities, healthcare facilities and assets,
road and transport facilities, irrigation facilities, water supply and sanitation infrastructure,
power facilities, and communication infrastructure.
 Households
Collection of data from individual households for the reference year in which significant
flooding/erosion occurred (2006). Data collection included household characteristics,
characteristics of the recent flood(s), losses to residential structures, hypothetical damages
to the residential structures for higher floodwater levels, agricultural assets and incomes,
measures of crop losses, livestock losses, fish/shrimp losses, other non‐agricultural
household income sources (e.g., home‐based businesses, outside employment) and health
impacts.
 Local businesses (non‐home‐based)

Collection of data from individual businesses for the particular reference year in which
significant flooding/erosion occurred (2006). Data collection included business data,
characteristics of the recent flood(s), direct and indirect damages to business structures,
equipment and inventory, impact on revenues and employment, and expenditures for flood
protection.
In addition, the field work in each study area included collection of baseline vulnerability data
and focus group discussions on preparedness:
 Collection of baseline flood vulnerability data

In consultation with local district and/or commune officials and existing databases, data
were assembled to document key aspects of flood and erosion vulnerability. The data
included: demographic information on population composition and growth, number of

households, household assets, economic activities, poverty statistics, land ownership

information, land use and cropping patterns, fisheries production, animal husbandry, and
inventories of the types of structures and infrastructure in the area. In addition, the
baseline incorporated data on business activities, educational facilities and assets,
healthcare facilities and assets, road and transport facilities, irrigation facilities, water
supply and sanitation infrastructure, power facilities, and communication infrastructure.
This data feeds readily into the flood data database in addition to being useful for the
formulation of flood risk management strategies.
 Focus group discussions on community preparedness

The survey team facilitated focus groups with local leaders, community/mass organisations
and men and women living in the study area who perform different economic activities and
are of different socio‐economic status. The discussions elicited qualitative information on
the beneficial and detrimental effects of floods and erosion. Prior to commencing the
discussions, a guidebook was prepared to standardise the approach and to promote
comparisons. The focus group discussions covered the history of flooding in each area,
participants’ perceptions of ‘good’ and ‘bad’ floods, benefits of floods for paddy cultivation
and fishing, traditional coping mechanisms and strengths and weaknesses of community
preparedness and resilience to flooding.
For each of these approaches a survey tool was developed. The tools consisted of
questionnaires (for the first three approaches) and a guideline (for the focus group discussions).
For focal areas where bank erosion is the main cause of damages, a revised methodology in
order to collect data at district level for each relevant river bank section, without the need for
household or business interviews or focus group discussions. The full set of tools consists of:
 Questionnaires for a Provincial and District Level Inventory of direct flood damages/bank
erosion;
 Questionnaires for District Flood vulnerability/bank erosion baseline database;
 Questionnaires for District flood events/bank erosion;
 Questionnaires for District Indirect flood/bank erosion costs;
 Guide for Focal Group discussion;
 Household questionnaire;
 Business questionnaire; and
 National Report outline.
These tools were translated into local languages and spreadsheets for questionnaire analysis
were prepared and provided to the survey teams, along with guidelines on how to prepare a
national report.
The data from the surveys, interviews, group discussions, and field observations, combined with
the data in the governmental damage reports, serve as the basis for the damage functions and
damage‐probability curves.

Overview of data sources for damage assessment using a GIS
For the absolute approach to flood damage (see Section 3.3.4) the data requirements are lower
than for the relative approach. A Geographic Information System (GIS) is required for the
relative approach, combining spatial data with tabular data. Figure 3‐7 shows the difference
schematically for agriculture, housing and infrastructure damage and risk assessments.
Table 3‐5 offers an overview of the data sources used to determine flood damages in the
Cambodian focal areas. Maps were produced at both district and commune levels, although the
commune level data for housing damages were partly based on district‐level information
(average house values).
Table 3‐5 Data sources used for mapping damages in the Cambodian Mekong Delta.
Category Definition Source

Information on flood characteristics: Hydrology and Topography
ISIS maximum water levels for Max. water levels for different seasonal ISIS
different return periods of periods (e.g., early flood season, rainy season) hydraulic
flooding and various return periods (e.g., once every 2, model
5, 10, 25 or 100 years) (MRC)
Digital Elevation Model (DTM, Topographic data for the whole MRB MRC GIS
DEM) data
Information on land use and values
District and commune/village ‐ MRC GIS
boundaries data
Village locations and number of ‐ MRC GIS
households per village data
Paddy coverage (Japan Coverage (areas) of early floods and of MRC GIS
International Cooperation seasonal rice and rainy season rice cultivation data, JICA
Agency (JICA) land use map)
Average house values per district ‐ FMMP‐C2
surveys
Damage functions
‘Damage ‐ floodwater depth’ Relation (mathematical function) between FMMP‐C2
curves for damages to housing the local flood level (flooded areas) and the surveys
and rice harvests at the district damage on housing or rice cultures
level

Figure 3‐7
Data required
Agriculture Housing Infrastructure & relief

(annual/ perennial crops, (houses, offices, sheds, cultural (for education, health, irrigation,
livestock, poultry, fertilisers, buildings, industrial units, fisheries, transport, communi-
chemicals, erosion etc.) markets, warehouses etc.) cation, industry, drinking water/

sanitation, relief etc.)
Flooding reports: country/ province damage data simple
Flooding reports: district/ commune damage data Absolute damage ($) -

flood level graphs
Hydrological dbase: measured flood levels (also model data can be used)
‐ 91 ‐
Survey: Survey: Survey:
agriculture data/ values housing data/ values infrastructure data/ values
GIS: land use GIS: land use GIS: land use Relative damage (%) -
= agriculture = village and household data = infrastructure data flood depth graphs
Hydraulic model (ISIS) data: calculated flood levels

complex
GIS base data: DT M (DEM), administrative boundaries
The absolute and relative flood damage assessment approaches require different data sets.
May 2010
3.9.5 Derivation of damage functions
For the estimation of damages for specific floods, the following data sources and equations
need to be combined (also see Section 3.3.2 for similar information):
a) Data on land use and values;

b) Data on flood characteristics; and
c) Damage functions.
During the first stage of the study (in 2008) household survey were carried out to investigate
flooding extents and flood damages in several districts. This survey showed that for some items
(e.g., houses, roads) the maximum annual flood depth largely determines the amount of
damages. In agriculture there is a difference in damages between the ‘early flood’ crop and the
‘rainy season’ crop. Damage to the ‘early flood’ crop is mainly caused by June – July floods,
while damage to the ‘rainy season’ paddies is the result of October flooding.
Government data from the commune level and MRC survey flood damage data taken from
three Cambodian districts can be used to derive the relationships between floodwater levels
and flood damages for the three categories of infrastructure and relief, housing, and agriculture.
Figure 3‐8 shows the results for Koh Andet District.

Koh Andeth: Direct & indirect damages Infrastructure & Relief

0.8
Damage (Million USD)

damage
fit
0.6
0.4
0.2
0.0
3.0 3.5 4.0 4.5 5.0
max water level at Borey Cholsar (M+MSL)
Koh Andeth: Direct & indirect damages Housing

0.04
damage
fit
0.03
0.02
0.01
0.00
3.0 3.5 4.0 4.5 5.0
Koh Andeth: Damages Agriculture

1.60
damage
fit
1.20
0.80
0.40
0.00
3.0 3.5 4.0 4.5 5.0
Figure 3‐8 District flood damage curves for three categories of damages in Koh Andet District in Cambodia.
Damage functions for housing from the household and business survey
In addition to the damage curves derived from the districts data, damage functions were
derived using data from the household and business survey that relates damage to floodwater
depth in and around houses. This approach followed the “relative damage approach” (see

Section 3.3.4). Thus, functions include water depth and damage fraction. Figure 3‐9 shows the
relative damage function derived for housing damages for Koh Andet District.
House damage curve - Koh Andet District (Takeo)
70%
60%
0.9444x
y = 0.0351e
R2 = 0.9881
50%
Temporary
Semi-temporary
Overall
40%
Damages
30%
20%
10%
0%
0.50 1.00 1.50 2.00 2.50 3.00 3.50
Flood water depth in yard (m)
Figure 3‐9 Example of a damage function for housing, relating housing damage (as % of the total value) to
floodwater depths.
Damage functions for rice paddy cultivation
Based on information collected during the Focal Group Discussions, damage curves for rice
crops in relation to flooding depths in the districts of Koh Thom, Koh Andet and Kampong
Trabek have been estimated both for early flood season rice paddies and for rainy season
paddies. The results are presented in the graphs below, showing three levels of damage,
depending on the timing of the flood. In this case, damages were most directly dependent on
the time of year in which the flooding occurred. For other regions, damage to rice paddy
cultivation can also be dependent on the duration of the flooding.

Damage Curve for Early Flood Paddy, Cambodia
800
700
600
1 of June
500 1 of July
Damage (US$/ha)
1 of August
400
300
200
100
0
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Flooding depth (m)
Figure 3‐10 Relative flood damage curves for rice paddies, Cambodia.
3.9.6 Creation of damage maps
There are essentially two types of damage maps:

1. Damage maps based on absolute damage curves; and
2. Damage maps based on relative damage curves.
Examples of both and explanations of the differences are presented in Section 3.3.4.
Figure 3‐11 depicts the steps for creating the (absolute or relative damage curve based) maps.
The data for creating the damage maps can be combined in several ways, but no matter which
analysis is used, the (absolute or relative damage curve based) maps should be similar.
1. Damage maps based on absolute damage curves

Absolute damage curves are created with historic flood damage data, relating
floodwater depths for a specific area to damage (see Figure 3‐8). The flood level data at
the time/date of flooding can be taken from measurements at a gauge station in a
district or determined using a hydraulic model covering that same district. Absolute
damage curves are linked to the long‐term flood levels from the same station,
generating in a damage probability curve. Damage figures for certain flood return
periods can be interpolated from this curve. In the example in Figure 3‐11, the map
shows district boundaries plus the damages for specific flood return periods in the
district with the data represented on a ‘pie’ chart on one map and a bar chart on the
other.
2. Damage maps based on relative damage curves

