829 © IWA Publishing 2013 Water Science & Technology | 68.

4 | 2013

Methodology for qualitative urban flooding risk

assessment
João P. Leitão, Maria do Céu Almeida, Nuno E. Simões and André Martins

ABSTRACT

Pluvial or surface flooding can cause significant damage and disruption as it often affects highly João P. Leitão (corresponding author)
Maria do Céu Almeida
urbanised areas. Therefore it is essential to accurately identify consequences and assess the risks André Martins
Laboratório Nacional de Engenharia Civil (LNEC),
associated with such phenomena. The aim of this study is to present the results and investigate the Portugal. Av. do Brasil,
101, 1700-066 Lisbon,
applicability of a qualitative flood risk assessment methodology in urban areas. This methodology Portugal
E-mail: jpleitao@lnec.pt
benefits from recent developments in urban flood modelling, such as the dual-drainage modelling
concept, namely one-dimensional automatic overland flow network delineation tools (e.g. AOFD) and Nuno E. Simões
University of Coimbra, Portugal. Rua Luís Reis
1D/1D models incorporating both surface and sewer drainage systems. To assess flood risk, the Santos – Pólo II da Universidade,
3030-788 Coimbra,
consequences can be estimated using hydraulic model results, such as water velocities and water Portugal

depth results; the likelihood was estimated based on the return period of historical rainfall events. To
test the methodology two rainfall events with return periods of 350 and 2 years observed in
Alcântara (Lisbon, Portugal) were used and three consequence dimensions were considered:
affected public transportation services, affected properties and pedestrian safety. The most affected
areas in terms of flooding were easily identified; the presented methodology was shown to be easy
to implement and effective to assess flooding risk in urban areas, despite the common difficulties in
obtaining data.
Key words | 1D/1D urban drainage modelling, flood modelling, risk assessment, surface urban
flooding

INTRODUCTION

Urban pluvial flooding has the potential to cause significant
damage and disruption as it often occurs in highly urbanised
areas (residential, commercial or industrial occupation).
Due to the density and complexity of the urban built
environment, it is essential to accurately identify flood-vul-
nerable areas. Flooding frequency is expected to increase
due not only to urbanisation but also to expected climate
changes (Ugarelli et al. ).
Recent developments in urban flood modelling (e.g. Leitão
; Maksimović et al. ) allow carrying out of reliable
simulations of the whole drainage system, including both
sewer and surface drainage systems. Developments are based
on the dual-drainage modelling concept (Djordjević et al.
) to simulate the interactions between the two systems.
This concept is essential to accurately simulate flooding con-
ditions, including interactions between the two drainage
systems during flood events. Implementation can be supported
by one- or two-dimensional (1D or 2D) representations of the
doi: 10.2166/wst.2013.310
surface drainage system. The 1D approach represents the sur-
face drainage system as a set of 1D flow paths and 2D terrain
depressions, representing flood-prone areas (Maksimović
et al. ). The latter approach represents the surface as a
2D mesh; the water flows over the mesh, accumulating in ter-
rain depressions (e.g. Chen et al. ; Innovyze ). To set up
a dual-drainage model, the sewer system is coupled with a sur-
face drainage system (1D or 2D approach). Simulations are
then carried out simultaneously on both systems’ models.
Despite the good results that can be obtained using 1D/2D
models, run-time may be extremely long (Leitão et al. ),
especially if the number of mesh elements is large. 1D/1D
models are fast but results may be less detailed and less accu-
rate than those obtained using 1D/2D models (Leitão et al.
; Simões et al. ).
Several references can be found (e.g. Apel et al. ;
Douglas et al. ) with examples of usage of urban drainage
modelling to estimate the consequences of flooding and

Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf

by guest
 830 J. P. Leitão et al. | Methodology for qualitative urban flooding risk assessment
consequently assess flooding risk. In this study, a 1D/1D
urban dual-drainage model is used to assess urban flooding
risk caused by intense rainfall and subsequent surface water
runoff and limited sewer hydraulic capacity. The adopted
risk approach is based on the combination of likelihood
and consequence levels, using a risk matrix – a qualitative
risk assessment method. The likelihood level is estimated
based on the return period of the rainfall event. Consequence
level is obtained from urban drainage model results. Five
levels of both likelihood and consequence are defined by the
risk analyst. Taking into account these levels and the risk
matrix, it is straightforward to identify how severe the risk is
for the event. Different actions may be considered to tackle
the flooding risk, according to set risk acceptance criteria.
This paper presents an application of a qualitative flood
risk assessment methodology to a real case, a densely urba-
nised catchment located in Lisbon (Portugal) where
frequent pluvial flooding events have occurred in recent
years (Leitão ).
METHODOLOGY
Urban flood modelling
Digital elevation models and geographic information
systems in surface flow modelling
Realistic representation of terrain plays a fundamental role
in modelling surface processes, including surface runoff
(Leitão ). Terrain representation is increasingly used
as input for overland flow model applications, such as 1D
overland flow network delineation, 2D representation
of surface runoff and dual-drainage models. Advanced over-
land runoff analyses require a good quality Digital Elevation
Model (DEM) (Leitão et al. ).
Dual-drainage and 1D/1D model
The Automatic Overland Flow Delineation (AOFD) meth-
odology (Maksimović et al. ) delineates DEM-based
1D overland flow networks automatically, allowing setting
up of 1D/1D models quickly and requiring small compu-
tational resources. Using this methodology, accurate
hydraulic simulations of urban drainage systems can be car-
ried out (Leitão et al. ).
The AOFD methodology searches the DEM for local
low points in order to define terrain depressions. A bound-
ary is delineated for each terrain depression, identifying a
Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf
by guest
Water Science & Technology | 68.4 | 2013
flood-prone area. The natural exit point of the flood-prone
area is identified as the termination criterion for the flood-
prone area boundary delineation process. This exit point
represents the starting point for an overland flow path. Start-
ing at a flood-prone area exit point or at a surcharged
manhole, the overland flow path delineation process is
identified by analysing preferential flow directions based
on terrain slope, taking into account the presence of build-
ings and other man-made features represented in the DEM.
Once the 1D overland flow network is defined, the data
are imported into a hydraulic simulation model and auto-
matically coupled to the sewer network of the study area.
The sewer network model used herein was Infoworks CS
(Innovyze ).
Risk assessment
Risk assessment includes the processes of risk identification,
risk analysis and risk evaluation (ISO Guide  ). The
risk matrix is a commonly used qualitative approach for esti-
mating risk in situations where data are insufficient for
detailed analysis or when consequences to be taken into
account are of non-numeric nature. This method is also
widely used for ranking or screening risks when more
than one risk is identified, providing a way to select those
that require further analysis or to identify those that are
broadly accepted (ISO  ). Application of the risk
matrix requires the definition of appropriate scales for
expressing both likelihood and consequence levels of poten-
tial events, and definition of the matrix to derive a risk level
for each combination of likelihood and consequence.
Uncertainty is also an important aspect to take into
account when dealing with urban flood modelling and flood-
ing risk assessment. Several studies (e.g. Freni et al. ;
Fontanazza et al. ; Leandro et al. ) focussed on the
uncertainty associated with the evaluation of potential
flood damage in urban areas. According to Deletić et al.
(), uncertainty sources are as follows: (i) input data; (ii)
model structure; (iii) parameter uncertainty; and (iv) natural
and operational uncertainty. All these sources of uncertainty
contribute to the overall uncertainty associated with the
presented risk assessment methodology; however its quanti-
fication is out of the scope of this study and will be
addressed in future developments.
Estimation of likelihood
The occurrence of a flooding event depends on various fac-
tors associated with the sewer system layout, operational
 831 J. P. Leitão et al. | Methodology for qualitative urban flooding risk assessment Water Science & Technology | 68.4 | 2013

