GIS tools for quantitative flood damage assessment and its [627289]

BABEȘ -BOLYAI UNIVERSITY
FACULTY OF ENVIRONMENTAL SCIENCE AND ENGINEERING
CLUJ -NAPOCA, ROMANIA

PhD Thesis

GIS tools for quantitative flood damage assessment and its
uncertainties in data -scarce environments

Scientific coordinator : PhD student: [anonimizat]. Alexandru Ozunu Eng. Iulia Crăciun

2018

Summary

Introduction ………………………….. ………………………….. ………………………….. ………………………….. …………….. 5
Scope and objectives of the thesis ………………………….. ………………………….. ………………………….. ………….. 9
Outline of the thesis ………………………….. ………………………….. ………………………….. ………………………….. .. 12
List of figures ………………………….. ………………………….. ………………………….. ………………………….. …………. 13
List of tables ………………………….. ………………………….. ………………………….. ………………………….. …………… 15
Abbreviations ………………………….. ………………………….. ………………………….. ………………………….. ………… 16
Part I – Theor etical considerations regarding flood risk assessment ………………………….. ………………. 17
1. Theoretical concepts and statistical data regarding the flood risk ………………………….. ………………. 17
1.1. Floods ………………………….. ………………………….. ………………………….. ………………………….. ………….. 17
1.2. Flood risk terminology ………………………….. ………………………….. ………………………….. ……………… 19
1.3. Flood damage terms and concepts ………………………….. ………………………….. ………………………….. 20
1.3.1. Type of damages ………………………….. ………………………….. ………………………….. …………………. 21
1.3.2. Factors that influence flood damage ………………………….. ………………………….. ………………… 22
1.4. Flood hazard in Europe and Romania ………………………….. ………………………….. ……………………. 23
1.4.1. Flood hazard in Europe ………………………….. ………………………….. ………………………….. ………. 23
1.4.2. Flood hazard in Romania ………………………….. ………………………….. ………………………….. ……. 25
1.5. Flood risk legislation in Romania ………………………….. ………………………….. ………………………….. . 27
2. General methodological approach regarding flood damages assessment ………………………….. …….. 29
2.1. Review of the current methodologies in the field of flood damage assessment …………………… 30
2.2. GIS tools and methods applied in the field of flood damage assessment ………………………….. .. 32
2.3. Hazard analysis. Tools and methods ………………………….. ………………………….. ………………………. 34
2.3.1. Hydraulic models ………………………….. ………………………….. ………………………….. ……………….. 34
2.3.2. GIS tools and plugins ………………………….. ………………………….. ………………………….. ………….. 36
2.4. Exposure analysis and land use data ………………………….. ………………………….. ………………………. 38
2.4.1. Elements at risk ………………………….. ………………………….. ………………………….. ………………….. 38
2.4.2. Land use datasets ………………………….. ………………………….. ………………………….. ……………….. 39
2.4.3. Remote sensing data ………………………….. ………………………….. ………………………….. ……………. 44
2.5. Damage analysis ………………………….. ………………………….. ………………………….. ……………………….. 46
2.5.1. Damage functions ………………………….. ………………………….. ………………………….. ……………….. 47
2.5.2. Tools for consequence analysis – Damage models. ………………………….. …………………………. 48
2.6. Flood damage uncertainties ………………………….. ………………………….. ………………………….. ………. 57

3
Conclusions ………………………….. ………………………….. ………………………….. ………………………….. ……………. 59
Part II. Flood damage assessment and its uncertainties in data -scarce environments. Applications
and results ………………………….. ………………………….. ………………………….. ………………………….. ……………… 60
3. Case study 1. Flood damage assessment and uncertainties analysis for the 2006 flood event in
Ilișua basin in Romania ………………………….. ………………………….. ………………………….. ………………………. 61
3.1. Introduction ………………………….. ………………………….. ………………………….. ………………………….. …. 61
3.2. Study area ………………………….. ………………………….. ………………………….. ………………………….. ……. 62
3.3. Methodology ………………………….. ………………………….. ………………………….. ………………………….. … 63
3.3.1. Hazar d analysis ………………………….. ………………………….. ………………………….. ………………….. 64
3.3.2. Exposure and Vulnerability analysis ………………………….. ………………………….. ………………… 66
3.3.3. Damage analysis ………………………….. ………………………….. ………………………….. …………………. 69
3.4. Uncertainty analysis ………………………….. ………………………….. ………………………….. ………………….. 72
3.5. Conclusions ………………………….. ………………………….. ………………………….. ………………………….. ….. 74
4. Case study 2. GIS tools for large -scale analysis of direct economic flood damage. The case study
of Romania ………………………….. ………………………….. ………………………….. ………………………….. …………….. 76
4.1. Introduction ………………………….. ………………………….. ………………………….. ………………………….. …. 76
4.2. Study area ………………………….. ………………………….. ………………………….. ………………………….. ……. 77
4.3. Methodology ………………………….. ………………………….. ………………………….. ………………………….. … 79
4.3.1. Hazard analysis ………………………….. ………………………….. ………………………….. ………………….. 81
4.3.1.1. The GFI method ………………………….. ………………………….. ………………………….. …………… 82
4.3.1.2. Estimate of the water inundation depth using the Geomorphic Flood Index.
Methodology. ………………………….. ………………………….. ………………………….. ………………………….. . 84
4.3.2. Exposure analysis. Data and methods ………………………….. ………………………….. ………………. 88
4.3.2.1. Satellite images processing ………………………….. ………………………….. ………………………… 89
4.3.2.2. Image segmentation ………………………….. ………………………….. ………………………….. ……… 93
4.3.2.3. Feature -based description and classification ………………………….. ………………………….. . 93
4.3.2.4. Training data. Land use data processing ………………………….. ………………………….. ……. 95
4.3.2.5. Validation and classification results ………………………….. ………………………….. …………… 98
4.3.3. Damage analysis ………………………….. ………………………….. ………………………….. …………………. 99
4.4. Results and discussions ………………………….. ………………………….. ………………………….. ……………. 102
4.5. Conclusions ………………………….. ………………………….. ………………………….. ………………………….. … 109
5. General conclusions ………………………….. ………………………….. ………………………….. ………………………. 111
6. Potential applications for stakeholders and future recommendation ………………………….. ………… 114
References ………………………….. ………………………….. ………………………….. ………………………….. ……………. 117

4
Annex 1. Hazard analysis. Flow direction and flow accumulation maps ………………………….. ……….. 131
Annex 2. The pan -European flood hazard map (JRC map) with a return period of 100 years …… 132
Annex 3. Landsat 8 image processing results ………………………….. ………………………….. ………………….. 133

5
Introduction
The advantages of the areas situated along watercourses have influenced over time the
development of civilizations in flood -prone areas. Even though these areas offer many resources,
being ideal for agriculture and development, river transportation and ene rgetic exploitation, the
settlements situated in floodplains are exposed to flood hazard. Over time, several flood protection
measures along rivers, such as hydrotechnical constructions, dams, embankments, regulation of
watercourse, stabilization of riverb eds, etc., have been developed and implemented in these areas.
However, the flood hazard and risk cannot be completely eliminated and a residual risk is still
present; furthermore, the rapid urbanization that took place in the past years has increased the
exposure of the population and economy, transforming flood events in catastrophes. On the other
hand, climate change increased the frequency of extreme flood events as well as their intensity,
producing great damages worldwide (de Moel, 2012; Samela et al. , 2017). In Europe, in the last
three decades floods were responsible for 39% of the total economic damages related to natural
hazards. In Romania the flood damages registered for this period represent 86% of the total damage
caused by natural hazards, Rom ania being one of the most affected country regarding floods
(CRED EM -DAT; Arghiuș et al., 2011). The extreme flood events are especially affecting the less
developed countries where higher losses were registered. The causes of these negative impacts on
less developed countries and with limited financial resources can be attributed to the lower living
standards, low awareness and level of education of the population regarding natural hazards as
well as the lack or pour implementation of proper protection an d prevention measures against
floods.
Hence, considering that the climate change will continue to increase the frequency and
intensity of the flood hazard and the socio -economic growth will increase the exposure to this
hazard, the improvement of flood ri sk assessment and management is becoming an important task
not only at local and national level but also at pan -European and global scale (Cirella et al., 2014;
Albano et al., 2015; Scorzini and Frank, 2015).
The flood risk assessment, that has been foc using traditionally on hazard analysis and
protection measures, has moved recently to a risk -based approach focusing not only on flood
control and reduction but also on the damage analysis, the risk being a combination between hazard
and its consequences; this change is supported by the adoption of Directive 2007/60/EC (Flood
Directive) of the European Parliament and Council (EU Directive 2007/60/CE, 2007). According

6
to the Directive 2007/60/EC, Art.1, “The purpose of this Directive is to establish a framew ork for
the assessment and management of flood risks, aiming at the reduction of the adverse
consequences for human health, the environment, cultural heritage and economic activity
associated with floods in the Community.” Furthermore, all member states mu st transpose this
Directive and must identify flood prone areas and develop hazard and risk hazard maps as well as
flood risk management plans (EU Directive 2007/60/CE, 2007; te Linde et al., 2011).
In this contex t, there is a need for a more comprehensive approach of flood risk
management that should properly evaluate and quantify all the aspects of flood risk assessment
such as hazard, exposure and vulnerability. These aspects play an important role in the decision –
making process, the identification of hi gh flood risk areas, the planning and the selection of
suitable objectives and design of the appropriate of flood management strategies and risk reduction
for flood risk management (Cirella et al., 2014; Scorzini and Frank, 2015). In this light, there is a n
emerging need for assessing flood consequences and for adopting different scenarios and courses
of action.
There are many ways to express the consequences, e.g. economic damages that can be
produced direct or indirect, loss of lives, environmental damag es or health impact. The ongoing
research studies on quantitative risk analysis are focusing mostly on the estimation of direct
economic damages because the latter are having the greatest impact on societies. Furthermore, the
ex-post estimation of these da mages offers important knowledge regarding the high -risk areas and
the potential damages that can occur to the stakeholders, such as the government, the insurance
industry and local authorities (Merz et al., 2010; Meyer et al., 2013).
The flood risk maps can be developed using two approaches: the qualitative approach,
which is based on the experts knowledge and experience from past flood events and the
quantitative approach which is using models in order to obtain damage and risk information that
can be mo netary quantified.
The general approach for quantitative flood risk assessment includes the determination of
hazard characteristics (water depth and velocity, flow), exposure and its vulnerability in order to
address the assessment of subsequent consequen ces for different probabilities of the hazard. The
results can be represented by hazard and risk maps, which are an important tool in the management
of floods and decision -making process (Albano et al., 2015). In particular, the increasing
development of F ree and Open -Source S oftware (FOSS), mos tly in the field of Geographic