Relative damage curves are created with damage figures related to the flood depths at
the damage locations. (For example, see Figure 3‐9 (housing) and Figure 3‐10 (rice
paddies).) The damage data may come from a field survey or may be based on
professional estimates. A land use map depicts the spatial distribution of different land
use areas (such as rice growing areas). A flood map (see Chapter 2) shows the water
depths in an area for a certain flood return period (based on hydraulic modelling

results). By combining the land use map (e.g., rice paddies, villages or roads) with the
flood map (water depths at specific flood return periods), the areas with rice paddies
and certain water depths for different flood return periods are specified. This
information can be combined with the relative damage curves, yielding a measure of
the damages from the flooding of areas like paddies. Maps like those in Figure 3‐11
shows the damages for specific flood return periods in the area with the data
represented on a ‘pie’ chart on one map and a bar chart on the other. Maps can also be
generated by shading such as those in Figure 3‐12 and Figure 3‐13.
3.9.7 Results of damage assessments
The damage assessments results for district level focal areas in Cambodia have been generated
using the absolute damage approach. Table 3‐6 shows the results for Koh Andet District for four
floods of four different probabilities. The results can also be shown on maps. Figure 3‐12 and
Figure 3‐13, for example, display the total damage and the damage per hectare for rice paddies.
Table 3‐6 Damage data for different return periods for Koh Andet District. Damage is shown in million
USD.
Probability (1/year)
Damage Category 1% 5% 10% 50%
Housing 0.05 0.04 0.03 0.00
Infrastructure and relief 1.14 0.75 0.69 0.10
Agriculture 2.15 1.58 1.48 0.35
Total 3.35 2.37 2.20 0.45
By combining the above damage assessments with the probability estimates, the overall risk can
be assessed, as will be shown in Section 4.5.
As the ‘relative damage curve’ method uses spatial information available on maps, the level of
detail that may be obtained is higher than with the ‘absolute damage curve’ method. In both
cases care needs to be taken if data from different detail levels is combined.
To create maps based on relative damage curves, there is a need to perform a series of
calculations both in a spreadsheet and in a GIS. For details on the steps and calculations in the
spreadsheet, see the spreadsheet with the necessary guidelines that is included on the CD/DVD
available at the MRCS FMMP Office in Phnom Penh.

Figure 3‐11: Steps (simpliified) to create maps based on

n absolute and relative damagge – flood depth
h curves.
ood Risk Assessm

MRC Flood
d Management and
P Com
Textb
box 2: Creatin
ng Specialised
d Damage M
Maps Based on
n Relative Da
amage Curvees
Camb
bodia, creatin
ng a housing damage
d map based on rellative damage curves
The ISIS‐based flo oodwater or basin wate r depth map ps can be co ombined witth village
locatiion maps, to generate wa ater depth innformation (b
by return perriod) for eachh village.
By taking floodwaater damage curves for ddamage to ho omes (Figuree 3‐9), combiining this
with ddistrict‐averaage home values and dataa for the num
mber of house eholds per vil lage, it is
possibble to calcullate the floo
od damage t o housing per village an nd per comm mune. By
transfferring the data from thee spreadsheett to a GIS, th
hese results can
c be displa yed on a
map.
bodia, creatin
Camb ng a rice padd
dy damage m ap based on relative damage curves
To caalculate damages to rice harvests forr different flo ooding return periods, thhe Japan
Internnational Coopperation Age
ency (JICA) la nd use map showing the rice cultivatiion areas
can be used as a basis.
b The pad
ddies that aree flooded oncce every two years are coonsidered
to be ‘early flood’ paddies, while the other paddies (on higher groun nds) are consiidered to
be ‘raainy season’ paddies. Byy combining the floodwater depth maps m (by floood return
period) with thesee rice paddy maps in a G GIS, transferriing the GIS data
d to a spreeadsheet
and bby combining the data witth the rice daamage curvess (of e.g. Julyy for ‘early floood’ rice;
see Figure 3‐9) in the spreadshheet, it is posssible to calcu
ulate the dam
mage to the rrice crop.
Thesee results can be transferre
ed to a GIS too display dam mages to the rice
r crop on a map by
districcts or commu
unes (see Figu
ure 3‐13).

Figure 3‐12 Example of a damage map based on relative damage curves, showing tootal damage to rice paddy
fields in 3 disstricts and theirr communes in Cambodia.
ood Risk Assessm

MRC Flood
d Management and
P Com
Figure 3‐13 Example of o a damage map based on reelative damage curves, showing damages to rrice paddy
fields perr hectare in 3 diistricts in Cambbodia and their communes.

CHA
APTEER 4
FLOO
OD RISSK ASSSESSMENT
4 FLOOD RISK ASSESSMEENT
4.1 In
ntroduction
In risk asseessment, the results of bboth hazard and damage e assessmentts are combined. This
chapter describes the approach
a to risk assessmment (Sectionn 4.2) and thhe several opptions for
presenting the assessment results (SSection 4.3). Section 4.4 describes
d how w the resultss of a risk
assessmentt can be used d to evaluatee the effects of measures and the levvel of accepttable risk.
The final seection demonstrates how risk is determ
mined for a ca
ase study are a for the LMBB.
4.2 G
General apprroach
Flood risk iss generally de

efined by thee simple equaation,
Rissk = Probabilitty x Damage (4..1)
For example, if an area has a probabbility of flood

ding of 1% pe
er year (or 1//100 per yearr) and the
expected damage from such floodinng is USD20 million, the risk r would eqqual USD 200 0,000 per
mber reflects the average expected economic damaage per year, it is also
year. Since this risk num
pected econoomic damage.
referred to as yearly exp
In flood risk assessment, both the pprobabilities and consequ uences of floood events haave to be
assessed. Figure
F 4‐1 sh
hows concepttually the prrocess of risk
k determinattion for a rivver and it
includes thee following stteps:
1. The starting point is informatioon on the freequency of ceertain river flfloodwater diischarges.
This information can be obttained from historical data d or proobabilistic annalysis of
hydrollogical processses.
2. Next, the relationsship betweeen the floodw water discharge and thee water (floo od) depth
shouldd be derived d. This can be done ussing a hydra aulic model or a stage‐discharge
relatio
onship. Thesee first two ssteps are desscribed in more
m detail inn Chapter 2 on flood
hazardd assessment.
3. To assess the floodd damage a rrelationship between
b the flood characcteristics and the level
od damage iss needed; i.ee., the ’damaage function’ (see Chapteer 3 on flood
of floo d damage
assessment).
4. By commbining thesee steps the reeturn period of
o a certain damage
d level can be assesssed.
Figure 4‐1 shows cleverrly, using sevveral graphs (starting in the t lower rigght hand corrner) how
each of thee steps yields a set of dataa that fits into
o the next step in an inteegrated proce
ess. In the
picture, eacch graph sharres a commonn data axis withw a graph used in the prrevious step.
In Section 4.4
4 this appro
oach is elaborrated for areaas in the LMB
B.
ood Risk Assessm

MRC Flood
d Management and
P Com
Water Water
depth h depth h
3) Damag
ge raulic
2) Hydra
assessme
ent modelinng
Dama
age D Discharrge Q
Discharrge Q
4) D
Determinatio
on of the risk
Damag
ge D
1) Probaabilistic
analysiss of river
dischargges
Return perio
od
T
Return
perio
od T
Figure 4‐1 Conceptu

ual approach for determinationn of flood risk.
4.3 Presentatio
on of the ressults of risk aassessmentss
4.3.1 Economic risk
To estim mate the flo ood risk of an area, infformation iss needed on n the probaabilities and
consequeences of floo od events that could affectt that area. For
F a full prob
babilistic ana lysis, a large
number of flood even nts should bee analysed wiith numericall models. One e useful way to do this is
with a M Monte Carlo simulation.
s The simulatio n takes existting data andd then uses t he equation
that a reegression analysis suggests for existingg data to rand
domly generaate data for mmissing data
points ussing a randomm selection. In this way, fl ood hazards and damage values can bbe estimated
for all evvents and can
n be combined to estimatee the risk.
Howeverr, rather thann looking at every

e possiblee type of floo
od, the standard approachh is to select
a limitedd number off potential flooods which are represen ntative for th
he spectrum of possible
floods. FFor example, the annual flood,
f the 1/22, 1/5, 1/10 and 1/100 (rread: once peer 100) year
events are used as the
t standard return periood or freque ency measure es I flood esttimates. For
each of tthese flood return periods, models ca n be used to o generate a hydraulic sim
mulation and
damage assessment.
Table 4‐1 Illustrative summaryy of damage annd probability calculations for a range of floodd scenarios.
Return period (year) 2 5 10 100

Probability of exceedance/floodingg
1/2 1//5 1/10 1/100
(per yeaar)
Flood daamage (USD) Damagee1/2 Damaage1/5 Dam
mage1/10 D
Damage1/100
Differentt flooding sceenarios can be defined by their return period or frequency; i.e., the average
recurrence interval between
b two floods with a certain floo
odwater leve el and the coorresponding
damage. Another meeasure often used is thee probability of ‘exceedan nce’ of an eevent with a

specified damage; i.e., the probability (per year) that a certain damage value will be exceeded.
These two metrics can be stated in terms of each other as follows:
Probability of exceedance = 1 / return period (4.2)
This means that the event with a 100 year return period has a probability of exceedance
(flooding of greater magnitude and damage) of 1/100 per year.
The relationship between the return period and the damage can be plotted in a graph (see
Figure 4‐2 – left). The longer the return period, the more extreme the event and the higher the
damages. However, even with higher return periods, damages will not increase exponentially. In
fact, the expected damages begin to reach a plateau (the curve will ‘flatten’) as the maximum
potential damage is reached and additional water adds little more to significant harm. The
relationship between the probability of flooding and a given damage value can also be displayed
graphically (see Figure 4‐2 ‐ right). Events that occur with a low probability and large damages
are shown on the left side of this graph.
The figures below are based on the assumption that damages only occur for floods that occur
(on average) once per year or with a smaller probability. However, it is also possible that
damages occur multiple times per year in the same area. In that case, a certain set of damages
could be exceeded with a frequency that is larger than once per year. For example a certain
damage value can be exceeded twice per year if such floods occur twice or more per year. In
that case the right side graph in Figure 4‐2 could display frequencies/return periods that are
larger than one. The data on damages is collected for each flooding event and the events data
note the number of such floods.
Damage ($) Damage ($)
1 2 5 10 25 100
Return period 1/100 1/10 1/5 1/2 1 Probability of
(year) exceedance
(1/year)
Figure 4‐2 Relationship between return period and damage (left) and probability of flooding and damage
(right).
Using this information, it is possible to determine the expected value of damages for an area, as
presented below. The unit of risk is measured in annual monetary damages and is referred to as
expected economic damage. Note that the risk/expected damage can also be determined for a
single unit of land7; e.g., per hectare. In such case the risk is presented in terms of expected
monetary damages per year per land area such as in USD/ha/year. The risk for a whole area is
relevant when one wants to know the absolute level of risk and consider risk reduction
measures (see also Section 4.4). The risk per hectare is of interest when one wants to compare
relative risk between areas.
7
It is preferable to estimate the damage per hectare for the actual surface of land that has been flooded.
However, the exact size of the flooded area is not always known and the damage needs to be averaged over a
larger administrative area than only the flooded area.