conditions (e.g. sedimentation, blockages, etc.), spatial rainfall
distribution, antecedent catchment and system conditions
(e.g. soil moisture and available sewer hydraulic capacity),
downstream boundary conditions (e.g. receiving water level),
etc. The complexity and time varying nature of these factors
are severe restrictions to determining the return period of
flooding events. In addition, data on these events is also
scarce. Herein events are characterised by the probability of
the associated rainfall event given the limitations associated
with obtaining the likelihood of each flooding event.
Likelihood classes were defined by probability intervals.
Table 1 presents the five classes scale for likelihood and
associated probability ranges adopted in this study.
The probability of a rainfall event can be based on its
return period (Tr), and calculated as the probability of occur-
rence of that event in one year. From the return period of the
rainfall event, its frequency (1=T) is calculated and the prob-
ability estimated using the Poisson distribution.
Estimation of consequences
The consequence dimensions selected for this study are
associated with public and socio-economic functional
impacts, namely continuity of public transportation services,
number of affected properties and pedestrian safety. The
selected metrics to express consequences are, respectively,
interruption of public transport routes, number of affected
properties and the combination of water depth and velocity
(hazard rating). The hazard rating was defined by HR Wall-
ingford et al. () based on the fact that pedestrians may
have problems if water level or flow velocity are too high.
The hazard is computed using Equation (1):
HR ¼ dðv þ 0:5Þ þ DF (1)
where HR is the hazard rating, d is the depth of flooding (m),
v is the velocity of floodwater (m s1) and DF is a debris
factor (in urban areas is 0 or 1 to water depths up to
0.25 m or water depths higher than 0.25 m, respectively).
The values assigned to the variables in each class are
defined in Table 2.

Any event affecting the objectives or performance of urban Risk matrix

water systems, independent of its type, can result in poten-
tial consequences of different sorts, herein called The risk matrix and risk scale adopted are presented in
dimensions: public health and safety, environmental Figure 1 and Table 3, respectively, including the acceptance
impacts, urban water infrastructures and services, structures criteria for each risk level. The risk matrix presented in
and other infrastructural impacts. Moreover, consequences Figure 1 was adopted after discussion and agreement
can also incorporate socio-economic impacts on public among the interested stakeholders.
and private property, including those derived from disrup-
tions to society. Estimation of the consequences of a Table 2 | Consequence dimensions classes
flooding event is also challenging due to the large variability
of conditions that can be found in urban areas. The different Dimensions

dimensions of consequence have to be evaluated using com- Effect on public Number of

parable scales; the classes of consequence should have the Consequence transportation affected Pedestrians
class services (>1 h) properties safety (HR)
same impact from the decision maker’s point of view, for
all consequence dimensions. The usual decision criterion 1 Insignificant No routes 0 properties [0; 0.125]
affected affected
to assign a value of risk to the event is the maximum risk
level, thus the higher value in all dimensions of consequence 2 Low 1 route affected 1 to 10 [0.125;
properties 0.75]
associated with the event.
affected
3 Moderate 2 to 3 routes 10 to 100 [0.75; 1.25]
Table 1 | Flooding event likelihood classes
affected properties
affected
Likelihood class Probability range (%)
4 High 3 to 5 routes 100 to 1,000 [1.25; 2.5]
1 Rare [0; 0.2] affected properties
2 Unlikely [0.2; 1] affected
3 Moderate [1; 2] 5 Severe >5 routes More than [2.5; þ∞]
affected 1,000
4 Likely [2, 10]
properties
5 Almost certain [10; 100] affected

Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf

by guest
 832 J. P. Leitão et al. | Methodology for qualitative urban flooding risk assessment Water Science & Technology | 68.4 | 2013

Figure 1 | Risk matrix considered in this study.

Table 3 | Risk scale, risk acceptance criteria and actions for risk reduction (adapted from catchment) are available for recent events, affecting public
AS/NZS (2004))
transportation services and properties.
Acceptance and
Class Risk level tolerability level Action for risk reduction Flood-prone areas, overland flow paths and urban
1 Low Broadly Not likely to be required. drainage model
acceptable Managed using routine
region procedures, unlikely to The flood-prone areas and overland flow paths in the study
require specific application
area were delineated using the AOFD tool (Maksimović
of resources or measures.
et al. ). The DEM used has a resolution of 1 m
2 Medium Tolerable Costs and benefits are to be
region taken into account and (square cells) and a vertical accuracy of less than 0.1 m. In
opportunities to be total, 867 flood-prone areas were identified, mainly located
balanced against potential at the downstream and flat area of the catchment; the over-
adverse consequences.
land flow paths (approximately 730) are mainly along the
3 High Intolerable Risk cannot be justified. streets (Figure 3).
region Specific actions and
The relevant results obtained using the urban drainage
responsibilities to be
defined and implemented. model were water level in the nodes (flood-prone areas)
Risk reduction measures are and the overland flow paths and flow velocity for each
essential. path. Water depth results were validated by direct measure-
ments during the flooding event, measurements of level
marks in the buildings after the event, and using photos of
CASE STUDY: ALCÂNTARA, LISBON (PORTUGAL) the event. Based on this, the results obtained using the
1D/1D model were assumed to accurately represent the
Catchment area flooding event.