7
Information S ystem (GIS) for flood hazard and flood consequence analysis has proved to be
particularly useful for flood hazard and risk analysis, being able to provide the necessary fea tures
for hazard and risk mapping, for spatial data analysis and visualisation of the results.
An important element that must be considered when a flood damage assessment is done is
the scale of the analysis. Each scale needs different input data, differen t methods, and is used for
different goals. There are three scales that are largely used: micro -, meso -, and macro -scale. The
micro -scale analysis is done at local level needing very detailed information, while the meso -scale
(regional/basin level) and mac ro-scale (national, international level) are using aggregated data
needing less detailed data (Messner and Meyer, 2005).
The hazard analysis is usually done by using hydraulic models combined with GIS tools in
order to estimate the characteristics of the flood. However, this approach is rather challenging
being time -consuming and needing detailed data that can limit the application in data scarce
regions and/or at large -scale. Therefore, the recent studies are developing new methods to identify
floodplain s in large and ungauged watersheds using publicly available data, such as the knowledge
of the geomorphology of the basin derived by public available DEM (Digital Elevation Mode l) by
using GIS systems (Nardi et al., 2013; Samela et al., 2017; Manfreda et a l., 2018). Moreover, many
tools and models were also developed worldwide for the damage estimation using the GIS
environments.
The estimation of total costs associated to flooding, considering also their distribution
within the territory, reflects context – and event -specific characteristics, such as extent and type of
flood event (e.g. flash flood, river flood, dam -break event, etc.), spatial variability and quantity of
exposed elements, resolution and type of input data, cultural or geographical differenc es, and so
on. Hence, increased effort s should be devoted to analyze the sensitivity of risk model parameters
in order to integrate and reduce uncertainties in decision -making processes in order to allow
decision -makers and stakeholders to make more inform ed and better decisions.
Current limitations in flood risk assessment
The current studies regarding flood risk assessment present several limitations such as: the
lack of data for extended areas, the analysis being restricted and not complete; the use of coarse
resolution data that can affect the analysis results in small areas; the use of high -computational
models that are costly and time -consuming, and cannot be applied over large areas or in data –
scarce environments.

8
The hazard analysis requires a wide range of data that may lack in many regions or they
are not consistent over large areas. The unavailability of data, particularly in ungauged basins,
limits the analysis to certain areas, this fact leading at the same time to the development of
heterogeneo us datasets. This situation represents an issue for the large -scale analysis, where the
use of hydraulic models in order to provide the necessary information may prove challenging.
Therefore, the existing hazard and risk maps have a limited extend, the res ources needed to conduct
this analysis in data -scarce environments being unaffordable and time -consuming (Samela et al.,
2018)
In the past years, the collection of exposure data has improved due to the increasing
availability of Earth Observation (EO) data sets, which triggered the interest for high -resolution
data development. However the current approaches in the field still work with coarse resolution
data which may be further improved.
Moreover, different approaches (qualitative and/or quantitative) are used for the
development of flood risk maps. The qualitative approach presents the information in a more
simplistic way, being useful for fast risk communication and identification and prioritisation for
future in -depth analysis of high -risk areas. Howeve r, it cannot be used for cost -benefit analysis,
not offering information about the amount of damage that may occur and about the uncertainties
related to the flood risk assessment process. On the other hand, the use of a quantitative approach
has a higher accuracy. It can provide information that can be monetarily quantified for cost -benefit
analysis of the mitigation measures and it can assess and communicate the uncertainties of damage
calculations (Albano et al., 2017c).
The hazard and risk maps are done using different models, the diversity of free and open
tools increasing over the past years. However their transferability from one area to another can
induce uncertainties in the results, the application of these models therefore needing thorough
validat ion and sensitivity analyses. Therefore, the uncertainties estimation must be considered in
the interpretation of the results and must be included in the flood risk management process having
an important role in the decision -making process and for cost -benefit analysis (Scorzini and Frank,
2015).
Therefore, the present work offers some solutions in order to overcome a part of these
limitations by developing a cost -effective approach for flood risk assessment in data -scarce

9
environments. This is done by usi ng high -resolution data and free tools that are easy to apply in
different areas and at different scales.
Scope and objectives of the thesis
The main goal of this thesis is to quantitatively analyze the flood damage and its related
uncertainties in data -scarce environments. For this purpose, a methodology that can offer cost –
effective (instead of expensive and time -consuming approach) and repeatable (transparent)
solution for the flood area delineation and damage calculation is proposed. This approach could
support different stakeholders in flood hazard delineation, damage estimation and flood risk
analysis in different territorial and socio -economic contexts. Therefore, the presented approach
could represent a starting point that can be useful for knowledge transfer from the scientific
community to stakeholders (e.g. practitioners and decision makers).
In this context, the questions addressed in this thesis refer to: how the free and open -source
GIS features potential could be used in quantitative flood dam age assessment; which are the
uncertainties related to flood damage assessment and how they can be analyzed; how to assess the
flood damages in data -scarce environments; how the accuracy of large -scale application could be
improved using p ublicly available datasets and EO data;
This thesis is focused on the damage assessment in data -scarce environments, applying a
method that is able to use the existing data in small areas and extend it over areas where the
information is missing. In this way, useful knowl edge can be provided in order to conduct a
complete flood damage assessment, including all the flood affected areas. Moreover, the
assessments from this thesis were done using high -resolution data (EO) that improve the accuracy
and reliability of the resul ts.
The main objectives of this thesis are:
– To present a theoretical backgroung regarding the flood risk and damage assessment
and the current tools and methods used in the literature;
– The quantitative estimation of the economic damages at large -scale and for data-scarce
environments, using pan -European depth -damage functions ;
– The application of a GIS -based model for water -depth calculation at large -scale and for
data-scarce environments, using high -resolution data;

10
– The use of the machine learning techni ques and Landsat 8 images in order to provide
high-resolution land use data over large areas with limited availability of data;
– The use of free and open source tools for flood damage quantification ;
– To conduct an uncertainty and sensitivity analysis regar ding the flood damage
assessment .
Therefore, in this thesis two case studies were conducted. The aim of the first case study
was to conduct a flood damage assessment and to quantify the uncertainty and sensitivity regarding
the damage functions that are u sed for damage calculation. Romania is one of the countries in
which the damage functions are lacking, therefore the application of damage functions from other
regions can induce great uncertainties in the results. The hazard analysis was done using HEC
RAS hydraulic model and the exposed elemen ts were obtained from the CORINE Land Cover
database. The flood damage was calculated using the functions proposed by Huizinga, 2007 (JRC
functions) for different European countries and the free and open source tool named FloodRisk.
Given the fact that the function transferability is a c ommon practice in areas for which they have
not yet been developed, the uncertainty analysis represents an important part of flood risk
assessment. It offers a better knowledge on how the model choice influences the results and the
reliability of the resul ts, this information being used in the decision -making process.
For the second case study, a large -scale flood damage analysis (for a return period of 100
years) was conducted for the entire Romanian territory. The aim was to develop a preliminary
simplified methodology for flood risk assessment that can be applied at large -scale and in data-
scarce environments, using high -resolution data and GIS tools. First, for the hazard an alysis a
geomorphic method was applied, using the G eomorphic Flood Area GIS tool (GFA), for rapid and
cost-effective flood extent and water -depth mapping over the entire territory of Romania.
Furthermore, a DEM with a resolution of 30 m was used. The resol ution of current analysis in the
field are ranging from 100 m to 1 km, therefore one aim of this study was to provide a hazard
method that is offering high -resolution hazard maps. Furthermore, this approach generates an
extended flood map, covering the ent ire territory including secondary and minor rivers which
usually are not considered in large scale analyses due to the lack of available data.
For the exposure data, a new land use map was developed by using machine learning
techniques and Landsat 8 multi spectral satellite images with a resolution of 30 m. The available
land use data for large scales (such as the CORINE Land Cover database) has a low resolution

11
(100 m) containing limited land use classes, thus decreasing the accuracy of the results (Albano et
al., 2015). In the last years the growth availability of high -resolution satellite imagery (e.g. Landsat
8, ERS, ASAR and TerraSAR -X) represents an important source of information in data -scarce
environments, allowing the automated detection of differe nt land use classes and the development
of high -resolution land use maps. Therefore, the proposed method can be used for large -scale
detection of elements -at-risk (represented by land use maps) in areas with limited availability of
data. Finally, the econo mic damages were calculated using the pan -European JRC depth -damage
curves and the FloodRisk QGIS plugin.
In this thesis, free and open source tools were used for flood hazard and damage analysis,
which allow the exchange of information in the scientific community, offering the possibility of
further development and improvement of knowledge through a transparent and collaborative
approach. Therefore, the scientific findings become available to the large public in order to be used
for better practices and i mplementation of flood risk management. Moreover, the input data that
have been used (e.g. DEM, CORINE L and cover, Urban Atlas, Landsat 8) have a free availability,
being products of EU projects. The proposed methodology was applied for the case study of
Romania; however, due to the use of free tools and data that are required as input, it can be easily
applied in other areas with limited availability of data.
In conclusion, the proposed methodology is using harmonised and homogeneous data in
order to provi de reliable and comparable results in data -scarce environments. The damage and
uncertainty analysis done in the first study is offering a better understanding of damage functions
applicability and their impact on the results. The large -scale approach is us ing high -resolution data
in order to extract the necessary information for hazard and exposure mapping in areas with limited
availability of data. Therefore, this thesis brings new improvements to the current methodologies
in the field, offering high -resol ution and comparable results using free and open source GIS tools
and information which is easily available worldwide.