MRC Flood
d Management and
P Com
The risk//expected daamage can be b determineed by calculaating the are ea under thee curve that
shows th he relationsh hip between probability oof flooding and a the dam mage (Figure 4‐3). In the
figure the risk can be measured by summing a reas A, B, C, D and E. For a more preccise estimate
also the aareas ‘a’, ‘b’, ‘c’, ‘d’ and ‘e
e’ can be appproximated byy triangles an
nd added to fiind the total
risk (if th
he curve can nnot be descrribed by an eequation for use of a mo ore exact meeasure using
integral ccalculus).
Note thaat just multipplying the dam

mages from TTable 4‐1 witth their probability of floooding would
lead to d
double countting and an overestimatio
o on of the dammage. Figure e 4‐3 shows w why: by just
multiplyiing Damage1//2 with a retuurn period off ½ per year one
o would ‘e extend’ the a rea E all the
way to tthe y‐axis, overlapping
o partly
p A, B, C and D. Insstead, one should calculaate just the
additionaal risk from each
e flood (p
probability off 1/2 – probaability of 1/5)) x Damage1//2. The same
applies too the other damages/pro
d babilities.
Damage [US$/ha]
A
d
B
C
D e
E
1/100 1/25 1/10 1/5 1/2 Probabilityy [-]
Figure 4‐3 The area below the curvve, the total riskk or expected damage
d (for exaample, in USD/yyear), can be
mated by summing the areas off the rectangless, A, B, b, C, c, D,
approxim D d, E and e.
When a large numbeer of events with

w differentt return perio ods is known, the stepwisse curve will
proach a conttinuous curve
likely app e. In that casee the risk/expected value can be approoximated by
just summming the areas of blockss A, B, C, D and E and the t triangles a, b, c, d annd e can be
omitted.
The general experience is that the t events w with the higghest probabilities (or low
west return
periods) form a majo or part of thee risk. Althouugh damagess for events with
w high retturn periods
t effect is ooften largely compensated by the low probability.
(e.g. 1000 years) can be very high, this
The resu ults of risk caalculation ca

an be shownn on maps. For F example the expecteed economic
damage for different areas can be shown. Thee advantage of the use of o flood risk m
maps is that
they offfer insight in the spatia al/geographicc distributio
on of flood risks and tthus to the
contributtion of speciific areas to the overall fflood risks. This
T implies that ‘hot spoots’ with the
largest ccontribution to the overa all risk level can be easily identified. Maps can aalso make it
easier too see the relationships between floooding and the presencce or absencce of flood
protectioon works as well as poin nt to other i nsights. Exam
mples of riskk maps for t he LMB are
included in Section 4..5.
4.3.2 Risk to life
An important indicato or in risk eva

aluation and decision‐makking is the rissk to life (losss of life). To
estimatee this type of risk, mortality data and rrelated flood
d event data are required.. The loss of

life can bee estimated with the ap proaches described in Section 3.6. A Again a table can be
designed to o show the probability of flooding and
d the loss of life for severral events or scenarios
(similar to Table
T 4‐1).
The risk to life is generally referred to as societal risk. It can

n be shown iin an FN curvve, which
shows the probability
p off flooding leaading to a certain numberr of fatalities.. Both axes are
a shown
at a logarith
hmic scale, meaning
m that the differencce between two
t axis‐unitss is a factor 10.
1 Figure
8
4‐4 shows ana example ofo an FN curvee .
Prob
bability of
o
exce
eedance e (1/yr)
-3
10
-5
10
-7
10
Fatalitie
es
10
0 00
10 10
000
Figure 4‐4 Example of an

a FN curve.
In addition, the risk of flooding to the individual (death of a person) caan be calcula ated. The
individual risk due to floooding can bee compared to individual risks
r for otheer activities su
uch as the
risks of trafffic or disease. These resuults can be used
u as inputs for decisionn‐making (se ee Section
4.3 for furthher details).
4.4 R reduction measures and risk eva

Risk aluation
4.4.1 Effects of anti‐flooding andd flood prepaaration measu

ures on risk
The results of a flood risk assessmennt could be very

v o evaluate thhe effects of flood risk
useful to
education measures.
m Basically there are two categories of floo
od risk reducttion measure
es:
1. Measurres to reduce
e the probabi lity of floodin
ng;
2. Measurres to reduce
e the consequuences of floo oding.
Figure 4‐5 offers

o an overview of posssible measure es and schemmatically showws the effectss of these
measures on
o risk. It assumes that m measures to reeduce the probability willl completely eliminate
the damagee up to a prottection level with a certain probability of flooding. In other words, floods
with larger probabilitiess (smaller retturn periods) are controlled and do noot cause anyy damage.
8
A FN curve isi very similar to the curve thaat shows the probability of excceedance resultting in a certainn damage
value. The diffference is that the damage vaalue (in this casse loss of life) is shown on the horizontal axis of the FN
curve.
ood Risk Assessm

MRC Flood
d Management and
P Com
For exammple, this is the case when n dykes or emmbankmentss are built to provide prottection to an
area up tto a certain leevel; e.g., for a major floo d of 50 year return periodd.
Measurees to reduce the t conseque ences of floooding such ass adaptation of land use ccontrols and
building improvemen nts, reduce th he consequennces of an eve ent (total dam
mages) for a ggiven return
period. M
Measures can n be combined, so there w will be a combbined reduction in risks.
Note thaat while compensation an nd insurance are also ofteen mentioned d as measurees to reduce
the conssequences, th hese measure
es do not redduce the totaal damages. They
T are onlyy a financial
reallocattion of the flo
ood damages costs.
Measurres to reduce flooding prob bability Measure es to reduce flooding

f conseequences
‐ Strenggthening of dyykes and emba ankments ‐ Land usse planning
‐ River w
widening (“roo om for rivers””) ‐ Seconddary flood defeences
‐ Retenttion and emerrgency storage ‐ Flood proof
p buildings (e.g. raising buildings
above ground levelss)
Damag
ge ($) Damage
e ($)
1
1/100 1/10 1/5 1/2 1//100 1/10 1/5 1/2
Probability of
P Pro
obability of
exceedance excceedance
(1/year) (
(1/year)
Figure 4‐5 Effects off measures on flood

f risk.
4.4.2 Basics of co
ost‐benefit an
nalysis
Cost‐ben
nefit analysiss is used to evaluate whhether a cerrtain measurre achieves eenough risk
reduction
n. A project should be undertaken if it results in an increase of economicc welfare. In
other woords, the benefits generatted by the pr oject should exceed its coosts. Further information
on cost ‐ benefit analysis in the context
c of floood managem ment is given in the workk by Penning
Rowsell aand Chattertton (1977) orr in the guideelines for cost‐benefit ana
alysis (CBA) ddeveloped in
the UK (M
MAFF, 1999, 2000) and Au ustralia (BTREE, 2001).
Investmeent projects can be sysstemically annalysed with h individual effects estim mated and,
whereveer possible, given
g a mone etary value. Cost ‐ benefit analysis provides
p an ooverview of
distributional effects and uncertaainties whilee pointing to alternativess. In theory, all relevant
impacts, including inntangible onees, must be taken into account
a in a competent cost‐benefit
analysis. Too often, however,
h in practice,
p the analysis is often
o narrowe ed to tangiblle monetary
effects.
In such a ‘limited’ co
ost ‐ benefit analysis,
a the measurable economic be enefits of ann activity are
compareed with the measurable
m costs
c of the aactivity. If the benefits arre higher thaan the costs,
the activvity is consid
dered attracctive (it geneerates an increase in ecconomic bennefit). If the
benefits are lower than
t the coosts, the act ivity is not attractive. In the conteext of flood
managem ment this means
m that costs of i ncreasing saafety againsst flooding (e.g., dyke
strengtheening of flood proofing)) are compaared with th he decrease in the yearrly expected
economic flood damaages to determine whethher an investment is attra active. In the cost figure,
differentt types of costs have to be
b included: costs of inve estment (fixe
ed and variabble) and the

costs of maintenance and management. The benefit measure is the reduction of the expected
economic damage.
Restating this in terms of an equation, a flood risk reduction investment is attractive in the
terms of a cost ‐ benefit analysis if:
I  E (D ) (4.3)
where: I = investment in project/measure [e.g in USD/yr]

ΔE(D) = reduction of expected economic damages [e.g. in USD/yr]
The Greek letter delta here is used to mean ‘the change’ but of course the change in expected
economic damages here is best described as the reduction in the damages since the
investments are designed to result in changes that reduce damages.
The reduction of the risk/expected economic damages can be calculated with a simple equation
looking at the before and after changes:
E ( D )  E ( D ) 0  E ( D ) N (4.4)
where: E(D)0 = expected economic damage in the original situation [e.g. in USD/yr]
E(D)N = expected economic damage after measures/project completion [e.g. in
USD /yr]
The expected economic damage can be determined through the approaches outlined in Section
4.2.
The change in the expected damage value due to different types of risk reduction measures can
be determined according to the principles presented in Section 4.3.1.
The above approach works if both the investments and the benefits can be expressed in
monetary units for the same unit of time (e.g., USD/yr). In many cases, investments in flood risk
reduction are one‐time costs where the benefits are expected to continue over a long period,
such as for a dyke construction project, whereas the benefits are expressed on an annual basis.
To conduct a cost benefit, the time element needs to be standardised for both the investment
costs and the benefits. The way to standardise the benefits is to calculate their net present
value. Any costs and benefits that occur over future time periods need to be discounted to
determine their net present value. The following general formula (as presented in Appendix 3)
can be used to do that:
n
Bi  C i
NPV  
1 (1  r ' ) t (4.5)
where: NPV = net present value [e.g. USD]

Bi = benefits in year i [e.g. USD]
Ci = costs in year I [e.g. USD]
t= time [year]
r’ = reduced interest rate [‐]
The reduced interest rate r’ is generally used for discounting. It is equal to the rate of economic
growth minus the rate of inflation. When one assumes an infinite time horizon the discount
factor approximates 1/r’ and the net present value of the benefits becomes ΔE(D)/r’.