The area selected to test the methodology presented com- Rainfall events
prises two contiguous urban catchments in Alcântara,
Lisbon (Portugal), with an area of approximately 1 km2. Two storms were selected to test the qualitative risk assess-
The southern part of the catchments is relatively flat and ment methodology. The characteristics of the two storms are
includes mainly commercial and residential buildings presented in Table 4.
(1,302 buildings in total), whereas the northern part of the
catchment is significantly steep (elevation ranging from Flooding event consequences
135 to 0 m) and not densely urbanised (Figure 2).
Several flooding events have been reported in this area, Effect on public transportation services
making this area an interesting case study to test the meth-
odology presented. Some historical data about flooding To assess the impact of flooding in public transportation,
consequences (mainly in the downstream area of the identification of existing tram and bus routes in the study

Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf

by guest
 833 J. P. Leitão et al. | Methodology for qualitative urban flooding risk assessment Water Science & Technology | 68.4 | 2013

Figure 2 | Public transportation network in the study area.

Figure 3 | Flood-prone areas and overland flow paths obtained using the AOFD methodology.

area was carried out; two tram and eight bus routes were
found to cross the study area (Figure 2). The metric used
to take into account the disruption caused to the public
transportation service is the number of routes affected
when the water depth is higher than 0.3 m for more than
1 h. To assign a consequence class for each flood-prone
area and overland flow path, a spatial intersection between
the flood-prone areas and the public transportation routes
was obtained using the ArcGIS software (ESRI ).
Number of affected properties
The metric considered to take into account property
damage is the number of flooded properties, i.e. number of

Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf

by guest
 834 J. P. Leitão et al. | Methodology for qualitative urban flooding risk assessment
Table 4 | Characteristics of the two selected rainfall events
Total
duration
Storms Date (min.)
Storm A 17–18 February 800
2008
Storm B 7 April 2008 930
properties affected by water depth higher than 0.2 m. The
properties in the study area have similar characteristics
(three floors, 80 years old on average). Although occupation
can vary, no difference was considered between commer-
cial, residential buildings or other activity. Depth–damage
curves (e.g. Dutta et al. ; Penning-Rowsell et al. ;
De Moel & Aerts ) were not considered due to the
lack of specific curves for the Portuguese type of buildings
and occupation. To identify the buildings affected by a flood-
ing event, for each flood-prone area and overland flow path
with more than 0.2 m water depth (obtained using the 1D/
1D model), the buildings that intersected a flood-prone
area or overland flow path were selected using the ArcGIS
spatial intersection tool (ESRI ).
Pedestrians safety
The metric defined for assessing pedestrians’ safety during a
flooding event is a combination of water depth and flow vel-
ocity (HR Wallingford et al. ). The hazard was
computed using Equation (1). The HR class values (see
Table 2) range from class 1, representing flooding areas
with very shallow flowing water or standing water, to class
5, representative of flooding areas with deep very fast flow-
ing water.
RESULTS AND DISCUSSION
Assessment of flooding likelihood
The first step to estimate the likelihood of the storms con-
sidered in this study was to estimate the catchment’s time
of concentration using the Schaake equation (Schaake
et al. (), cited in Matos ()) and validate by hydraulic
simulations considering constant intensity long rainfall
events; the value obtained was approximately 1 h. This
time of concentration was used as rainfall duration for deter-
mination of the rainfall events’ return period from the
Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf
by guest
Water Science & Technology | 68.4 | 2013
rainfall Intensity–Duration–Frequency (IDF) curves for
extreme rainfall events in Portugal (Brandão et al. ).
Max.
intensity
The probability and likelihood for each of the two rainfall
in 1 min Total depth events were then estimated (Table 5).
(mm h1) (mm)
120 90.2
Assessment of flooding consequences
90 16.8
Applying the selected methodology, flood-prone areas and
overland flow paths were classified according to the conse-
quence dimension classes defined (Table 2).
Water depths higher than 0.2 or 0.3 m were obtained for
flood-prone areas, and taken into account to estimate flood-
ing consequence and flooding risk for public transportation
services and affected properties’ consequence dimensions.
Figures in Table 6 represent the flood-prone areas for each
consequence class.
Figure 4 shows the affected areas in terms of affected
public transportation services and properties for storm A.
Considering the pedestrian safety consequence dimen-
sion, both flood-prone areas and overland flow paths were
considered; it was assumed that flood-prone areas have
standing water, i.e. zero velocity.
Table 7 presents the results for both Storm A and Storm
B. Most flow paths are classified as having insignificant con-
sequence (class 1). The flow paths with higher consequences
are located in the upstream part of the study area, due to
higher slopes, which imply higher flow velocities.
Estimation of risk
In Figures 5 and 6 the results of the risk estimation for the
flood-prone areas and overland flow paths in the study
area are presented.
Most flood-prone areas have a low-risk level (Figures 5(a)
and 6(a)), even for Storm A, classified as an unlikely rainfall
event. The number of medium- and high-risk level flood-
prone areas and overland flow paths is higher for Storm B
than for Storm A (comparing Figures 5(b) and 6(b)). This
is explained by the higher likelihood of Storm B when com-
pared with the likelihood of Storm A, despite the lower
water levels and flow velocities originated by Storm
Table 5 | Considered rainfall events probability and likelihood class
Storms Return period (year) Probability (%) Likelihood class
Storm A 350 0.3 2
Storm B 2 39.3 5
 835 J. P. Leitão et al. | Methodology for qualitative urban flooding risk assessment Water Science & Technology | 68.4 | 2013