12
Outline of the thesis
The thesis is structured as follows:
Chapter 1 presents the main concepts related to floods, flood risk and damage, as well as
the situation in Europe and Romania regarding flood hazard. In addition, Chapter 1 presents the
flood risk governance in Romania, showing the leg al framework of flood risk assessment.
Chapter 2 describes the methodological approach reg arding flood damage assessment. The
available methodologies are described, highlighting the current gaps in the field and how these
gaps can be overcome. The main components of flood damage assessment are described as well
as a review of the tools, metho ds and datasets that are used.
Chapter 3 presents a quantitative flood damage assessment approach and an uncertainty
analysis for a basin in Romania. The approach is using free and open source tools and damage
functions for the damage estimation. This stud y is focusing on the applicability and transferability
of depth -damage functions in areas with limited availability of data that don’t allow the
development of site specific functions. An uncertainty analysis was conducted in order to underline
the influen ce that these functions have on the damage results.
Chapter 4 presents a large -scale flood damage assessment methodology for data -scarce
environments, combining EO high -resolution data (30 m ) and free and open -source GIS tools. A
case study for the entir e Romanian territory was conducted. A GIS-based model (GFA tool) that
is using a geomorphic method was applied in order to develop an extended and high -resolution
water depth map for the study area. A machine learning technique was used to perform a
superv ised classification , in order to obtain the land-use map from Landsat 8 images over the entire
territory of Romania potentia lly exposed to flood hazard . The flood damages were calculated using
a GIS tool, combining the previous estimated hazard and exposure data and the damage functions.
This chapter introduces modelling as a crucial tool for flood risk and damage assessment, and more
important, reveals the advantages of using modelling c ombined with free available datasets in
order to develop hazard and risk maps in data -scarce environments.
Chapter 5 and 6 bring toghether the conclusions from the case studies and put forward
recommendations and potential applications for stakeholders an d outline future potential research
topics derived from the present thesis.

13
List of figures

No. Title of the figure Page no.
1. Factors that contribute to flooding
2. The frequency of natural hazards in Europe in the last 30 years (1988 – 2018)
3. The economic damages caused by natural hazards in Europe in the last 30
years (1988 – 2018)
4. The frequency of natural hazards in Romania in the last 30 years (1988 –
2018)
5. The economic damages caused by natural hazards in Romania in the last 30
years (1988 – 2018)
6. Schematic representation of the flood damage assessment process
7. HEC RAS data processing
8. Exposed elements identification
9. CORINE Land Cover 2012 in Europe
10. Urban Atlas 2012 in Europe
11. Residential land use comparison in Cluj Napoca city, Romania; a. CLC
residential class b. Urban Atlas residential classes
12. Rhine Atlas depth -damage functions
13. FloodRisk plugin components
14. Localization of the study area; a. Romania – georgaphical map; b. Ilișua
Catchment
15. Schematic representation of performed flood damage assessment
16. Representation of data analysis in HEC RAS
17. Flood hazard map for Ilișua Basin developed in this study
18. Identification of the exposed elements for the city of Căianu Mic
19. The improved CORINE land use map developed for Ilișua basin
20. The JRC functions for different countries, introduced in the FloodRisk plugin
21. Graphic representation of the results of damage calculation using different
JRC depth -damage functions and the reported damages
22. Differences in the shape of damage functions (Netherlands -JRC and UK –
JRC)
23. Study area; a. Danube River – Europe location; b. The Lower Danube –
Romania
24. Schematic representation of the applied methodology
25. Schematic representation of the hazard analysis methodology
26. Water depth map developed in this study (GFI hazard map)
27. Comparison between the JRC hazard map and the GFI hazard map
obtained in this study
28. Landsat 8 satellite image, band 1
29. QA band processing
30. Satellite image, band 1, with no clouds covered areas
31. Urban Atlas land use of Cluj -Napoca city; a. Vector file of Urban Atlas land
use; b. Raster file of Urban Atlas land use

14
32. Input and output data for t he validation of the algorithm. Sofia city, Bulgaria
(accuracy 0.76)
33. Land use map developed for the flood prone area using Landsat 8 satellite
imagery
34. Graphic representation of damage results for the next scenario s: a. JRC
wate r-depth and CORINE land use; b. GFI water -depth and Landsat 8
landuse
35. Graphic representation of damage results for the next scenarios: c. GFI
water -depth and CORINE land use; d. JRC water -depth and Landsat 8
landuse
36. Damage value comparison betw een the f our scenarios
37. Comparison between urban damages
38. Comparison between industrial damages
A1. Flow direction
A2. Flow accumulation
A3. The pan -European flood hazard map (JRC map)
A4. Lansat8 Band 1 – Romania
A5. Lansat8 Band 2 – Romania
A6. Lansat8 Band 3 – Romania
A7. Lansat8 Band 4 – Romania
A8. Lansat8 Band 5 – Romania
A9. Lansat8 Band 6 – Romania
A10. Lansat8 Band 7 – Romania
A11. Lansat8 Band 8 – Romania
A12. Lansat8 Band 9 – Romania
A13. Lansat8 Band 10 – Romania
A14. Lansat8 Band 11 – Romania

15
List of tables

No. Title of the table Page no.
1. Damage types and examples
2. Impact of flood events in Central -Eastern Europe countries
3. The CORINE Land Cover database 2012 nomenclature
4. The Urban Atlas database 2012 nomenclature
5. Landsat 8 Operational Land Imager (OLI) and Thermal Infrared Sensor
(TIRS)
6. Comparative description of the damage models
7. Land use classes reclassification and site -specific assets value
for the selected case study
8. Correspondences between JRC depth damage curves and
CORINE land -use classes
9. Results of damage calculation using different JRC depth -damage
functions and the reported damages
10. Relative error and function uncertainty determination
11. Values of the optimal thresholds and relative performance measures
obtained in calibrating the classifier over the investigated basins
12. Examples of pixels values and their meaning for QA band (Landsat 8)
13. Reclassification of Urban Atlas classes
14. Reclassification of CORINE land use classes
15. Correspondence between Landsat 8 land use classes and Urban Atlas
codes
16. Assets value for the study area
17. Damage results for the next scenarios: a. JRC wa ter-depth and CORINE
land use; b. GFI water -depth and Landsat 8 landuse
18. Damage results for the next scenarios: c. GFI w ater-depth and CORINE
land use; d. JRC water -depth and Landsat 8 landuse
19. Results of confusion matrix between CORINE Lan d Cover and Urban
Atlas

16
Abbreviations

ASAR Advanced Synthetic Aperture Radar
CLC CORINE Land Cover
CONHAZ Costs of Natural Hazards
DEM Digital Elevation Model
DSM Damage Scanner Model
EEA European Environment Agency
EM-DAT Emergency Events Database
EO Earth Observation
ERS European Remote -Sensing
EU European Union
FLEMO Flood Loos Estimation MOdel
FLORA FLOod and Roughness Analysis
FOSS Free and Open -Source Software
FSO Feature Space Optimisation
GDP Gross Domestic Product
GFA Geomorphic Flood Area
GFI Geomorphic Flood Index
GIS Geographic Information S ystem
GLCM Gray -Level Co -occurrence M atrix
GRASS Geographic Resources Analysis Support System Software
HIS-SSM High -water Information System – Damage and Casualty Module
HEC -RAS Hydrologic Engineering Center's – River Analysis System
JRC Joint Research Centre
MCM Multi -Coloured Manual
NUTS (fr.) Nomenclature of Territorial Units for Statistics
OCSVM One-Class Support Vector M achine
OLI Operational Land Imager
QA Quality Assessment
QRAS Quantitative Risk Assessment Software
RAM Rhine Atlas Model
RT Random Tree
SVM Support Vector Machine
TIRS Thermal Infrared Sensor
TUFLOW Two-dimensional Unsteady FLOW
VHR Very -High -Resolution
WD Water D epth

17
Part I – Theoretical consideration s regarding flood risk assessment

1. Theoretical concepts and statistical data regarding the flood risk
1.1. Floods
Floods are hydrodyna mic phenomena characterized by “temporary covering of land by
water outside its normal confines” (Gouldby et al., 2005) as a result of an increase in water level
and water flow (Arghius, 2008). There are a couple of criteria regarding flood classification, in th e
following being presented the most important and relevant for this study .
Regarding the ir geographical distribution, the floods can be divided in (Johnson et al.,
2016; Blöschl et al., 2015) :
– Coastal floods, which occur when the coast is flooded by the sea and are caused by
strong winds, high tides, storm surge;
– Fluvial floods, which occur along rivers and are cau sed by rainfall and/or snowmelt.
Regarding the flood genesis, they can be divided into three categories:
– Natural floods , caused by weather conditions and their natural evolution; this type of
flood can be caused by rainfalls, snowmelt, ice break forming ice jam s;
– Anthropogenic floods, that are a result o f human activities, can be caused by dam
failures , failures of spillway structures, bad management of reservoirs , deforestation ;
– Natural -anthropogenic (mixed) floods, caused both by human activities and natural
phenomena; for example, extreme rainfall that can cause dam failure s (Arghius, 2008;
Istomina et al. , 2005 )
Depending on the manifestation of the floods (temporal and spatial scale), they can be
divided into two categories:
– Slow on -set floods . This type of flood usually occur s in large catchments as a result of
prolonged rainfall. The water level increases gradually and can be forecasted, therefore
the authorities and the population have the possibility to prepare and take protection
measures. Usually they can last days or even weeks (Arghius, 2008; Johnson et al.,
2016 ).
– Flash floods . This type of flood occur s suddenly (usually in small c atchments) and last s
a short period of time, the water level having a rapid increase, therefore the time for
warning the exposed population is limited. The manifestat ion is violent, occurring with

18
high velocity, having the potential to cause great economic damages as well as loss of
lives (Arg hius, 2008). They can be caused by very intense rainfall an d/or dam failures .
The genesis of a natural flood is dependent on a number of factors ( Fig. 1) that characterize
the catchment. For example, the climate influences the local weather conditions which can cause
floods due to heavy rainfall or snowmelt. The geology and the land use of the area are two other
important factor s, along with the soil texture , influencing the runoff conditions and the infiltration
characteristics. The slops influence the water velocity, the flow accumulation represents the low
areas where the water ca n accumulate and produce floods. T he distance from the drainage channel
indicates the flood potential for the cl ose areas and the density of drainage network s indicates the
flow accumulation potential (Jhonson et al., 2016; Kabenge al., 2017).