MRC Flood
d Management and
P Com
4.4.3 Economic optimisation

o
A specific form of thee cost ‐ benefit analysis iss economic optimisation.

o This approacch is used to
determinne an optimaal level of pro
otection whe n both the in nvestments in n protection and the risk
are depeendent on th he level of protection. Thiis is the case
e, for example, for flood prone areas
that are protected byy dykes. The choice of prrotection leve el is dependeent on the hheight of the
dyke.
The Deltaa Committeee in the Netheerlands has aapplied econo

omic optimisa ation to floodd protection
systems. The Commiittee began its i work to i nvestigate possibilities
p fo
or new safetty standards
followingg a major floo
od that cause
ed enormous damage in th he Netherlands in 1953.
In econoomic optimisaation, the inccremental in vestments in n more safety are balancced with the
reductionn of the risk. In the example of the Neetherlands, th he investmennts consist of the costs to
strengtheen and raise the dykes. This simple appproach assumes that floo oding can onlly occur due
to floodwwaters exceeeding the floo od defences. Thus, each dyke d height corresponds
c to a certain
probabiliity of floodin
ng (the highe er the dykess the smallerr the probability of floodding) and an
associateed damage. ByB summing the t costs andd the expecte ed damage or risk, the tottal costs can
be plotteed as a functiion of the saffety level. A ppoint can be determined where
w the tot
otal costs are
minimal. This is conssidered the optimal poinnt. (See Figure 4‐6.) At this t optimum m point, the
correspoonding dyke height
h is know
wn. Because tthe statistics of the waterr levels are allso known, a
correspoonding protecction level in terms of a prrobability can n be defined.
This approach was applied after the Dutch 19953 storm su urge to determine an opptimal safety
level forr the largest flood prone e area, Southh Holland (vvan Dantzig, 1956). Eventtually these
results h have been used to derive safety standards for flood de efences throoughout the
Netherlaands. For coastal areas, thhe optimal leevel of protection has been chosen w with flooding
probabiliities betweenn 1/4,000 per year and 1//10,000 per year.
y For Duttch river basiin areas, the
safety staandards weree set at 1/1,2
250 per year aand 1/2,000 per
p year.
Hig
gh probability Small probability
p
Figure 4‐6 Principle of economic op

ptimisation.

Economic optimisation
o can also be applied to other
o types of
o problems of various sccales. For
example, thhe optimal level of raisingg a house to minimise floo od damages ccan be deterrmined by
taking into account the additional coosts of raisinng the buildin
ng, the damaage to the strructure in
case of floo
oding and the
e probabilitiess of several water
w levels th
hat would leaad to floodingg.
4.4.4 R valuation for loss of liffe

Risk
The estimattes of risk to life can be i ncluded in risk valuation and decisionn‐making. It iss possible
to define ceertain standaards that indiicate the acceptable level of individuaal or societal risk. (See
Section 4.3 3.2 for definiitions). Moree informationn on this topic is found in several texts
t (see
Vrijling et al.,
a 1998; Jonkkman et al., 22009), but a summary
s of the main conccepts is given
n below.
For individu ual risk, a threshold leve l can be set to indicate thet tolerablee risk due to a certain
activity, succh as a probability of deatth of 10‐5 per year. If the actual
a calcula ted risk is higgher (e.g.,
10‐4 per yeaar) the situatiion is not accceptable and risk reduction measures aare necessaryy.
For societall risk, a limit can

c be deter mined througgh use of a limit line in thee FN curve (ssee Figure
4‐7). The lim
mit line sets the
t acceptab le number off fatalities versus the probbability of a harm
h such
that any nuumber of fatalities beyondd this are considered unaccceptable. In oother words,, an event
with larger consequences is acceptaable only witth a smaller probability.
p I n planning protection
p
measures, the
t calculated risk can bee compared with w the limit line. If the caalculated riskk is below
the limit line, the situa ation is acceeptable. If th
he calculated
d risk is aboove the limitt line the
situation is not acceptab ble and addit ional measurres are needeed.
Acceptable societal riskk Unacceptablle societal rissk

Probability of Probability of
ance (1/yr)
exceeda exceedance (1/yr)
10
-3 Limit line
e 10
-3 Limit line
Calcullated
Callculated
-5 -5 risk
10 risk
k 10
-7 -7
10 10
Fattalities Fataliities
10 100 1000 10 100 1000
Figure 4‐7 Societal risk of fatalities duee to flooding an

nd the limit line
e of acceptable harm.
The above concepts fo or limits for individual an

nd societal risk
r are curreently explore ed in the
Netherlands where new w safety stanndards are de
eveloped thaat meet socieety’s acceptable limits
for risk to life. At pressent, the Neetherlands ap
ppears to bee the only coountry that uses this
approach too safety standdard for floodding.
4.4.5 R matrix
Risk
A tool used to evaluate risk is the ris k matrix (see

e below). It prrovides a ran king of the probability
(vertical axis) and expected damagees (horizontal axis). Based d on the maggnitude of probability
and expectted damagess, a certain rrisk can be rankedr as accceptable (shhown on the e chart in
green), unaacceptable (re
ed) or subjectt to further discussion
d on its acceptabiility (orange) based on
subjective determinatio ons regardingg the risks. In I order to categorise
c thhe damages, different
damage categories (eco onomic damaage, loss of life, environmental or soocial impactss) can be
taken into account.
a
ood Risk Assessm

MRC Flood
d Management and
P Com
The choice of the boundaries between the cattegories is an n arbitrary on

ne. The outcoomes of risk
assessmeent can only provide inforrmation on thhe level of rissk and not onn the accepta bility, which
is a subjeective determ
mination.
H
High
Probability
MMedium
LLow
Low Mediium High
H
Consequuences
Figure 4‐8 Structuree of a risk matrixx.
4.5 Risk assesssment for the LMB
4.5.1 Introduction and approa

ach
This secttion presentss the results of risk assesssment within n the contextt of the MRBB; combining
od hazard analyses (Chappter 2) and daamage assesssments (Chappter 3) from
the resullts of the floo
the areass under studyy. Note that benefits
b of floooding are no
ot discussed in
i detail heree, but can be
relevant in an overall risk valuation.
Figure 4‐‐9 outlines thhe approach that was fo llowed in this project to estimate rissk. From the
time series output off the ISIS hyddraulic modeel (the graph in the uppe
er left of the figure), the
probabiliity of different water levels can be a nalysed and presented graphically (shhown in the
graph in the upper right). The relationship bbetween floo odwater leveels (or flood depth) and
damagess (shown graaphically in the lower l eft) is also known. By combining tthe damage
relationsship with thee flood probbability data,, the damagge probabilityy curve can be derived
(shown in the graph in the lower right
r of the figgure).
Hydrolo
ogical data Flood depth - Probability
3.5
8.5
Flood / water level (m + MSL)
8
3
7.5
Flood / water depth
(m + field level)
7
2.5
6.5
6
2
5.5
5
1.5
4.5
4
1
1910
1914
1918
1922
1926
1930
1934
1938
1942
1946
1950
1954
1958
1962
1966
1970
1974
1978
1982
1986
1990
1994
1998
2002
2006
1/100
1/50
1/10
1/5
1/2
Year P
Probability (-)
Dam
mage Risk
70 2000
1800
Damage (% of house value)
60
1600
50
Damage ($ / ha)
1400
1200
40
1000
30
800
20 600
400
10
200
0 0
0 0.5 1 1.5 2 2.5 3 3.5
1/100
1/50
1/10
1/5
1/2
Flood / w ater depth (m + field level) P

Probability (-)
Figure 4‐9 Approach

h for estimation
n of the risk.

The resultss obtained ussing this appproach for th

hree districts in Cambodi a designated d as focal
areas in thee FMMP Stage 1 are illustrrated below. These are the districts of Koh Andet, Koh
K Thom
and Kampo ong Trabek.
4.5.2 R
Results of risk assessment
As part of the
t risk assesssment the flooods that occurred betwe een the yearss 1910 and 2006 were
simulated. The damage curves from m Section 3.8 were used to estimate t he damages for these
flood even nts. By combining thesee results, th he following g probabilityy damage cu urve was
generated. From the figgure it can bbe seen that Koh Andet has h the higheest damages for most
return perioods; especially for eventss with a relattively large probability of flooding (40% ‐ 60%).
The Koh Tum district hass the lowest ooverall risk cu
urve.
Figure 4‐10 Probability damage

d curve foor three Cambo
odian districts.
The risk (annual expected value off the damagges) for the three districtts can be de erived by
estimating the area under the probbability damage curve. (S See explanattion in Sectio
on 4.3.1).
hows both th
Table 4‐2 sh he absolute exxpected value and the exp pected value per square kilometre.
k
From thesee results it beccomes clear tthat Koh And
det has the highest risk.
Table 4‐2 Results of

o risk calculatioons for the thre
ee focal areas in Cambodia.
Expected eeconomic dam mages Exp

pected econoomic damagess per ha
District
(mill ion USD/yr) USD/(km2*yr))
(million U
Koh Andet 0.77 00.0022
Koh Thom 0.42 00.0009
Kampong Trabek
T 0.69 00.0014
The results of the risk caalculations caan also be vissualised on maps.

m Figure 44‐11 shows thhe results
for the averrage flood rissk calculationns for the focal areas in Ca
ambodia andd Viet Nam. The size of
each circle shows the magnitude
m oof the risk fo
or the districtt. The distribbution over the
t three
ood Risk Assessm

MRC Flood
d Management and
P Com
damage categories (h housing, infra

astructure andd relief, agricculture) is sho
own within eeach circle. It
hat the distribution over the
is clear th t damage ccategories vaaries between n districts. Foor Koh Andet
and Kam mpong Trabekk, the largestt part of the risk is to aggriculture. For Koh Thom,, the largest
share of the risk is to infrastructurre.
Figure 4‐11 Flood riskks by district forr selected areass in Cambodia and
a Viet Nam.
The figurre shows cleaarly that the risks on the Vietnamese side are much higher. Thhis is due to
the fact tthat the Vietn
namese areas are more deensely populated.