Table 6 | Flood-prone areas affecting public transportation services and properties

Number of flood-prone areas affecting public

transportation routes Number of flood-prone areas affecting properties

Consequence class Storm A Storm B Storm A Storm B

1 860 (99.2%) 866 (99.9%) 695 (80.2%) 797 (91.9%)

2 0 0 166 (19.1%) 69 (8.0%)
3 5 (0.6%) 1 (0.1%) 6 (0.7%) 1 (0.1%)
4 0 0 0 0
5 2 (0.2%) 0 0 0

Figure 4 | Flood-prone areas and overland flow paths affecting public transportation services and properties (results for Storm A). (a) Public transportation services, (b) properties.

B. This shows that risk assessment results can differ from
consequence estimation results as likelihood is also taken
into account for risk assessment. Figure 5 also shows that
severe or medium flooding consequences do not necessarily
match high-risk zones.
High-level risk areas are mainly the result of high runoff
velocities, representing a problem to pedestrians. Urban
Table 7 | Overland flow paths affecting pedestrian safety
Number of overland flow paths affecting
pedestrian safety
drainage system managers should look at these areas with
high priority and analyse potential measures to reduce this
risk of flooding.
Another aspect often discussed in flood management is
flooding extent. Flooding extent should always be con-
sidered in combination with water depth. For the study
area, a large flood-prone area could have a high risk due
to its large extent (Figure 5(a) (1)). However, since the
water depth in this area is small, not significantly affecting
public transportation services, properties or pedestrian
safety, even for the flooding event originated by Storm A,
a low flood risk level was obtained.

Consequence class Storm A Storm B

1 594 (81.6%) 680 (93.4%)

2 48 (6.6%) 39 (5.4%)
CONCLUSIONS AND FINAL REMARKS
3 31 (4.3%) 4 (0.5%)
The methodology adopted is easy to implement and effective
4 48 (6.6%) 5 (0.7%)
in assessing flooding risk in urban areas. It can be of interest
5 7 (0.9%) 0
to urban drainage system managers to deal with flooding

Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf

by guest
 836 J. P. Leitão et al. | Methodology for qualitative urban flooding risk assessment
Figure 5 | Flooding risk originated by Storm A. (a) Whole catchment, (b) detail.
Figure 6 | Flooding risk originated by Storm B. (a) Whole catchment, (b) detail.
risk, since it allows identification of priority flood-prone
areas, thus improving decision making on allocation of
resources to minimise flooding risk.
The risk matrix method is a practical method to estimate
risk. Herein, one advantage is the proper definition of
classes helping to reduce the subjectivity of the results. In
addition, such qualitative methods may help reduce the
false notion of accuracy that might exist when dealing
with complex methods. Furthermore, it can be applied
using existing data and be upgraded when additional data
become available.
It is noteworthy that the probability of the rainfall event
originating flooding and the probability of that flooding
event can be significantly different, as catchment character-
istics, such as soil moisture, spatial and temporal variation
of the rainfall period, and the sewer system capacity, may
Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf
by guest
Water Science & Technology | 68.4 | 2013
influence the flood magnitude, and therefore its probability.
The simplification adopted of assuming that the likelihood
of the urban flooding is equal to that of the rainfall event
is considered acceptable for intense rainfall events and,
often, this is the adopted procedure due to difficulties in esti-
mating the real return period and subsequently the
likelihood of the flooding event.
Another advantage of the risk matrix is the use of differ-
ent consequence dimensions, herein three were used. Data
availability is also one factor that can limit the number of
dimensions.
Combining the risk matrix method with results obtained
using advanced urban drainage models, such as 1D/1D or
1D/2D models, valuable outputs can be produced for
urban drainage systems managers. The selected method-
ology, which uses a 1D representation of the surface flow
 837 J. P. Leitão et al. | Methodology for qualitative urban flooding risk assessment
network, was shown to be reliable and easy to implement;
further developments include coupling with real-time fore-
casting systems, which, if well calibrated can increase
effectiveness of risk reduction measures and thus avoid sig-
nificant damage and disruption. The 1D representation of
the surface has significantly shorter running times and
smaller data storage requirements when compared with
the 2D representation of the surface flow, which make it
more suitable for real-time modelling and forecasting
(Leitão et al. ; Simões et al. ; Wang et al. ).
The 1D/1D approach used herein represents the hydraulic
processes that take place during pluvial flood events,
except when surface flow is multi-directional. The 1D/2D
approach accuracy is highly dependent on the size of the
surface representation mesh (cell size).
REFERENCES
Apel, H., Aronica, G. T., Kreibich, H. & Thieken, A. H.  Flood
risk analyses – how detailed do we need to be? Natural
Hazards 49, 79–98.
AS/NZS  HB 436:2004 Risk Management Guidelines.
Companion to AS/NZS 4360:2004. Australian Standards,
New Zealand Standards.
Brandão, C., Rodrigues, R. & Pinto da Costa, J.  Analysis of
Extreme Phenomena – Intense Precipitation in Portugal
Mainland (in Portuguese). DSRH-INAG, Lisbon.
Chen, A. S., Djordjevic, S., Leandro, J. & Savic, D.  The urban
inundation model with bidirectional flow interaction
between 2D overland surface and 1D sewer networks. In
NOVATECH 2007, Lyon, France.
Deletić, A., Dotto, C. B. S., McCarthy, D. T., Kleidorfer, M., Freni,
G., Mannina, G., Uhl, M., Henrichs, M., Fletcher, T. D.,
Rauch, W., Bertrand-Krajewski, J. L. & Tait, S.  Assessing
uncertainties in urban drainage models. Physics and
Chemistry of the Earth, Parts A/B/C 42–44, 3–10.
De Moel, H. & Aerts, J.  Effect of uncertainty in land use,
damage models and inundation depth on flood damage
estimates. Natural Hazards 58, 1–19.
Djordjević, S., Ivetić, M., Prodanović, D., Savić, D. & Maksimović,
C.  SIPSON – Simulation of interaction between pipe
flow and surface overland flow in networks. Water Science
and Technology 52 (5), 275–283.