Fig. 1. Factors that contribute to flooding (adapted after Jhonson et al., 2016
and Kabenge et al., 2017)
Flood hazard is characterized by the magnitude and probabilit y of occurrence. The
magnitude refers to the intensity of the flood and depends on the flood parameters such as water

19
depth, water velocity, duration of the flood, maximum discharges , etc. The probabilit y ind icates
the recurrence period of a flood with a certain magnitude (Olfert and Schanze, 2007).

1.2. Flood risk terminology
In literatur e, the terminology regarding flood risk assessment can be different and in some
cases it’ s used with different meanings , which can lead to wrong interpretations and can produce
confusion regarding these terms among different research groups (Arghiuș, 2008; Gouldby et al. ,
2005). This variation in the interpretation of one term can be caused by the fact that this
terminolog y is used in different scientific disciplines, being therefore adapted to respond to
different needs. Therefore, researcher s from different fields may understand differently the risk
assessment process. This problem has been approached by some European pro jects, such as
FLOODsite and CONHAZ in order to adopt a unitary and common terminology (Albano et al.,
2015a). In the FLOODsite project two definitions regarding the term “risk” are described which
are widely used and accep ted in the scientific community :
R = P x C (1)
R (Risk); P (Probability); C (Consequence)

R = H x E x V (2)
R (Risk); H (Hazard); E (Exposure); V (Vulnerability)

In the first definition the term „risk” is defined as the product between the probability of
occurrence of the event and the consequences of this event on the population, economy or
environment ( Ozunu and Anghel, 2007; Klijn, 2009; Scorzini et al ., 2015; B otezan et al., 2015 ).
When referring to flood risk assessment the term „frequency” of occurrence is also used , which
must not be confused with the term „probability”. Therefore, the probability is defined as „the
chance of occurrence of one event compared to the population of all events” (Samuels and
Gouldby, 2009). It can be express ed as a number from 0 to 1 or, as a percent from 0 to 100%, and
usually refers to a specific time frame (for example annual exceedance probability) (Samuels and
Gouldby, 2009; A rghius, 2008). The frequency is defined as „the expected number of occurrences

20
of an event within a specific number of events, often related to a time frame” (Samuels and
Gouldby, 2009), for example annual occurrence frequency.
The consequences represent the impa ct of a certain event and refer to economic damages,
loss of lives, number of people affected, environmental damages, etc. (Gouldby et al., 2005).
In the second definition the risk depends on three factors: hazard, exposure and
vulnerability.
The hazard refers to the characteristics and frequency of the flood; the chara cteristics of
the flood provide information regarding the magnitude of the event and are rep resented by factors
such as water depth, water velocity and flooded area (Albano. et al., 2015a; Winsemius et al., 2013).
The frequency of the flood can provide information about the area s that will be flooded, helping
in the process of identifying the potential exposed elements. For example, a flood with a return
period of 1000 years will affe ct a larger area then a flood with a return period of 10 or 100 years.
However, a hazard doesn’t always result in a disaster, this depending on the exposed factors and
their vulnerability (Gouldby et al., 2005).
The exposure refers to the population and the economic assets situated in the flooded area,
being in this way exposed to the hazard. If in the flood affected area there are no soc io-economic
elements that are exposed, we cannot speak of a hazardous outcome or risk.
The vulnerability refers to the susceptibility of the population and the potential assets
affected by a flood. It is a very complex component that can change in space and time (de Moel,
2012; Kaźmierczak and Cavan, 2011 ; Barredo et al., 2006 ; Ștefănescu et al., 2018 ). The
consequences o f an event depend on the vulnerability of this elements at risk, in other words it
depends on the way they will behave and on their inherent characteristics that make them to
resistant or not to flood impact (Messner and Meyer, 2005).

1.3. Flood damage terms and concepts
In the flood risk assessment process , the hazard characteristics (water depth and velocity,
flow), the exposure , the vulnerability and subsequently the consequences are calculated for
different probabilities. Most commo nly, there are used three probabilities: 0.1 (10 years return
period), 0.01 (100 years return period) and 0.001 (1000 years return period). Therefore, by
associating the consequences with the corresponding probability, the damage -probability curve
can be developed . The area under this curve represent s the expected annual flood damage (Messner

21
et al., 2007). On the other hand, the flood damage assessment involves the quantification of
consequences for one flood event with a certain probability.
1.3.1. Type of damages
Floods affect the exposed elements (such as population, economic assets, the environment),
producing damages. In general, the potential damages produced by floods are classified in four
categories: direct tangible damage, direct intangible damage, indirect tangible damage and indirect
intangible damage (Scorzini and Frank, 2015). Direct damage occurs immediately after the hazard,
as a result of the direct contact between the hazard and the exposed elements. These damages are
further classified in tangible damage, when they can be assessed in monetary terms, and intangible
damage, when the quan tification in monetary terms is not possible or i s difficult. Indirect damages
usually refer to long term effects, and can affect also the sectors outside the flood plain area and
can be caused by direct damage or production interruption, which involves additional costs after
the occurrence of the event. These damages are further classified in tangible and intangible damage
(Jongman et al., 2012; Meyer et al., 2013; Merz et al., 2010) . In Table 1 examples of damages for
each category described above are presented .
Table 1. Damage types and examples (Meyer et al., 2013)
Tangible damage Intangible damage
Direct damage Destruction of buildings,
infrastructure, goods and crops Loss of lives, injuries
Indirect damage Production loss, cost of traffic
disruption, disruption of public
services Trauma, increased vulnerability of
the survivors, negative effects on the
environment

Even though for a complex and comprehensive flood damage assessment, all the damages
described above must be considered and analysed, Meyer et al., 2013 and Merz et al., 2010
highlighted the fact that most of the research in the field focuses on direct tangible damage. This
can be explain ed by the fact that more data are available for direct tangible damage assessment as
well as more mo dels and methods, this type of damage being therefore easier to quantify.
Furthermore, the stakeholders, such as decision makers or the insurance industry, may be more
interested in gathering this type of damage data.

22
1.3.2. Factors that influence flood damage
The flood damage magnitude is determined by the flood impact and the exposed element ’s
resistance. Therefore, the damage influencing factors are divided into impact and resistance
parameters (Merz et al., 2010, Sterna, 2012, Thieken et al., 2005).
The impact parameters are represented by the flood characteristics which influence the
flood impact and action:
– The area of inundation is used to identify the affected elements at risk;
– The water depth is the most used and important parameter, the amount o f damage
produced increasing with the water level;
– The duration of flooding influence s especially the damages to building s and
agriculture . The longer the water persist s in these areas, the buildings will deteriorate
and more crops will fail and therefore the recovery costs will increase;
– The velocity of the water can lead to physical destruction of structures, increasing the
probability of structural failure. Furthermore, this parameter can influence the loss of
lives increasing the number of deaths. The v elocity is important especially in the case
of flash flood s and/or dam failures ;
– The time of occurrence influence s the damage to agriculture, this damage being
determined by the season in which the flood occur s. The time of day is also important,
the flood s occurring during night having the potential to produce more damage due to
the fact that the warning process is more difficult and the people can be more confuse d
about the preparedness measures;
– The contamination of water , as a results of Natech events, can increase the damage
(Messner et al., 2007; Thieken et al., 2005; Merz et al., 2010).
The resistance parameters refer to the characteristic s of the exposed elements, such as type
of the construction, building material, etc. These characteristics can in dicate the capability of the
structural element s to resist to the flood impact and determine their vulnerability (Merz et al.,
2010).

23
1.4. Flood hazard in Europe and Romania
1.4.1. Flood hazard in Europe
Over time, floods have caused great socio -econ omic losses in Europe, in particular in the
countries crossed by large rivers such as the Elbe, Danube and Rhine. In the last three decades
(1988 – 2018) the natural hazards with the highest frequency in Europe a re represented by floods .
The CRED EM-DAT database recorded a number of 506 ev ents since 1988, representing 37% of
the most important natural hazard events that occurred in this period. They are followed by storms
with 27% and extreme temperature with 16%. However, regarding the mortality caused b y natural
hazards, floods caused a number of 2770 deaths, lower than extreme temperatures (143 .954 deaths)
or earthquakes (28 .532 deaths) (Fig. 2) . Moreover, 39% of the total economic damages related to
natural h azards are are caused by floods (CRED EM -DAT ) (Fig.3) .

Fig. 2. The frequency of natural hazards in Europe in the last 30 years (1988 – 2018)
(data from CRED EM -DAT , 2018 )

24

Fig. 3. The economic damages caused by natural hazards in Europe in the last 30 years
(1988 – 2018) (data from CRED EM -DAT, 2018)
Among the most catastrophic flood events , the following can be included: the flood from
1970 on the Danube and Tisza rivers affecting Romania and Hungary and producing damages of
585 million US$; the floods from 1997, affecting Pol and, Germany and Czech Republic a nd
producing total damages of 5. 900 million US$; in 2000 major floods occurred in UK, Italy and
France produ cing total damages of around 10. 000 million US$; the floods from 200 2 produced
damages of around 27. 000 million US$ affecting many countries (Barredo, 2006; Kundzewicz et
al., 2013 ). In the past three decades Central -Eastern Europe has been the most affected part of
Europe , the flood events having a great impact on population and economy (Table 2 ).
Table 2 . Impact of flood events in Ce ntral-Eastern Europe countries (adapted after Raška, 2015)
Country Major floods between 1990 and 2014
Casualties Damage (M US$) Worst events
Czech Republic 100 5.744 1997, 2002, 2013
Slovakia 65 306 1998
Slovenia 1 270 2012
Hungary 10 881 1999, 2010
Romania 426 3.047 1991, 2005, 2006,
2010
Bulgaria 74 510 2005, 2014
Poland 104 7.380 1997, 2001, 2010
Lithuania 4 0 2005, 2010

25
The adoption of Directive 2007/60/EC (Flood Directive) had an important role in the flood
risk management , requiring a uniform and harmonized approach between countries, this Directive
being implemented in each EU member state. The implementation of the Flood Directive requires
three steps: flood risk assess ment, flood hazard and risk mapping and flood risk m anagement
planning (Müller, 2013). The aim of the Flood Directive is to reduce the flood risk and
consequences by the implementation of a common framework, offering also a better coordination
of national flood risk management with the joint EU policy (Prie st et al., 2016). The traditional
approaches focus more on the hazard and exposure components of the risk, and less on the
vulnerability and thus on consequences, mainly due to limitations in available data and knowledge
on damage processes and influencing factors. Therefore, the Flood Directive came to fill this gap
by requiring a holistic approach of flood risk management, including all components of the flood
risk (Albano et al., 2017 a).