CHA
APTEER 5
REFFEREN
NCES
5 R
REFERENCES
S
[1] Abt S.R

R., Wittler R.JJ., Taylor A.,, Love D.J. (1989) Human stability inn a high floo
od hazard
zone,Waterr Resources Bulletin
B Vol. 225 No.4, pp. 881‐890.
8
[2] Adamsoon, P.T. (2007). Strengthenning of the Floood Managem ment capacitty for the Kokk River
Basin in Chiiang Rai Province, Thailan d. Hydrologiccal Baseline Study.
S Inform
mation and Knnowledge
Managemeent Programm me, MRC, Sepptember.
[3] Ahern M.,

M Kovats S.R., Wilkinso n P., Few R., Matthies F. (2005) Glo bal health im
mpacts of
floods: epid
demiologic evvidence, Epiddemiologic Reeviews Vol. 27
7 pp. 36‐46.
mous (2009), Development

[4] Anonym D t and use of DACA
D in de lo
ower Mekongg basin – execcutive
summary.
[5] Bennet,, G.(1970) Bristol Floodss 1968 – Con ntrolled Survvey of Effectts on Health of Local
Communityy Disaster, Briitish Medical Journal Vol. 3, pp.454‐458.
[6] BTRE (2 2001). Econo

omic Costs oof Natural Disasters
D in Australia. Reeport 103, Bureau
B of
Transport anda Regionaal Economicss, Disaster Mitigation
M esearch Worrking Group, Paragon
Re
Printers. Caanberra: Commonwealth oof Australia.
[7] Clausen, L.K. (1989) Potential Damm Failure: Esstimation of Consequences

C s, and Implications for
Planning. Unpublished
U M.Phil. thesiis at the Schhool of Geog graphy and PPlanning at Middlesex
M
Polytechnicc collaborating with Binniee and Partnerrs, Redhill.
[8] Dutta, D.,

D Herath S., Musiake K. (22003). A mathematical flo
ood loss estim
mation, Journal of
Hydrology, 277:24 –49.
2003). HAZUSS: Multi‐hazarrd loss estimaation model methodology

[9] FEMA (2 m y – flood mod
del.
Federal Emergency Mannagement Ageency, Washin ngton.
[10] Hajat S., Ebi K.L., Kovats S., M

Menne B., Edw
wards S., Haines A. (20003) The huma an health
consequencces of flooding in Europee and the implications for public heal th, Journal of
o Applied
Environmenntal Science and
a Public Heealth Vol 1 No
o. 1 pp. 13‐21
1.
[11] Hosking, J.R.M. and

d J.R. Wallis (11997). Region
nal frequencyy analysis: an approach ba
ased on L‐
moments. Cambridge
C University Presss, Cambridge e, U.K.
[12] Jonkmaan S.N. (2007

7) Loss of life estimation in
n flood risk asssessment, P hD thesis, De
elft
University.
[13] Jonkman S.N., Bockarjova M., KKok M., Bernnardini P. (20

008) Integrateed Hydrodynnamic and
Economic Modelling
M of Flood
F Damagge in the Neth
herlands, Ecological Econoomics 66, pp. 77‐90.
[14] Jonkman S.N., Penning‐Rowselll E. (2008) Human instab

bility in floodd flows, Journ
nal of the
American Water
W Resources Associatioon (JAWRA) Vol.
V 44 No. 4 August 10088, pp. 1‐11.
[15] Jonkman S.N., Vrijliing J.K., Kok M. (2008) Flo ood risk assessment in th e Netherland
ds: A case
study for dyyke ring South Holland, Risisk Analysis Vol.
V 28, No. 5, pp.1357‐13773.
nen R.A., Hepojoki A., Huhhta H.K., Louh

[16] Karvon hio A. (2000) The use of phhysical modells in dam‐
break analyysis, RESCDAM
M Final Reporrt, Helsinki University of Technology,
T H
Helsinki, Finla
and.
ood Risk Assessm

[17] Karvonen R.A., Hepojoki A., Huhta H.K., Louhio A. (2000) The use of physical models in dam‐
break analysis, RESCDAM Final Report, Helsinki University of Technology, Helsinki, Finland.
[18] Kelman I. (2002) Physical vulnerability of residential properties in coastal, eastern England,
PhD dissertation, Cambridge University.
[19] Kelman, I. and R. Spence (2004) An Overview of Flood Actions on Buildings. Engineering
Geology 73. pp. 297–309.
[20] Kok M., Huizinga H.J., Vrouwenvelder A.C.W.M., van den Braak W.E.W. (2005)
Standaardmethode2005 schade en slachtoffers als gevolg van overstromingen, HKV report
PR999.10 ‐ draft translation is available as: Standard method for Predicting Damage and
Casualties as a Result of Floods.
[21] Leenders, J.K., Roelevink, A., Parodi, G. and J. Wagemaker (2009a). Development and
Demonstration of DACA (Damage and Casualties Assessment) in the Chiang Rai Region, Lelystad:
HKV Consultants and ITC.
[22] MAFF (1999). Flood and Coastal Project Appraisal Guidance. Economic Appraisal (eds. K.
Riddell and C. Green). London, UK: Ministry of Agriculture, Fisheries and Food, Flood and
Coastal Defense Emergencies Division.
[23] MAFF (2000). National Appraisal of Assets from Flooding and Coastal Erosion. Halcrow
Maritime, HR Wallingford, Flood Hazard Research Centre for the Ministry of Agriculture,
Fisheries and Food, Flood and Coastal Defence Emergencies Division, London, UK.
[24] Merz B., Kreibich H., Thieken A., Schmidtke R. (2004) Estimation uncertainty of direct
monetary flood damage to buildings, Natural Hazards and earth System Sciences Vol. 4, pp. 153
– 163
[25] Messner, F., Meyer, V. (2005): Flood damage, vulnerability and risk perceptions; HKV:
Implementation of DACA.
[26] Messner, F., Penning‐Rowsell, E., Green, C., Meyer, V., Tunstall, S., Veen, A. van der (2007).
Evaluating Flood Damages: Guidance and Recommendations on Principles and Methods.
FLOODsite Consortium, Wallingford, UK.
[27] Parker D.J., Green C.H., Thompson P.M.,(1987) Urban flood protection benefits – a project
appraisal guide. Gower Technical press.
[28] Penning‐Rowsell, E.C., Chatterton, J.B., (1977). The benefits of flood alleviation – a manual
of assessment techniques, Saxon House, ISBN 0566001908.
[29] Roos, W. (2003) Damage to buildings. Delft Cluster report DC1‐233‐9.
[30] Royal Haskoning (2008) Flood damages and flood risks in focal areas. Annex 2 to the draft
Stage 1 evaluation report, (August 2008).
[31] Stuyt L.C.P.M., Reinders J.E.A., van der Hoek E.E., Hermans A.G.M., Muinck Keizer de M.,
Kamperveen J.P., Hoogvliet M.C., Gerritsen A.L., Hounjet M.W.A., Bontenbal M.P., Hoefsloot
F.V.J., Icke J. (2003) Environmental impact of flooding, Delft Cluster Report DC1‐233‐8.

[32] Thieken, A., Muller, M., Kreibich, H., Merz, B. (2005). Flood damage and influencing factors:
new insights from the August 2002 flood in Germany, Water Resources Research, 41:1‐16.
USBR (United States Bureau of Reclamation) (2001) Downstream hazard classification
guidelines. Acer Technical Memorandum no. 11.
[33] Van Dantzig D. (1956) Economic decision problems for flood prevention, Econometrica Vol.
24 pp. 276‐287.
[34] Van der Veen A., Steenge A.E., Bockarjova M., Logtmeijer C.J.J. (2003) Structural economic
effects of large scale inundation: a simulation of the Krimpen breakage, Delft Cluster report
DC1‐233‐12.
[35] Viessman, W., G.L. Lewis and J.W. Knapp (1989). Introduction to Hydrology. Third edition,
Harper Collins Publishers.
[36] Vrijling J.K., van Gelder P.H.A.J.M. (2002) Probabilistic design in hydraulic engineering. TU
Delft lecture note WA5310.
[37] Vrijling, J.K., Hengel, W. van, Houben, R.J. (1998) Acceptable risk as a basis for design,
Reliability Engineering and System Safety Vol. 59 pp. 141‐150.
[38] Watkins, L.H. and D. Fiddes (1984). Highway and urban hydrology in the tropics. Pentech
Press Ltd, London.

AP
PPE
END
DICE
ES
MRC Flood Management and Mitigation
M Program
mme Componentt 2: Structural Me
easures and Floodd Proofing
Appendix 1 Approa
ach to Flood Damage and Risk Assesssment in Euurope
This Appendix presents the Europeann Union’s Flood Directive along with p reliminary studies and
guidelines for
f its implem
mentation prooduced by thhe European Exchange Cirrcle on Flood Mapping
(EXCIMAP) on behalf of several Europpean countries, the FLOODsite researcch project thaat directly
follows thee EU Directivve, and the Dutch Damaage and Casualties Assesssment (DAC CA) as an
introduction to the management of fflood damage e and flood risk in Europe..
In 2007, thee European Parliament

P annd the Council of the European Union ppromulgated Directive
2007/60/EC C9 on the asse
essment and managemen nt of flood risks. This direcctive compelled the 27
member sttates of the European U Union to (preeliminarily) assess their fflood risks (b
by 2011),
prepare flo
ood hazard and flood riskks maps (by 2013), and to prepare floood risk man nagement
plans (by 20015).
The Flood Directive