Douglas, I., Garvin, S., Lawson, N., Richards, N., Tippet, J. &
White, I.  Urban pluvial flooding: a qualitative case study
of cause, effect and nonstructural mitigation. Journal of Flood
Risk Management 3 (2), 112–125.
Dutta, D., Herath, S. & Musiake, K.  A mathematical model
for flood loss estimation. Journal of Hydrology 277 (1–2),
24–49.
ESRI  ArcGIS 10 Desktop Help. Environmental Systems
Research Institute, California, USA.
Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf
by guest
Water Science & Technology | 68.4 | 2013
Fontanazza, C. M., Freni, G. & Notaro, V.  Bayesian inference
analysis of the uncertainty linked to the evaluation of
potential flood damage in urban areas. Water Science and
Technology 66 (8), 1669–1677.
Freni, G., La Loggia, G. & Notaro, V.  Uncertainty in urban
flood damage assessment due to urban drainage modelling
and depth-damage curve estimation. Water Science and
Technology 61 (12), 2979–2993.
HR Wallingford, Flood hazard Research Centre – Middlesex
University and Risk & Policy Analysts Ltd  R&D
outputs: flood risks to people – Phase 2. FD2321/TR2
Guidance Document, DEFRA, London, UK (Online:
accessed 15 March 2012).
Innovyze  InfoWorks CS Technical Review (Online: accessed
15 March 2012).
ISO 31000  Risk Management. Principles and Guidelines.
International Standards Organization, Geneva, Switzerland.
ISO Guide 73  Risk Management. Vocabulary. International
Standards Organization, Geneva, Switzerland.
Leandro, J., Leitão, J. P. & de Lima, J. L. M. P.  Quantifying the
uncertainty in the Soil Conservation Service food
hydrographs: a case study in the Azores Islands. Journal of
Flood Risk Management 6 (3), 279–288.
Leitão, J. P.  Enhancement of Digital Elevation Models and
Overland Flow Path Delineation Methods for Advanced
Urban Flood Modelling. PhD Thesis, Imperial College
London, London, UK.
Leitão, J. P., Boonya-aroonnet, S., Maksimović, Č ., Allitt, R. &
Prodanović, D.  Modelling of flooding and analysis of
pluvial flood risk – demo case of UK catchment. In: Flood
Risk Management: Research and Practice (P. Samuels, S.
Huntington, W. Allsop & J. Harrop, eds). CRC Press/
Balkema, Leiden, The Netherlands.
Leitão, J. P., Boonya-aroonnet, S., Prodanović, D. & Maksimović,
Č .  The influence of Digital Elevation Model resolution
on overland flow networks for modelling urban pluvial
flooding. Water Science and Technology 60 (12), 3137–3149.
Leitão, J. P., Simões, N. E., Maksimovic, C., Ferreira, F.,
Prodanovic, D., Matos, J. S. & Sá Marques, A.  Real-time
forecasting urban drainage models: full or simplified
networks? Water Science and Technology 62 (9), 2106–2114.
Maksimović, Č ., Prodanović, D., Boonya-aroonnet, S., Leitão, J. P.,
Djordjević, S. & Allitt, R.  Overland flow and pathway
analysis for modelling of urban pluvial flooding. Journal of
Hydraulic Research 47 (4), 512–523.
Matos, M. R.  Methods to Analyse and Calculate Pluvial
Flows in Urban Drainage Systems – Study Based on
Portuguese Experimental Urban Catchments (in Portuguese).
PhD Thesis, LNEC – Laboratório Nacional de Engenaria
Civil, Lisbon, Portugal.
Penning-Rowsell, E., Johnson, C., Tunstall, S., Tapsell, S., Morris,
J., Chatterton, J. & Green, C.  The Benefits of Flood and
Coastal Risk Management: a Manual of Assessment
Techniques. Flood Hazard Research Centre, Middlesex, UK.
Schaake, J. C., Geyer, J. C. & Knapp, J. W.  Experimental
examination of the rational method. Journal of the
Hydraulics Division 93 (HY6), 353–370.
 838 J. P. Leitão et al. | Methodology for qualitative urban flooding risk assessment Water Science & Technology | 68.4 | 2013

Simões, N. E., Leitão, J. P., Maksimovic, C., Sá Marques, A. &
Pina, R.  Sensitivity analysis of surface runoff generation
in urban flood forecasting. Water Science and Technology
61 (10), 2595–2601.
Ugarelli, R., Leitão, J. P., Almeida, M. C. & Bruaset, S. 
Overview of climate change effects which may impact the
urban water cycle (PREPARED 2011.011 report).
PREPARED: enabling change project.
Wang, L., Simões, N., Rico-Ramirez, M., Ochoa, S., Leitão, J. &
Maksimović, Č .  Radar-based pluvial flood forecasting
over urban areas: Redbridge case study. IAHS Red Book
Series 351, 632–637.

First received 28 November 2012; accepted in revised form 4 April 2013

Downloaded from https://iwaponline.com/wst/article-pdf/68/4/829/439941/829.pdf

by guest