1.4.2. Flood hazard in Romania
In Romania, floods are the natural hazard with the highest frequency, Romania being one
of the most aff ected countries in Europe (Arghiuș et al., 2011). Moreover, “ Romania is also the
second country in Europe exposed to this type of natural disaster in ter ms of mortality rate
(42.6%)” (Kovacs et al., 2017). According to CRED EM-DAT, in the last 30 years, the economic
damages caused by natural hazard s are related to floods (86%) and droughts (14%) (Fig. 4, Fig.
5). Regarding the loss of lives, the natural ha zards that caused the highest number of deaths are
extreme temperatures (529 deaths) and floods (436 deaths). In the past years several flood events
had major consequences on the population and economy: the flood from April 2000 in the Crișul
Alb basin in west of the country produced total damage s of 5,5 million euro; the fl ood from August
2002 on Slanic r iver, Buzau County with precipitation exceeding 100 l/m2 caused damages of
292.500 euro; the floods from July 2008 on the Suceava, Moldova and Bistrița ri vers from the
Siret b asin caused great damages, on the Suceava r iver the historical rates were exceeded, 107
cities were affected, the reported damages being around 200 million euro. Also the floods from
2010 with damages of 800 million euro affected 37 counties in Romania (Albano et al., 2017 a).
In this context the development and implementation of improved flood risk management
approaches and policies has become a priority. To serve this purpose the legislation in the field
was constantly improved. In the next section the flood risk governance in Romania is described.

26

Fig. 4. The frequency of natural hazards in Romania in the last 30 years (1988 – 2018)
(data from CRED EM -DAT, 2018)

Fig. 5. The economic damages caused by natural hazards in Romania in the last 30 years (1988 –
2018) ( data from CRED EM -DAT, 2018 )

27
1.5. Flood risk legislation in Romania
The flood risk management involves the implementation of different protection and
mitigation measures, polic ies and procedures regarding flood risk identification and assessment.
The legislation in the field is an important tool that support s the implementation of all these
measures and procedures with the aim of reducing the impact of floods. The most important
legislation that regulate s the flo od hazard in Romania is presented bellow .
The Water Law 107/1996 had as a main purpose the protection of water resources and
aquatic ecosystem s against pollution as well as flood defense. However , regarding flood defense,
the document was lacking details , pointing out just the need for structural defense measures,
prevention and intervention measures, as well as the need for the elaboration of certain strategies
(The Water Law 107/1996, art.67; Al bano et al., 2017 a).
The Law 310/2004 was adopted for modifying and completing the Water Law. It addressed
issues such as protection measures against floods, highlighting the need for an efficient flood risk
reduction. This law states that warning and protection actions must be undertaken for the entire
national territory, on stages of development and must be available to the public authorities.
Furthermore, it forbid s the placement of buildings, economic and social objectives in the flood
prone areas . The construction of protective structures against floods must be maintained and
repaired under the above mentioned law (The Law 310/2004, art. 44, 49, 69 and 85).
The Water Law was also completed by the Law 112/2006 that require s the development of
the N ational Strategy for Flood Risk Management. It also highlights the importance of flood
protection plans and flood risk identification ( The Law 112/2006, art. 43).
The Law 146/2010, which further modifies the Water law, has the aim to transpose the
Directive 2000/60/CE of the European Parliament and Council regarding the flood risk assessment
and management. It includes the concept of flood risk management with the a im of reducing the
negative impact of floods. For each river basin the following information must be provided: a
preliminary flood risk assessment, flood hazard and risk maps, flood risk management plans. The
management of emergency situation s regarding fl oods is also taken into consideration ( The Law
146/2010).
Furthermore, the flood events that took place in the past years highlighted the need of a
comprehensive and holistic approach regarding the flood risk management, that could support
decision makers to prioritize intervention and to select alternative risk mitigation options. In this

28
context the Di rective 2007/60/EC regarding flood risk assessment and management was
transposed in 2010 at national level by the G.O. 846/2010 , regarding the National Stra tegy for
Flood Risk Management for medium and long term. The main aim of this Strategy was the
mitigation of damages and life loss prevention. It includes prevention, protection and preparedness
activities, emergency situation act ivities and activities tha t take place after the flood event (G.O.
846/2010).
The implementation process has three stages: preliminary flood risk assessment,
development of hazard maps and flood risk maps, and the final stage regarding the development
of flood risk management plan s. All these stages were accomplished and as a result The National
Plan for prevention, protection and mitigation of flood effects was developed (N.I.H.W.M ., 2016).
However, all three of flood risk components are acknowledged (hazard, exposure,
vulnerability) and should be analyzed in order to provide objective results and a rational procedure
for quantitative risk and cost -benefit analysis, as prescribed by the Flood Directive: according to
Art. 7 of the Directive 2007/60/EC, flood reduction management plans have to include measures
for flood risk reduction, taking also into account “relevant aspects such as costs and benefits”.
Hence, it is evident that only traditional qualitative risk assessments, commonly used in Romania,
are not alone sufficient to comply with these prescriptions and that more sophisticated quantitative
methods should be adopted (Albano et al., 2017 a).
Romania is also a member of the International Commission for the Protection of the
Danube River (I.C.P.D. R.) since 1995 when the Law 14/1995 regarding the collaboration for the
protection and sustainable usage of the Danube River was adopted. According to the Law 14/1995,
art. 2: “Water management cooperation shall be oriented towards sustainable water manage ment,
relying on criteria for a stable, environmentally sound development, which are at the same time
directed to wards : maintain ing the overall quality of live; maintain ing continu ous access to natural
resources; avoid ing long lasting environmental damage and protect ecosystems; exercise ing
preventive approach”.

29
2. General methodological approach regarding flood damages assessment
The classical approach for flood damage assessment (Fig. 6) has the following three steps:
hazard analysis, exposure analysis and vulnerability analysis (de Moel et al., 2015; Bubeck and
Kreibich, 2011; Scorzini and Frank, 2015).

Fig. 6 . Schematic representation of the flood damage assessment process

30
Before starting a flood damage assessment it’s very important to establish the aim and scale
of the study as well as the available data and resources. Depending on these factors, the appropriate
approach for the study can be sel ected (Messner et al., 2007).
The flood damage assessment plays an important role in the decision makin g process,
nowadays the focus being on the development of improved methods and tolls that will help the
stakeholders in the flood management process and in the implementation of the Flood Directive.
Therefore, European projects such as FLOODsite, Danube Fl oodrisk or CONHAZ were
developed, providing information regarding best practices in flood damage assessment (Messner
et al., 2007; Eleuterio, 2012).

2.1. Review of the current methodologies in the field of flood damage assessment
The deterministic method for flood damage assessment presented above represent s the state
of the art in t he field and has been applied in several research studies. Bubeck et al., 2011 analysed
the future changes in flood damage for the Rhine River basin, using two damage mo dels. The land
use data were represented by the CORINE Land C over (CLC) database and for the future land use
projection the Land Use Scanner model was used. The estimation s of relative flood damage differ
by a factor of 1.4, while the land use projections provide damage results that differ by a factor of
3. Jongman et al., 2012 conducted a comparative flood damage assessment for two case studies
using seven flood damage mode ls. Also a sensitivity analysis was done in order to determine the
uncertainties induced by depth -damage functions and maximum damage values. The results
showed that the modelling approaches have a great variation, the depth -damage functions being
the main source of uncertainties. Also, the study suggests that attention must be paid when
aggregated land use data (such as CLC database) are used, the generalisation of this data inducing
over or underestimation of flood damage (Jongman et al., 2012). Escuder -Bueno et al., 2012
proposed a flood risk methodology that include s social data and analyses the impact and efficiency
of non -structural measures on risk reduction.
The accuracy of the results depends mostly from the resolution of the input data, therefore
the use and development of high resolution data sets at large scale has become an important goal
of the studies in the field. Furthermore, the focus is to develop simplified methodologies that can
be applied over large areas offering in this way harmonised and consistent results regarding the
flood prone areas as well as the flood damage.

31
Several studies conducted by the Joint Research Centre of European Commission (JRC)
developed large scale methodologies in order to respond to the above identified need. B arredo and
De Roo, 2007 proposed a methodology of flood risk based on the standard definition of the flood
risk which states that the risk is the product between hazard, exposure and vulnerability. For this
large scale approach , homogenous data at European level are required. For the hazard analysis, the
flood hazard map of Europe, developed by De Roo et al., 2006 , was used. The map has a 1 km
grid size, being developed using a DEM with a resolution of 1 km. For the exposure and
vulnerability analysis the CLC 2000 and the GDP per capita at NUTS 3 level were used. This study
represent s a first attempt of a harmonised approach of flood risk assessment at large scale
providing a standardised risk index map. However, it is a coarse qualitative methodology that u ses
low resolution data and simplified methods and therefore offer s low a ccuracy results (Barredo and
De Roo, 2007). Winsemius et al., 2013 proposed a framework for flood risk mapping at global
scale using one cascade of models and data with a resolution o f 1 km. Lugeri et al., 2010 re –
processed the map developed by De Roo et al., 2007 in order to use it for a pan -European
quantitative flood risk assessment. The data were downscaled to a resolution of 100 m, and depth –
damage functions and maximum damage va lues were used in order to calculate the flood damages
(Lugeri et al., 2010). The same method was used by Feyen et al., 2012 in a study that assess the
impact of climate change for future flood damage in Europe. In this study, in order to obtain the
hazard map, topographic information and a 100 m DEM were used. The analysis was done for
different return periods and the available data were downscaled to a resolution of 100 m. For the
economic damages the depth -damage functions provided by Huizinga, 2007 and the CLC 2000
database were used. Fu rthermore, te Linde et al., 2011 estimated the future flood risk for the entire
Rhine basin taking into account climate change and by applying the same general methodology.
The damage estimations were done with a resoluti on of 100 m using the Damage scanner model.
Using the cascading model approach proposed in Barredo et al., 2007 study, Alfieri et al.,
2014, developed a pan -European hazard map for a return period of 100 years, downscaling the
available data to a resolut ion of 100 m. The pan -European hydrological model Lisflood was
calibrated and used for hydrological simulations and flood hydrographs development. The
hydraulic simulation was done for 5 km sections using the 2D hydraulic model Lisflood – ACC
and subsequently were put together, representing in this way the pan -European hazard map. More
information regarding the applied methodology are available in Alfieri et al., 2014. The limitation