D intrroduces a num mber of key principles that may be reelevant and applicable
a
to other reggions or riverr basins. Thesse principles include
i the fo
ollowing:
 Memb ber States sha all, for each riiver basin or portion of an
n internationaal river sub‐b basin lying
within their territoory, undertakke a prelimin nary flood rissk assessmennt along with h damage
estimaates of the potential
p adveerse consequ uences of futture floods ffor human he ealth, the
enviroonment, cultu ural heritage and econom mic activity. These shouldd consider as a best as
possibble issues succh as the to pography, th he position ofo watercoursses and theiir general
hydrollogical and geo‐morphoological charracteristics including flooodplains ass natural
retention areas, the effecti veness of existing human develooped flood defence
infrasttructure, the position of populated areas,a areas of economicc activity and d planned
long‐teerm develop pments, andd estimates of the imp pacts of clim mate change e on the
occurrrence of flood ds.
 In the case of an innternational rriver basin orr sub‐basin which
w is shareed with otherr Member
States,, Member Sttates shall eensure that exchange
e of relevant infformation takes place
between the conce erned compettent authoritties.
 Flood risk managem ment plans sshall take into o account co osts and beneefits, flood exxtent and
flood conveyance
c routes
r and arreas which have the pote ential to retaiin flood wate er such as
naturaal floodplains, environm mental objecctives, soil and a water managementt, spatial
planning, land use, nature conseervation, navvigation and port p infrastruucture.
 Flood risk managem ment plans sshall address all aspects of o flood risk m managementt focusing
on preevention, pro otection, and preparedness; including flood forecaasts and earlyy warning
systemms. They shou uld also take into accountt the characte eristics of thee particular river basin
or sub‐basin.
 Flood risk manage ement plans may also in nclude the promotion of sustainable land use
practicces, improvement of watter retention, as well as the controlleed flooding of o certain
areas in
i the case off a flood evennt.
 In the interests of solidarity,
s floood risk management plans establishedd in one Mem mber State
shall not
n include measures
m whiich, by their extent and impact, signifficantly incre ease flood
risks upstream
u or downstream m of other co ountries in thhe same riveer basin or sub‐basin,
s
unless these meassures have bbeen coordinated and an agreed soluution has been found
amongg the Membe er States conccerned.
 Memb ber States shall make avaailable to the e public the preliminary
p fflood risk asssessment,
the flo
ood hazard maps, the floo d risk maps and a the flood risk manageement plans.
 Memb ber States sha all encouragee active involvvement of interested partties in the production,
revieww and updatin ng of the floo d risk managgement plans.
9
http://eur‐leex.europa.eu/LLexUriServ/LexU
UriServ.do?uri=
=CELEX:32007L0
0060:EN:NOT
Guidelines Floo
od Risk Assessment App. 1 ‐ 1 May 2010
EXCIMAP
In order to carry out the tasks set by this directive, a large number of European countries
organised themselves into the European Exchange Circle on Flood Mapping (EXCIMAP). In 2007
they produced the Handbook on Good Practice for Flood Mapping in Europe10, which focuses on
the initial steps required by the Directive.
Flood maps are used to indicate the geographical areas which could be covered by flood waters
from all natural sources at several levels of probability. The Guideline distinguishes between: (i)
river flooding in floodplains; (ii) sea water flooding; (iii) mountain torrent activity or rapid run‐
off from hills; (vi) flash floods in Mediterranean ephemeral water courses (‘wadi’s’); (v)
groundwater flooding; and (vi) lake flooding.
EXCIMAP flood maps can show flood extent according to: (i) probability classifications; (ii)
historic events; (iii) flood depths; (iv) flow velocities; (v) flood propagation; and (vi) degree of
danger. These maps can be used for: (i) land use planning and management; (ii) water shed
management; (iii) hazard assessment; (iv) emergency planning and management; (v) planning of
technical measures; and for (vi) overall public awareness.
For flood risk management, flood maps are useful tools in answering questions such as: (i)
where is the greatest risk? (ii) where should investments be targeted? (iii) which investments
may have the highest cost effectiveness?
Flood hazard maps and related maps
EXCIMAP directs that flood hazard maps be prepared for low, medium and high probabilities of
flood occurrence. For each probability the extent, inundation depth/floodwater level, and
where appropriate the water velocity are indicated. These maps are to be prepared using
hydraulic models, statistical analysis and observations.
Flood danger maps indicate a degree of danger by combining information on probability level,
flood depth, velocity and debris content. These maps are not required by the Flood Directive,
but are used in some member countries for land use planning.
Flood event maps record historical flooding and can be seen as a first approach to flood hazard
mapping. The can be used in awareness raining, for follow‐up of flood hazard assessment, for
the calibration of models, and as an emergency management and planning tool.
Flood vulnerability maps
Flood vulnerability maps show the categories and assets at risk: (i) population, (ii) assets and
economic activity, and (iii) potentially affected installations that may release pollution. In the
Netherlands, these maps are prepared at the provincial level and are available to the public and
specialised services11.
10
http://ec.europa.eu/environment/water/flood_risk/flood_atlas/index.htm
11
www.risicokaart.nl
Guidelines Flood Risk Assessment App. 1 ‐ 2 May 2010

Flood risk maps
The EU Flood Directive defines ‘flood risk’ as the combination of the probability of a flood event
and of the potential adverse consequences for human health, the environment, cultural
heritage and economic activity associated with a flood event.
The basis of flood risk maps, therefore, is to indicate: (i) the number of inhabitants potentially
affected, (ii) the type of economic activities potentially affected, (iii) installations which might
create incidental pollution, and (iv) other pertinent information, such as high content of
transported sediments and debris and other significant sources of pollution.
Risk may be calculated as:
Risk = C * Ph
where: C = potential adverse consequence and

Ph = the probability of the hazardous process
Risk is expressed as a potential loss in a particular area or unit area (ha) within a given time
period (one year) using a second equation:
C=V*S*E
where: V, S and E are called the ‘vulnerability parameters’ for damage measurement:
V= value of element at risk (in monetary terms or human life)
S= susceptibility: damage effect on element at risk (as a function of
magnitude of hazard (e.g., floodwater depth‐damage and damage‐
duration relationship curves (with each factor ranging from 0 to 1)
E= exposure: the probability of the element at risk to be present while the
event occurs (factor ranging from 0 to 1)
FLOODsite12 was a research project on Flood Risk Analysis and Management Methodologies,
designed by the EU to support its Member States in complying with the requirements of the EU
Flood Directive, collecting data during 2004‐09. FLOODsite research analyses the physical,
environmental, ecological and socio‐economic impacts of floods from rivers, estuaries and the
sea. It considers flood risk as a combination of hazard sources, pathways and the consequences
of flooding on the “receptors” – people, property and the environment. The FLOODsite research
consortium included 37 of Europe’s leading institutes and universities and the project hired
managers, researchers and practitioners from a range of government, commercial and research
organisations, specialising in aspects of flood risk management. FLOODsite research is currently
applied in EU Member States in the form of models and technical support for river, estuary and
coastal sites in Belgium, the Czech Republic, France, Germany, Hungary, Italy, the Netherlands,
Spain, and the UK.
FLOODsite also prepared a discussion paper13 reviewing the flood damage evaluation methods
applied in the UK, the Netherlands, the Czech Republic and selected German States, followed by
a guideline for flood damage assessment14.
12
www.floodsite.net
13
Meyer, V.(2005): National flood damage evaluation methods
14
Messner, F. et al. (2007): Evaluating flood damages: guidance and recommendations on principles and
methods

Dutch Damage and Casualties Assessment Tool15
In the Netherlands, (HIS‐SSM), a standard method for flood damage and casualties assessment
has been used since its development in 1999. It is used to assess the change of risk following
national and regional decisions on water management measures in the Netherlands. Since
2007, regulations require that all Dutch water‐projects of national importance be evaluated
using this tool. It is also used in the production of the Dutch flood vulnerability maps mentioned
in Section 3.7.
The Damage and Casualties Assessment Tool (DACA) is a GIS‐based application in which land
use, inundation depth‐damage functions per land use and maximum damage values of each
land use are combined with a model‐generated flood hazard. This is in line with the ‘relative
approach’ described in Section 3.3.4.
These approaches are an advance on traditional flood measures and preparations. Traditionally,
flood planning focused on safety standards, such as dyke design levels or reservoir volumes
required to ensure predefined protection levels. Protection of the community against floods
occurring with a frequency of once in 1,250 years is an example of such an established level
under the flood protection law of the Netherlands. What is missing in this approach is the
amount of damage protection that the system offers. In other words, the approach disregarded
the efficiency of flood protection measures. While economic costs of alternative flood defence
options are usually considered in the decision‐making process, the benefits of flood protection
in the form of prevented damages should be taken into account too.
The new paradigm for flood risk management includes the economic analysis of costs and
benefits of flood protection and mitigation measures. Not only the safety of a defence system
and its associated costs are considered, but also the damages to be expected in case of its
failure. As a consequence of the application of cost‐benefit and risk analysis, safety standards
can better be adjusted to the specific circumstances, given that the costs of ensuring an overall
safety standard often considerably exceed the benefits in some areas.
15
Based on: Messner, F. and Meyer, V. (2005)

M Program
Appendix 2 Approa
aches for thee Assessmen
nt of Structural Damage to Buildingss
Flow velociity is an impo ortant factorr in damage to building since

s it impaccts the hydro odynamic
loads. Seveeral models have
h been deeveloped that relate the physical dam mage to housses to the
combinatio on of flood de epth and flow w velocity. As
A an example Figure A 1 presents the e damage
criterion fro
om brick andd masonry hoouses (Clausen, 1989). In combination
c with input frrom flood
simulationss, such criteria can be useed to develop p flood risk maps
m that exhhibit differentt levels of
expected damage.
d Otheer researche rs have inve estigated building vulneraability. Kelma an (2002)
and Kelman n and Spence (2004) inveestigated physical vulnerability of ressidential properties in
coastal areas and adde ed the influe nce of rise rate.
r Roos (22003) developped a comprehensive
probabilistic model for collapse of bbuildings, wh hich takes intto account tw wo failure mo odes that
can lead to partial or full collapse of a building. Th
hese are 1) thhe scour of fooundations, and
a 2) the
failure of walls.
w The moddel of Roos coonsiders hydrostatic and hydrodynam ic loads, as well w as the
impacts of waves and pounding deebris. Also the strength (rresistance) oof the buildin ng type is
considered. A general weakness
w of these ‘collappse of buildin
ngs’ models cconcerns the eir limited
practical vaalidation. Moost of the preesented relationships are e based on ttheory or sca ale model
tests.
Figure A 1 Damage crite erion for brick aand masonry houses as propo
osed by Clausenn (figure from Karvonen
K
et al., 2000)).
In general the
t advantage of the abovve approach is that the acctual flood coonditions (floo od depth,
flow velocitty) are taken into accountt. A related diisadvantage is that a lot oof detailed (GIS) data is
needed to relate
r the loccal flood condditions to the vulnerabilityy of an individdual building.
The degreee of damage will also be hhighly depen ndent on the type of stru cture that is affected.
For examplle wooden houses
h have different dam mage pattern ns than conccrete or brick houses.
Measures and
a constructtion techniquues, e.g. build ding houses on
o poles, will also affect th
he degree
of structuraal damage. This
T is why e xisting mode els and structtural damagee relationshipps should
not be directly transferrred to other regions or to
o use with diffferent housiing types witthout first
conducting a detailed asssessment of building type es and their vulnerability.
v
Guidelines Floo
M Program
Appendix 3 Fundam
mental issuees in Econom
mic Valuation
n
There are different rationales for eeconomic valuation. Financial valuatioons considerr damage
from a persspective of a single personn or firm, negglecting public affairs and focusing on the
t actual
financial bu urden that individual
i orr firm has to
o bear. Econ nomic valuattions have a broader
perspectivee and seek to o assess the immpact on nattional or regional welfaree, including im
mpacts on
intangible goods
g and se ervices. This broader ecoonomic persp pective is th e appropriatte one to
apply if calcculations of fllood damagee are to be used for public policy.
Several priinciples of economic

e vaaluation need to be considered in order to conduct a
comprehen nsive flood da
amage valuattion study in a consistent way. Some oof the most important
principles as
a regards th he economic valuation off flood dama age are speccified in the following
sections. Th on of the prinnciples of economic valuation is cruc ial for an ap
he applicatio ppropriate
examination of flood da amage, and tthe use of the eir results in flood risk annalysis. There
efore, It is
essential to
o follow these
e principles ass closely as practicable.
p
Economic points
p of view
w
Economic aspects
a of flood damagee valuation and a analysis require a deetermination of flood
damages to
o the nationaal economy. TThere is no matter
m of who would payy for the dam
mages and
who would receive the benefits
b (if anny) as result from
f the flood.
The analystt has to poinnt out the floood damagess in a view point of the ssociety as a whole.
w To
reflect a vieew point of the
t society, tthe economicc and/or shaddow prices off goods and services ‐
opportunityy cost are used
u in estim
mating flood damages and benefits.. Thus, anything that
reduces naational per ca apita wealthh is a “damage” and anyything that inncreases nattional per
capita wealth is a “benefit”.
In an econo
omic damage study, depreeciated values should be applied
a in ordder to reflect the value
of a good at the tim me when it is damaged by a flood d. Using repplacement co ost is an
overestimation of damaage from a brroader econo omic perspecctive, becausee replacemen nt usually
involves im
mprovementss: old house s/goods whiich are dam maged duringg a flood arre usually
substituted by new, morre productivee and better performing
p ones.
o
nomic prices for