32
of this approach is represented by the relatively low resolution of the input data that affect s the
analysis in small river areas and the overall accuracy of the results. However, this approach
represent s a first attempt to improve the existing methodologies of flood haza rd assessment at large
scale, offering guidelines for future analysis (Alfieri et al., 2014).
Furthermore, in order to carry out a comprehensive flood risk assessment , this
methodology was applied for different return periods and the consequences were ca lculated in a
study by Alfieri et al., 2016. The damages were calculated overlapping the 100 m resolution hazard
map with the population density map (in order to determine the affected population) and with la
land use map (in order to determine the economi c damages). For the economic damages the depth –
damage functions provided by Huizinga, 2007 and a refined version of CLC database were used.
An attempt to improve the previous approaches and offer more reliable results with a better
resolution, Sampson et al., 2015 proposed a model that provide s hazard maps at a resolution of 90
m and includes a detailed river basin network.
As illustrated above, even though there is an increased interest to improve the
methodologies and the quality of the data applied at l arge scale there are still gaps that make the
process difficult. Many studies focused on improving the hazard maps, developing methods that
offer better resolution results. However, the limited availability of data over large areas still
represents a probl em, therefore simplified method s are used. Furthermore, less attention is given
to quality and resolution of land use data, which represent s an important factor in the exposu re
analysis and subsequently in the flood damage analysis.

2.2. GIS t ools and methods applied in the field of flood damage assessment
In the past years the use of free and open source GIS (Geographical Information Systems)
software in the process of flood damage assessment has increased due to the fact that the GIS tools
are able to provide adequate spatial data proces sing, analysis and mapping. These features are
particularly useful for the development of flood hazard and flood risk maps as well as for the
visualisation of the results. Furthermore, the free availabil ity of data lead to an increase in the
development of tools and models that can be used for flood hazard and flood consequence analysis
(Steiniger and Bocher, 2009; Chingombe et al., 2015; Albano et. al, 2015). These tools can process
data and provide information that are further used for hydraulic modelling with the aim of
obtaining the flood characteristics such as flooded area, water depth and water velocity. For

33
example, the GIS tools can provide georeferenced input data (stream network, cross sections) for
the hydraul ic HEC RAS model, based on the DEM (Digital Elevation Model) (Gogoașe Nistorean
et al., 2016). The QRAS tool from the QGIS software (QGIS, 2017)c can be used for the
preparation of the geometry data needed by the HEC RAS hydraulic model. This approach is of ten
used for areas with ungauged basins where the information is limited and therefore the GIS tools
offer an alternative to obtain the geometry data that are needed. The results of the hydraulic models
can be used also as an input data in the GIS for fur ther analysis such as flood damage analysis. On
the other hand, more hazard tools and models were developed on the GIS platform with the aim
of offering a more user -friendly environment for data analysis, reducing also the computational
time and the uncer tainties induced by the using of different combined models. Such tools are
developed in the existing GIS software QGIS and GRASS (Geographic Resources Analysis
Support System Software). For example the GRASS tools ( r.hazard.flood , r.inund.fluv ,
r.damflood ) use the geomorphologic characteristics of an area in order to provide flood prone area
maps (Albano et. al, 2015; Marzocchi et al., 2014). Furthermore, the GFA (Geomorphic Flood
Area) plugin developed in QGIS can provide flood prone areas based on the geo morphic
information using the DEM of an areas. This plugin can be used for mapping the flood prone area
and the water depth in areas with limited availability of data. It can be applied for preliminary
flood risk assessment over large areas for a complete analysis that includes all the streams of the
basin (Samela et al., 2018) . Another QGIS plugin is FloodRisk, which can calculate the damage
produced by floods based on the distribution and value of elements at risk combined with the water
depth and water velocity ( Albano et al., 2017b ).
In flood damage assessment the distribution of elements at risk is an important factor that
can influence the magnitude of the consequences. The exposure and the vulnerability of the assets
can b e easily represented in the GIS environment using different tools that can provide exposure
and vulnerability maps. Therefore, GIS-based land use database s were developed (such as CLC )
using satellite images that can be processed using the GIS tools. In or der to obtain the assets that
are actually exposed to the hazard , the hazard data are combined with the vulnerability data and
the GIS environment is ideal to perform this type of spatial overlay analysis (Eleuterio, 2012).

34
2.3. Hazard analysis . Tools and methods
The hazard analysis implies the determination of the flood probability and inte nsity . The
flood intensity can be represented by many parameters such as water depth, wat er velocity,
duration and extent of the flood, however the most important and common ly used are the water
depth and water extent (Sole et al., 2013) . These parameters can be calculated using 1 -dimens ional
or 2-dimens ional hydrodynamic models. (de Moel, 2012; Messner et al., 2012). However, there
are ungauged location s where the data necessary for detailed modelling are not available, or the
resources and the time are limited and therefore more simplified methods are needed. Such a
methodology for the flood area estimation was proposed by Samela et al., 2017. This method is
using the basin’s geomorphology in order to obtain a large scale analysis of the flooded area in
scarce -data locations.
The hazard is represented by flood maps, illustrating the characteristics of the hazard (flood
extent , water depth) for different return periods or for a specific historical even t (de Moel, 2012).
Usually, these maps are plotted using the interface o f different software such as GIS, which can
provide adequate sp atial visualization of results.

2.3.1. Hydraulic models
The most common method to perform flood hazard analysis is to use one -dimensional (1D)
or two -dimensional (2D) hydraulic models which provide flood maps containing the flooded area,
water depth and w ater velocity. Usually the 2D model s are applied for small areas, being time –
consuming and the computationally expensive , while the 1D models can be a pplied for larger areas
and do n't require as many resources. There is a wide range of hydrological -hydraul ic models that
are used worldwide for flood simulations, such as HEC RAS, FLO -2D, MIKE FLOOD, RiverFlow
2D and FLORA 2D (Patro et al., 2009; Manfreda et al., 2015; Banks et al., 2014).
HEC -RAS (Hydrologic Engineering Centers’ River Analysis System) (Fig. 7 ) is the most
common free hydraulic model which can be used for river flood as well as dam flood simulations
(HEC -RAS, 2017) . The input data refers to the geometric features of the basin that can be obtain ed
with GIS tools using a high resolution DEM. In t his situation the accuracy of the obtained data
depend s mostly on the DEM resolution. The geometry data offers information about the river and
the floodplain. Other data that are needed are the discharge and the flow conditions. The model

35
contains a viewer for the visualisation of the results as well as a tool call ed RAS Mapper that can
transform the results in a 2D grid and a raster file. This output of the tool can be imported in a GIS
environment for further analysis such as exposure and consequence anal ysis (Kumar Sharma et
al., 2017; Banks et al., 2014; Albano et al., 2017b). In fact, many studies in the flood risk field
(Khattak et al., 2016; Mosquera -Machado and Ahmad, 2007; Ullah et al., 2016) combine the HEC
RAS model together with GIS tools. The mo del has a stable platform and the computational time
is reduced , however it has limitation s regarding its applicability in flat basin areas which can
provide less reliable results (Manfreda et al., 2015) .

Fig. 7. HEC RAS data processing

36
The FLO -2D is a commercial hydrological -hydraulic model that can perform
hydrodynamic analysis using the topographic features of an area. It is used for floodplain
delineation as well as for debris flow simulations. (Banks et al., 2014; Stancanelli and For i, 2015).
The MIKE FLOOD model, developed at the Danish Hydraulic Institute, can be used for
river flood, urban flood and dam break flood simulations. It’s a flexible and user friendly model
that contains several packages such as MIKE 11, MIKE 21 and MIKE urban. Mike 11 is a 1D tool
that can simulate the flow in the channel section, MIKE 21 is a 2D tool used for free surface flow
while MIKE URBAN is used for collection systems. It can be applied at any scale for flood risk
analysis and mapping (Patro et al ., 2009; MIKE FLOOD, 2017) .
RiverFlow 2D is an advance d commercial hydrological -hydraulic model based on a
flexible non -structured mesh and it can simulate river, urban and dam break flooding as well as
sediment and pollutant transport and mud -debris flow. It offers accurate results, being ex tensively
validated and it has efficient computational time s. (RiverFlow 2D, 2017 ).
FLORA 2D (FLOod and Roughness Analysis) is a hydraulic model that simulate s the flood
propagation in flat areas taking into account th e effect of vegetation. It is based on the “shallow
water equations” (Manfreda et al., 2015; Cantisani et al., 2014).
TUFLOW is a commercial model that performs 1D, 2D an d 3D modelling and is also using
GIS software for data processing (TUFLOW, 2017 ).

2.3.2. GIS tools and plugins
The hydraulic models are constantly developing and improving, advancing with the IT
technology, resulting m ore detailed models which offer a better accuracy of the results. However,
in order to use detai ed hydraulic models a con sistent and detailed data set is needed, but most often
such data are not avai lable. Moreover, the use of these models involve s high costs and their
application in large areas is difficult and time -consuming, currently representing an important
problem (Fre ni et al., 2010; Samela et al., 2017). Furthermore, a study done by Freni et al., 2010,
highlighted the fact that the use of detailed hydraulic models in the proces s of damage flood
assessment i s not j ustified most of the time, for reasons such as: additio nal costs and resources;
the advantages of their use are partially reduced by the fact that the use of depth -damage functions
leads to significant uncertainti es, the results mostly depend on the type of depth -damage function
used and not on the hydraulic m odel. The final results of the damage assessment when depth –

37
damage functions are used, shows little differences between using a simplified approach and using
complex hydraulic models (Freni et al., 2010).
Therefore, in order to determine the flooded area s in large watersheds with scarce -data, or
in watersheds with rivers where there are no hydrological stations in order to obtain the stream
flow, it is necessary to apply a simplistic approach which doesn’t need complex and detailed data
sets. In this cont ext the geomorphic features of an area can be a useful factor in the process of
flood prone area delineation. Lastra et al., 2008 demonstrated that the geomorphological method
is more consistent than the hydrological -hydraulic method, when it is applied ov er large areas and
in ungauged basins. Furthermore, the use of a digital elevation model (DEM) has become a
common practice in the developing of new and improved tools and methods for flood area
delineation. Manf reda et al., 2015 and 2018 used DEM informat ion and a line ar binary classifier
to analyse different morphological descriptors and their potential to identify flood -prone areas.
These studies revealed that two morphological features showed particularly good performance: the
distance from the nearest stream (D), which offer s better results for flat areas; and the Geomorphic
Flood Index (GFI), which compares water depth (h r) with the elevation difference (H), ln(h r/H).
The GFI index exhibits good performance when applied in large areas with limited data availability
(Manfreda et al., 2018).
A simplified methodology for the flooded area estimation was proposed in a stud y by
Samela et al., 2017 and i s based on the Geomorphic Flood Index (GFI). This method uses the
basin’s geomorphology in order to obta in a large scale analysis of the flooded areas, which contains
all the rivers, including the small ones which usually are not considered in large scale analysis due
to the lack of sufficient data. The analysis was conducted on the Ohio basin, U.S., and the results
showed a good performance of this approach based on the GFI index, offering accurate flood maps
for large areas with scarce -data (Samela et al., 2017).
Based on this methodology the GFI (Geomorphic Flood Area) tool was developed as a
QGIS plugin. All the input data can be obtained based on a DEM: DEM, filled DEM, flow direction
and flow accumulation. The tool run s a terrain analysis and subsequently the delineation of flood
prone areas is performed. The result is a binary map containin g the flood a rea.

38
2.4. Exposure analysis and land use data
The elements at risk are usually represented by population, buildings, or type of land use
such as industrial area or agricultural areas. These information, which are represented by maps
indicating the characteristic of the elements at risk and their location, are overlapped with the
information regarding the hazard (flood maps) and in this way the exposed elements are obtained
(Fig. 8) (de Moel et al., 2015; Albano et al., 2015).

Fig. 8 . Exposed elements identification

2.4.1. Elements at risk
The element s at risk represent the assets that can be affected by the flood. In order to be
able to assess them, they are classified into groups or classes based on their characteristics and the
damage assessment will be done for these classes. This is necessary because of the lack of data
that would be necessary for a detailed assessment, this being also time consuming and very difficult
to accomplish (Merz et al., 2010; Bubeck and Kreibich, 2011).
Usually , the elements at risk are classified into economic sectors (urban, industry,
agriculture), individual element s of each sector having the same characteristics. The reason to use
this type of classifications is the fact that the economic data regarding these elements are also
available in an aggregated fo rm for each sector and not for each object (Bubeck and Kreibic h,
2011).
Based on these sectors, land use maps were developed with different resolution s and
different n umber of land use classes. These maps are used for the representation of the elements

39
at risk. The most common database is CLC which is a pan -European database with a resolution of
100 m x 100 m. It provides data for most of the countries, containing datasets for the years 2000 ,
2006 and 2012 (see chapter 2.4.2. Land use data sets for more details).
The exposed elements are identified by overlap ping the land use data with the flood data.
After they are identified, ther vulnerability is determine d using depth -damage functions. The
functions will calculate the potential damage based on the economic value of the assets and the
water depth (de Moel an d Aerts, 2010).
Therefore, in order to perform a quantitative damage assessment, the assets ’ value must be
determined. There are two types of values that are currently used in literature: the depreciated
value , which i s the actual value of an asset at the moment of the event; the replacement value ,
which is the value needed to replace the asset that was damage d during the flood ( Merz et al.,
2010; Jongman et al., 2012 ). However, Merz et al. , 2010 , highlighted the fact that using
replacement values can lead to an overestimation of the potential damages .

2.4.2. Land use datasets
The land use is represented by maps at different scales and with different resolution s in
which the areas with the same characteristics are aggregated in homogeneous zones.
CORINE Land C over (CLC) database
CLC offers a pan-European land use data at a scale of 1:100 000. The project coordinated
by European Environment Agency (EEA) provided the first datasets for the reference year 1990
with the aim of offering standardized land use data. These data were subsequently updated for the
years 2000, 2006 and 2012. It has a coarse resolution and the number of classes regarding built -up
areas are limited. Even though it contains 44 land use classes, just two of them are referrin g to
resident ial buildings. These 44 land use classes are organized on three levels: at the first level the
land use is divided in five main categories, while at the second and third levels the main classes
are divided in to more detailed categories (Table 3 ). CLC is the most used database given its large
and free availability. Furthermore, it is the only harmonized European land use database. The CLC
database is available on the Copernicus website (Fig. 9) (Copernicus, 2018 ; Prastacos and
Chrysoulakis, 2011; Feran ec et al., 2007; Kuntz et al., 2014; Büttner, 2014).

40
Table 3 . The CORINE Land C over database 2012 nomenclature ( adapted after Yılmaz, 2010)
Class 1 Class 2 Class 3 Code
Artificial surfaces Urban fabric Continuous urban fabric 111
Discontinuous urban fabric 112
Industrial,
commercial, and
transport units Industrial or commercial units 121
Road and rail networks and associated land 122
Port areas 123
Airports 124
Mine, dump, and
constructions sites Mineral extraction sites 131
Dump sites 132
Construction sites 133
Artificial,
nonagricultural
vegetated areas Green urban areas 141
Sport and leisure facilities 142
Agricultural areas Arable land Nonirrigated arable land 211
Permanently irrigated land 211
Rice fields 213
Permanent crops Vineyards 221
Fruit trees and berry plantations 222
Olive groves 223
Pastures Pastures 231
Heterogeneous
agricultural areas Annual crops associated with permanent crops 241
Complex cultivation patterns 242
Land principally occupied by agriculture, with
significant areas of natural vegetation 243
Agro -forestry areas 244
Forest and semi –
natural areas Forests Broad -leaved forests 311
Coniferous forests 312
Mixed forests 313
Scrub and/or
herbaceous
vegetation
associations Natural grasslands 321
Moors and heathland 322
Sclerophyllous vegetation 323
Transitional woodland -scrub 324
Open spaces with
little or no
vegetation Beaches, dunes, sands 331
Bare rocks 332
Sparsely vegetated areas 333
Burnt areas 334
Glaciers and perpetual snow 335
Wetlands Inland wetlands Inland marshes 411
Peat bogs 412
Maritime wetlands Salt marshes 421
Salines 422
Intertidal flats 423
Water bodies Inland waters Water courses 511
Water bodies 512
Marine waters Coastal lagoons 521
Estuaries 522
Sea and ocean 523

41

Fig. 9. CORINE Land Cover 2012 in Europe (Copernicus, 2018 )

Urban Atlas database
The Urban Atlas database provides high -resolution land use maps for large urban areas
having a resolution of 10 m (Fig. 10 ). It contains data for the reference years 2006 (for 305
European cities) and 2012 (for 697 European cities). The updated version contains 27 urban land
use classes from which 6 are referring to residential buildings ( Table 4 ). It has a high accuracy
which make s it suitable for quality verification of other land use datasets such as CLC (Pazúr et
al., 2017 ; Copernicus, 2018 ). Compared with the CLC database the focus of Urban Atlas is more
on the built -up areas, providing a more detailed classification reg arding residential areas (Fig. 11 ).
The aim of this project was to provide pan -European comparable land use data with a high
resolution (Monte ro et al., 2014).

42

Fig. 10 . Urban Atlas 2012 in Europe
(Copernicus, 2018a )

Table 4 . The Urban Atlas database 2012 nomenclature
(Copernicus, 2018a)
Code Land use class
11100 Continuous Urban Fabric (S.L. > 80%)
11210 Discontinuous Dense Urban Fabric (S.L. : 50% – 80%)
11220 Discontinuous Medium Density Urban Fabric (S.L. : 30% – 50%)
11230 Discontinuous Low Density Urban Fabric (S.L. : 10% – 30%)
11240 Discontinuous Very Low Density Urban Fabric (S.L. < 10%)
11300 Isolated Structures
12100 Industrial commercial public military and private units
12210 Fast transit roads and associated land
12220 Other roads and associated land
12230 Railways and associated land
12300 Port areas
12400 Airports
13100 Mineral extraction and dump sites
13300 Construction sites
13400 Land without current use

43
14100 Green urban areas
14200 Sports and leisure facilities
21000 Arable land (annual crops)
22000 Permanent crops
23000 Pastures
24000 Complex and mixed cultivation patterns
25000 Orchards at the fringe of urban classes
31000 Forests
32000 Herbaceous vegetation associations
33000 Open spaces with little or no vegetation
40000 Wetlands
50000 Water

Fig. 11. Residential land u se comparison in Cluj Napoca city, Romania;
a. CLC residential class b. Urban Atlas residential classes