Apply econ f valuationn of damages
The econom mic valuation
n of goods annd services whichw are traded in the m market is a ta
ask that is
relatively easy to accommplish. Howevver, accordin ng to econom mic theory sevveral rules ne
eed to be
considered to identify thhe ”correct“ vvalue. As stated in most economic
e texxtbooks, the value
v of a
market goo od correspond ds to its scarccity price. An
nd a scarcity price emergees in the conttext of an
ideal comp petitive markket with maany competin ng actors involved and without govvernment
distortions. Since these ideal conditiions are neve er fulfilled in a real‐worldd market, adjjustments
are necessaary in order to calculate the ideal “sh hadow price“ of the goood under examination.
This means, real world market
m pricess must be translated into shadow
s pricees by excludin
ng all kind
of transfer payments likke taxes and ssubsidies and d by convertin ng monopolisstic market prices
p into
competitivee market pricces as well as by adjusting for the socia al value wheree market valuues fail to
account forr social beneffits and lossess (externalitie
es).
Time value of money

The time vaalue of mone ey is the wayy of thinking about the choice betweeen holding re eal goods
this year orr holding “mo
oney” and w waiting to purrchase those goods in a ffuture year. The
T “time
value” in thhe future refflects how ppeople generally think ab
bout the futuure and whether they
would ratheer save or have somethingg they will use.
Guidelines Floo
There is a simple way to think about this. If you have USD100 today and lend it to someone or
put it in a bank, you expect to be paid interest for the use of the money (for example, 10% per
year). This means that from USD100 today you will have USD110 one year later. In another
words, the value of USD100 today worth USD110 next year and can buy 10% more (or 10%
minus inflation).
In economic analysis the economist compares value of money at a different time instead of the
amount of money. To do this, economists use a ‘discounting’ technique for converting value of
money to a specific time/date. A number of terms, presented below, can describe this
relationship over time, using mathematical equations.
Compounding factor
This factor is used to determine future value (FV) of a present amount (PV) at the end of the nth
period at the interest rate of i:
FV = PV (1 + i)n
An easy way to think about the compound factor is to think of the ‘compound interest rate’
from bank interest, with the same interest multiple applied to the new amount at the end of a
period of time, such as every year. In the example above for USD100 in the bank, if the
compound rate is 10%, applied every year, at the end of 2 years, you would have, USD121.
Compounding factor of once per year with annual distributions

This is used to calculate the future accumulated value (FAV) at the end of nth period at an
interest rate of i with a sequence of equal payments (A).
(1 + i)n ‐ 1
FAV = A ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
i
Here, the easy way to think about this is for a period (n) of 2 years, like above. In the example of
USD100 in the bank, you would receive USD10 after the first year and USD10 after the second
year. Of course, the total amount you would have would depend on what you do each year with
the USD10 in interest. If you reinvest it at the same interest rate, then you have USD121 after 2
years as in the example above. But if you spend the interest each time you receive it, your
future accumulated value is only USD20.
Capital recovery factor

This is used to calculate the amount of each payment (A) at the end of each n period to recover
the present amount (PV) with the interest rate of i:
i (1 + i)n
A = PV ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(1 + i)n ‐ 1
Net Present Value (NPV)

Net Present Value is the present worth of the income stream (the continual annual payments,
A, above, or the benefits or value of a project) generated by an investment. It can be calculated
by the following equation:
n
( Bt  Ct )
NPV = 
t 1 (1 i)
t

where: B = benefit;
C= cost of investment and annual cost
T= time from year 1 to year n; and
I= discount rate
Note that this equation doesn’t actually use profit but calculates an annual ‘net benefit’ of the
project that is equal to the benefit minus the cost (B less C) in each year. The Greek letter sigma
represents that this is a summation from year one to the final year, n.
Constant price
The concept of constant price assures that a monetary sum taken from one year can be
compared with a sum from another year. Flood damage evaluation is carried out in a specific
year but the cost and damages are always estimated using data from previous years. To keep
the values constant, the values from each year (called the ‘nominal value’ for that specific year)
needs to be converted to a constant/fixed value using the current pricing value. To do this
means applying the same idea of interest rates as applied in the equations above to the year of
the measurement and bringing it up to the current year using information on comparative
prices over the years.
‘Constant’ or ‘fixed price’ refer to a value from which the overall effect of general price changes
has been removed. Using constant prices ensures that the past and future costs and benefits of
the identified alternatives are estimated in the same units as the costs and benefits measured
at the time of evaluation for decision.

M Program
Appendix 4 Analysis of Human Instability in Flood Flow

ws
For smallerr‐scale floods where veloccities are more importantt, loss of stabbility (i.e. top ppling) for
people in flood
f flows can
c be an im mportant reaason of lead ding to loss oof conscioussness and
drowning or o direct drow wning (death)). Criteria havve been deve
eloped to asssess this. These can be
used to asssess the flood risks at a local level. Different series s of expperiments ha ave been
undertaken n (Abt et al., 1989; Karvoonen et al., 2000). For th he available test data, Figure A 1
shows the combinations
c s of depth annd velocity th
hat resulted in instability.. It shows that people
lose stabilitty in flows with limited deepth‐velocityy multiples. The obtained critical depth‐velocity
multiples raange from 0.6 m2/s to ab out 2 m2/s. The T differencces in the repported test re esults can
be related to test circcumstances (ground surfface friction, test configguration) or personal
characteristtics (weight, height, clotthing). Further background informat ion regarding human
instability iss given in Jon
nkman and Peenning‐Rowse ell, 2008.
1,5
water depth (m)
data Abt
A et al.
data Rescdam
R
1 hv=1,35m2/s
hv=0,6m2/s
hv=2m
m2/s
0,5
0
0 1 2 3 4
flow
w velocity (m/s)
Figure A 2 Observations from experim mental series reggarding the com

mbination of waater depth and flow
velocity that resulted in ins tability.
Based on such
s tests criteria have bbeen develop ped that sho
ow different levels of dannger as a
function of water depth h and flow veelocity. An exaample of such a criterion is shown in the
t figure
above with floodwater velocity (in m meters or in feet per seco ond) on the x‐axis and flo
oodwater
depth (in meters
m or feett) on the y‐axxis. Distinctio
on is made between threee ‘zones’ rep
presenting
different leevels of dangger. The abovve criteria caan be used to o determine risk zones foor certain
flood conditions.
Guidelines Floo
MRC Flood Management and Mitigation Prog
gramme Componnent 2: Structurall Measures and Flood Proofing
Figure A 3 Criteria fo
or danger to pe
eople in flood fl ows (USBR, 200
01). The graph is two dimensioonal but
presents depth and velo ocity both in meeasures of meteers and feet.
Finally, in
ndirect flooding effects (e
e.g. on local ffood supply) and longer term health immpacts (e.g.
diseases)) can also contribute to looss of life. H owever, therre are no quantitative appproaches to
directly estimating th he effects off these proc esses on the e loss of life. Such lossess should be
directly ddependent on the area th hat is affecteed and the so
ocial and econ nomic condittions as well
as physsical charactteristics (saffety prepar ation, government and d communitty self‐help
infrastructure, demoggraphics, climmate and ecollogy, etc.).
Guidelines Flood Risk Assesssment App. 4 ‐ 2 May 2010

M Program
Appendix 5 The Best Practice G

Guidelines an
nd Project Phases/Stagees
In order to facilitate ma
anagement oof an enginee ering project, project mannagers norma
ally divide
projects intto phases such as the folloowing five phases:
1. on
Initiatio
2. Planninng/Development/Design
3. Production/Implementation
4. Monito oring/Control
5. Closuree
n idea to solvve or mitigate a problem,, create a prooduct or stru

A project sttarts with an ucture, or
some otherr objective. In
n the initiatioon phase, finances are mo obilised, a prroject team is formed,
equipment and tools are e purchased, and the ideaa begins to ta ake shape. Thhe second phase is the
planning/deevelopment/design phasee. The feasib bility of the id
dea is testedd, and, if succcessful, a
project plan is produce ed and the ddesign is prepared. In Ph hase 3, the pplans and de esigns are
implementeed; the prod duction take s place and the project is underwaay. Monitorin ng during
project imp
plementation may reveal the need to correct the planning andd/or design, and a make
adjustmentts. After completion, the project will beb closed; the project teaam disbands, accounts
are closed, and the prodduct or resultt may be hand ded over to a client.
Figure A 4 The phasess of an enginee ring project.
The Best Practice Gu uidelines aree almost exxclusively ap pplicable to Phase 2: Planning/
his phase cann be subdivided in various stages, listedd below.
Developmeent/Design. Th
a) minary/pre‐fea
Prelim asibility studyy
b) Feasib nd overall pla nning
bility study an
c) Prelim
minary design
d) Detaileed design andd detailed plaanning
e) Constrruction/bid documents
The numbeer and conten nt of the stagees may differr, depending mainly on prroject type orr country‐
specific preeferences. The preliminaary design stage,s for exxample, is ooften includeed in the
feasibility sttudy for engineering projeects.
Each section of the guiddelines appliees to one or more

m of the above
a stages.. In the guide
elines this
will be indiicated by dissplaying symbbols that reppresent each stage (showwn below) in the page
margins.
Guidelines Floo
Phase 2 contains all of the following five stages:
a) Preliminary/pre‐feasibility study
A pre‐feasibility study is the precursor to a feasibility and design study. Its main purpose
is to decide whether it is worthwhile to proceed to the feasibility study stage and to
ensure there is a sound basis for undertaking a feasibility study.
A pre‐feasibility study generally includes:

 Definition of achievable project outcomes;
 Analysis of the development situation and constraints the project is to address based on
collected data;
 Identification of related (government and other stakeholder) policies, programs and
activities;
 Preliminary assessment of the viability of alternative approaches; and
 Preliminary identification of likely risks to feasibility and benefits (including risks to
sustainability).
b) Feasibility study and overall planning
If a project is considered to be feasible based on the pre‐feasibility study, a more

thorough feasibility study can begin. A feasibility study defines the project and its
objectives in detail, and looks at these various forms of feasibility:
 Technical feasibility: Can the measures technically be realised in the local context?
 Operational feasibility: Are the measures to be implemented manageable by the local
people?
 Economic feasibility: Is the cost‐benefit analysis favourable?
 Social feasibility: Are the objectives and measures socially acceptable?
 Environmental feasibility: Are the environmental impacts acceptable?
 Political feasibility: Will the measures be supported by those with power?
 Overall feasibility: Will implementation of the envisaged measures result in
accomplishment of the project objectives?
Field surveys, hydrological and hydraulic analyses (in flood mitigation projects), social and
environmental assessments, stakeholder meetings, and costs estimates are among the basis for
answering the above questions. If the answers are positive, the operations/management
structure and management method can be defined, and any initial planning can be detailed.
c) Preliminary design
If a project is deemed feasible, the preliminary design stage can start. This stage
focuses on the technical measures, as follows:
 Site surveys and investigations and computer modelling provide the data for preliminary
design criteria;
 The design criteria can be translated into the preliminary design of structures and measures
in an integrated and balanced system in which the envisaged management activities are
linked to one another;
 The cost‐benefit analysis (construction and operation) and analysis of environmental, social
and political factors demonstrate whether the project is still viable.
If necessary, project planning can be adjusted based on new insights gained in this stage.

d) Detailed design and detailed planning
During the final design stage, the detailed architectural and engineering drawings
(the blueprints) of all physical components of the project are produced. Virtually
all design problems must have been resolved before the end of the final design
stage. Sufficient detail must be provided by the drawings and the report to allow reasonably
accurate estimates of construction and operating costs, as well as the construction scheduling.
e) Construction documents/bid documents
The detailed designs and construction scheduling are incorporated in construction

documents and bid specifications, giving the contractors the information they
need for construction.
Where sections of the guidelines refer to other than the above‐described phases
(e.g. the construction or monitoring phase), the following symbol is used.

Appendix 6 Flow Chart to Create Flood Maps with a GIS
METHODOLOGY TO CREATE FLOOD MAPS WITH ARCGIS BASED ON ISIS HYDROLOGICAL MODELING RESULTS
Legend:
Notes: This diagram assumes the creation of separate layers for Cambodia and Viet Nam.
● To create water level or water depth maps on the basis of output of the hydrological model (ISIS), Base data Performed action
the full diagram needs to be followed. originating from on data
● Is the ISIS node information already transferred to the GIS and are the water levels already included ISIS or MRC
in the Access tabel (.mdb; see A-7), than only the lower grey section needs to be followed for
Files/ processes to be updated
mapping of a different level/ depth. Derived or when new w.l.-values become
● For water levels not yet included in Access, both grey areas need to be redone. created data by available (whole scheme when
● If the ISIS schematisation changes, the whole diagram has to be followed again. FMMP-C2 ISIS configuration changes)
A-5: step
Section A (B-11): optional step, depending on base data set
ISIS node names (ID's) A-4 Transfer of ISIS water levels
and coordinates of Import and clean into 'georeferenceable' table
reservoir and cross- in Excel and transfer to
section nodes Access
A-7
(.csv)
ISIS node names
A-1 Split in Reservoir and
Import and process in (ID's), type, Import in ArcGIS,
Combine in e.g. Access (River) Cross-section
e.g. MatLab ('return coordinates, location georeference and save in
using queries points layers for Camb. &
A-3 period' data) and w.l.'s in one table shape-format (.shp)
A-6 Vietn.
A-2 (.mdb)
B-1
ISIS node names (ID's) B-2
and water levels Import and clean
(e.g. for return periods in Excel and transfer to
or scenarios) Access
(.csv) A-5
ISIS cross-section ISIS cross-section
B-3
nodes (names, nodes (names,
location and w.l.'s) of location and w.l.'s) of
Camb. Viet Nam
(.shp) (.shp)
Assign data of buffer Assign data of node

Draw 150 m buffer ISIS polylines of river
to cross-section polylines; within buffer to buffer ('spatial Add to ArcGIS project
polygons around cross-sections
remove polylines without node join'); remove buffers with >1 and clean
polylines (.shp)
data node inside
B-7 B-5
(B-9)
(B-12) (B-10)
ISIS reservoir nodes ISIS reservoir nodes

(names, location and (names, location and
Section B w.l.'s) of Camb. w.l.'s) of Viet Nam
Creation of (.shp) B-4 (.shp)
spatial data set
(B-11)
Assign data of node ISIS polygons of
within polygon to polygon Add to ArcGIS project
reservoirs
('spatial join'); remove poly- and clean
gons with >1 node inside
(.shp)
B-8
B-6
B-15
ISIS cross-section B-13 ISIS cross-section
polylines in Camb. with polylines in Viet Nam Convert polylines to
node ID and water with node ID and water points (e.g. 15 points/
levels levels polyline)
ISIS nodes, polyline C-1 ISIS nodes, polyline
(.shp) (.shp) points (and polygon points (and polygon
Put all points (original
nodes + derived points) points) in Camb. with points) in Viet Nam
(B-16) in one layer node ID and w.l.'s with node ID and w.l.'s
(.shp) (.shp)
Convert polygons to B-17
points (e.g. 15 points /
polygon)
B-14 Create raster of required

ISIS reservoir
C-4 ISIS reservoir polygons w.l. by interpolation (cell size
polygons in Camb. with Convert polygons to
in Viet Nam with node Section C as DTM); check ISIS whether
node ID and water rasters using required
ID and water levels water level value Creation of to use all points
levels
(.shp) C-2
(.shp) C-5 flooding maps
C-7
C-6 Rectangular area C-3 Rectangular area
ISIS reservoirs (raster) ISIS reservoirs (raster) Superimpose reservoir (raster) containing (raster) containing
in Camb. with w.l.'s in Viet Nam with w.l.'s raster on point-based focal area in Camb. focal area in Viet Nam
(.adf) (.adf) raster (using 'con') with w.l.'s with w.l.'s
(.adf) (.adf)
C-9
C-8
Mask of focal area in Mask of focal area in Remove the raster Focal area (raster) in Focal area (raster) in
Camb. Viet Nam outside the focal areas Camb. with w.l.'s Viet Nam with w.l.'s
(.shp) (.shp) ('mask') (.adf) (.adf)
C-10
C-13 Focal area (raster) in C-12

Focal area (raster) in DTM: 50m or 100m
Camb. with water Viet Nam with water Subtract DTM from focal cells, vert. accuracy ±
depths depths area rasters 0.5 m (?)
(.adf) (.adf) (.adf)
C-11

V3A BPG FRA Final FMMP 2

Uploaded by

Copyright:

Available Formats

V3A BPG FRA Final FMMP 2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

V3A BPG FRA Final FMMP 2

Uploaded by

Copyright:

Available Formats

Mekon

Best Practtice Guide

© Mekong River Commission

2 FLOOD HAZARD ASSESSMENT ............................................................................................ 11

Guidelines Flood Risk Assessment ‐i‐ May 2010

3 FLOOD DAMAGE ASSESSMENT........................................................................................... 69

4 FLOOD RISK ASSESSMENT ................................................................................................ 103

5 REFERENCES ..................................................................................................................... 117

Guidelines Flood Risk Assessment ‐ ii ‐ May 2010

Appendix 1 Approach to Flood Damage and Risk Assessment in Europe

Guidelines Flood Risk Assessment ‐ iii ‐ May 2010

Guidelines Flood Risk Assessment ‐ iv ‐ May 2010

Guidelines Flood Risk Assessment ‐v‐ May 2010

Guidelines Flood Risk Asseessment ‐ vi ‐ May 2010

ABBREVIATIONS AND ACRONYMS

ADCP Acoustic Doppler Current Profiler (Acoustic Doppler Profiler); an

Guidelines Flood Risk Assessment ‐ vii ‐ May 2010

JICA Japan International Cooperation Agency

Guidelines Flood Risk Assessment ‐ viii ‐ May 2010

Attribute data Also known as "count" data.

Areal rainfall Rainfall in a specific area expressed as an average depth of liquid

Colmatage The natural deposition of particles suspended in water on land

Exposure A measure of the people, assets and activities threatened by a flood

Guidelines Flood Risk Assessment ‐ ix ‐ May 2010

Flood control A structural intervention to reduce a flood hazard.

Flood damage Damage to people, property and the environment caused by a

Flood damage map Map of the predicted flood damage.

Flood hazard map Map of the predicted or documented extent/depth/velocity of

Flood proofing A process for preventing or reducing flood damages to

Hydrological hazard A hydrological event (discharge) that may result in flooding.

Hydraulic roughness Is the measure of the amount of frictional resistance water

Hyetograph A graphical representation of the distribution of rainfall over time.

Guidelines Flood Risk Assessment ‐x‐ May 2010

Mekong River Basin.

Meteorological hazard A meteorological event (storm) that may result in a hydrological

Raster Defines values for pixels occupying a particular rectangular area of

Reach Any length of a stream between any two points.

Routing A technique used to predict the changes in shape of water as it

Guidelines Flood Risk Assessment ‐ xi ‐ May 2010

Spatial data Data or information that identifies the geographic location of

(N‐year) synthetic series Simulated long term event‐related data series.

Guidelines Flood Risk Assessment ‐ xii ‐ May 2010

i) Project backgrround/Reportrt informationn

iv) Applications and Guideline s

B) Project Stage Symb bols: A projecct usually con

a) Preliminary/pre‐feasibility study stage

b) dy and overaall planning sttage

Any part of a guideline falling

Sometimes more than one

ood Risk Assessm

Volume 1 Final Report

Volume 2 Characteristics of Flooding in the Lower Mekong Basin

Volume 4 Project development and Implementation Plan

Volume 5 Capacity Building and Training Plan

This report is Volume 3A in the above series.

Guidelines Flood Risk Assessment ‐3‐ May 2010

1.2 Best Practiice Guideline

The MRCC Secretariatt has develop

for flood risk manageement in the Lower