44
2.4.3. Remote sensing data
The information regarding land use has a significant ro le in the process of flood damage
assessment. The use of land use data can be a major source of uncertainties in the results of the
flood damage assessment due to their accuracy, the level of detail, their availability and the ir
actualization frequency . Due to the rapid urbanization from the last years, there is a lack of up to
date land use data, especially in developing countries (Wieland and Pittore, 2016). Furthermore,
most of the available maps have a low resolution, containing limited land use classes, thus
decreasing the accuracy of the results (Albano et al., 2015).
Remote sensing techniques allow the acqu isition of information regarding land use
characteristics with high spatial, temporal and spectral resolution. Various remote sensing
platforms can serve this purpose, the most common being airborne and satellite sensors. In the case
of satellite remote se nsing, the data acquisition is usually done by Earth -orbiting satellites , which
use on-board sensors to record the electromagnetic energy reflected by the Earth at different
wavelengths. The collected data, also referred to as satellite imagery, are analys ed in order to
extract different features that characterize the Earth’s surface (spectral, textural, geomet rical and
elevation features). For pattern recognition object -oriented approaches are commonly used which
analyse segments (also referred to as objec ts) instead of pixels. In order to achieve this purpose
certain image segmentation procedures are applied (Frohn et al., 2009).
Therefore, in the last years there is an increased interest in using the satellite imagery with
the purpose of obtaining land us e data. The current availability of high -resolution satellite imagery
allow s the automated detection of different land use classes and different settlemen t types. This
trend led to the development of different approaches for land use classification based o n satellite
imagery (Ok, 2013). Biro et al., 2013, applied an object -oriented approach to obtain land use
classification from the TerraSAR -X satellite imag es. This method is based on three steps: image
segmentation, feature -based description based on train ing data , and classification. This approach
effectively identified eight land use classes, using the Feature Space Optimisation (FSO) to
calculate the best features to separate the classes (Biro et al., 2013). Ok, 2013, proposed a new
method for building i dentification from very -high-resolution (VHR) multispectral satellite images.
This method is based on information provided by the buildings’ shadows and it can be applied to
buildings with different characteristics. Liu et al., 2014 used the Landsat TM ima ges with a
resolution of 30 m to detect terrestrial land use conditions, showing the distribution of land use

45
changes and their impact on flood exposure. Wieland and Pittore, 2014 used four machine learning
algorithms to identify urban patterns from multis pectral satellite images. The association of these
algorithms with certain selected features (spectral, textural and geomorphic) provided goo d
performance and flexibility for the method, the best performance being exhibited by the Support
Vector Machine (S VM) and Random Tree (RT) algorithms. The study highlighted the potential of
the method to be applied over large scale areas. Zhang et al., 2014 proposed a method that
combines Landsat spectral data with multivariate texture information in order to detect the urban
built-up areas. The method is using the one -class support vector machine (OCSVM) classifier
which requires training data from one class. The image -based approach proposed Dornaika et al.,
2016 is using image segmentation and descriptor classifica tion to detect buildings from optical
aerial images. Wieland and Pittore, 2016 proposed a large scale application, combining object –
based approach with machine learning te chniques to identify settlement patterns from Landsat 8.
The SVM classifier was used, offering a good performance and indicating a good transferability.

Landsat 8 images
The Landsat Program offers multispectral data of the Earth’s surface, with a medium
resolution, the data being kept in a national archive and made available to the p ublic. This data can
be used in fields such as agriculture, geology, land use planning, etc. The l atest available data –
obtained by Landsat 8 satellite, launched on February 11, 2013 – are high quality, thus improving
and updating the existing database. T he system is composed of two instruments: the Operational
Land Imager (OLI) sensor and the Thermal Infrared Sensor ( TIRS); these instruments provide
images (each of them containing 11 spectral bands) with a resolution of 30 meters (for the visible
spectru m, near infrared spectrum and shortwave infrared spectrum), 100 meters (for thermal
infrared spectrum) and 15 meters for the panchromatic band (Table 5) (USGS, 2018; NASA,
2018 ).

46
Table 5 . Landsat 8 Operational Land Imager (OLI) and Thermal Infrared S ensor (TIRS)
(USGS, 2018 )
Landsat 8
Operational
Land Imager
(OLI)
and
Thermal
Infrared
Sensor
(TIRS)
Bands Wavelength
(micrometers) Resolution
(meters)
Band 1 – Ultra Blue (coastal/aerosol) 0.435 – 0.451 30
Band 2 – Blue 0.452 – 0.512 30
Band 3 – Green 0.533 – 0.590 30
Band 4 – Red 0.636 – 0.673 30
Band 5 – Near Infrared (NIR) 0.851 – 0.879 30
Band 6 – Shortwave Infrared (SWIR) 1 1.566 – 1.651 30
Band 7 – Shortwave Infrared (SWIR) 2 2.107 – 2.294 30
Band 8 – Panchromatic 0.503 – 0.676 15
Band 9 – Cirrus 1.363 – 1.384 30
Band 10 – Thermal Infrared (TIRS) 1 10.60 – 11.19 100 * (30)
Band 11 – Thermal Infrared (TIRS) 2 11.50 – 12.51 100 * (30)
* TIRS bands are acquired at 100-meter resolution, but are resampled to 30 meter s in the delivered data
product.

2.5. Damage analysis
The elements within the flooded area can b e affected to a different extent , depending on
their proximity to the river, the depth and velocity of the water in their locations as well as their
characteristics (type of land use, building material in the case of constructions). The degree of the
damage that can occur can be determin ed using damage functions. As mentioned before, the most
common used are the depth damage functions which determine the susceptibility of a cert ain
exposed element depending to the water depth (Jongman et al., 2012; Cammerer et al., 2013).
These functions are considered a standa rd approach and they are used by different damage
models and can be applied at different scales. Nowadays a wide range of damage functions have
been developed in many countries in Europe and around the world. However, they are speci fic to
the areas and the context for which they were created and therefore their use in other areas can
lead to uncertainties in the results, the validation proc ess being very important in these cases
(Scorzini and Frank, 2015; Cammere et al., 2013). A fir st attempt to develop pan -European
harmonized damage functions was made by the European Commission Joint Research Center
(JRC) (Huizinga, 2007). They developed functions for nine different European countries as well
as an aver age function that can be used for the countries were no specific functions exist (Huizinga,
2007; Jongman et al., 2012). Furthermore, JRC developed a comprehensive global database

47
containing depth -damage functions which can be used for global scale analyse s (Huizinga et al.,
2017).
2.5.1. Damage functions
The most common and used method for the estimation of the expected damage caused by
flood s are damage functions . It must be specified that these functions are used to calculate the
direct tangible damages, this ones producing the hi ghest costs. Moreover, as mentioned before,
more attention and interest is given to the economic damage, usually the risk analysis being limited
to the estimation of this type of damage (Merz et al., 2010; Messner and Meyer, 2005 ). These
functions calculat e the damage of a certain element or class of elements exposed to the flood
considering the type and characteristics of this elements and the impact parameters of the flood
(Molinari et al., 2014). The water depth is considered the most important parameter in determining
the potential damage. Therefore, the most used functions are the depth damage functions , which
calculate the expected damage in function of the water depth (Jongman et al., 2012). In Fig. 12,
the Rhine Atlas functions are presented which de termine the total damage for different land use
categories.

Fig. 12. Rhine Atlas depth -damage functions (de Moel and Aerts, 2010)

48
Depending on the data that are used to develop the damage functions, there are two
approach es that can be used:
– The empirical approach uses real data regarding the affected assets collected after a
flood event. Having the data about the type of assets, the water depth and the damage
of each asset or class of assets, depth damage functions can be developed for different
classes of assets. However, these type of data are specific for a certain area and a certain
hazard event and the use of the resulting functions in other situation s and areas may
induce great uncertainties in the results (Messner et al., 2007; Sterna, 2012 ).
– The synthetic approach focuses on develop ing functions based on standardised type s
of assets, for example land use classes, using “what if” questions. First, the typology
of the classes or constructions (in the case of micro -scale analysis) are establis hed based
on their similar characteristics, the monetary value of the assets is determined and their
susceptibility to certain water levels is assessed by experts. For example, it is
determined which will be the expected damage for a certain type of buildi ng having a
certain water depth (Merz et al., 2010; Messner et al., 2007).
The depth damage functions can be divided into two categories:
– Relative functions , calculating the damage as a percentage of the total value of the
affected asset;
– Absolute functio ns, calculating the damage in monetary terms based on the water depth
(Jongman et al., 2012; Messner et al., 2007).
Therefore, in order for the damage functions to be able to calculate the damage , the asset
value must be predetermined (Thieken et al., 200 5).

2.5.2. Tools for consequence analysis – Damage models .
The development and the use of damage models in the past years has increase d, becoming
an important tool for flood damage estimation and sub sequently flood risk assessment. The
diversity of these models is given by the way in which they were developed as well as the goal for
which they were developed. A first differentiation can be made depending on the type of estimated
damage. Most of these models are focusing on direct economic damages using da mage functions.
However, there are also models that are also considering the loss of lives in their assessment.
Another factor that must be considered is the scale of application. Each scale need s different

49
damage assessment approach es, for example at micr o-scale, the analysis is done at object level ,
while for meso and macro -scale , economic sectors and land use classes are used. Moreover,
depending on the scale, some models may need different input data with a different detail level s.
There are however mod els that can be applied for different scales and allow the user to choose the
level of detail of the input data, but in this situation attention must be given in order to correlate
the correct approach with the analysis scale that is used (H. Kreibich et al., 2011 ; Messner and
Meyer, 2005). In the following a description of the most common damage models used in the
literature are presented along with a comparative review (Table 6).
FloodRisk
FloodRisk (Fig. 13) is a Q GIS plugin that calculates the direct economic damages and the
loss of lives caused by floods and can be applied in analys es at different scales. It is a free a nd
flexible plugin based on a transparent and collaborative approach. The main input data are referring
to the hazard characteristics (water depth, water velocity), exposure characteristics (land use map,
maximum damage value of the assets) and vulnerability (depth -damage functions) . For the
susceptibility analysis, the mode l is using depth damage functions. The model provides a set of
depth -damage functions that were collected from the most commonly used models in this research
field: Hazus MH, Standard method (HIS – SSM model), Rhine Atlas model, Flemish model,
Damage Scanner model, JRC model, Multi -Coloured Model, FLEMO model. These models are
further described in this section. The provided functions are associated with land use cl asses’
codes of the Urban Atlas database. Therefore, if the user uses a different database, a reclassification
of the land use classes’ codes must be done. However, the mo del also allows the user to input and
use other functions . For the estimation of los s of lives, the model takes into consideration factors
such as the preparedness level of population and the existence of warning systems. The provided
results regarding economic flood damage assessment, contain quantitative damage values for each
land use class as well as damage and vulnerability maps. The results for the loss of lives analysis
contain the number of losses fo r different water depth ranges and maps representing the population
at risk and loss of lives ( Mancusi et al., 2015 ; Albano et al., 2017 b). The model is available at:
https://github.com/FloodRiskGroup/floodrisk .

50

Fig. 13. FloodRisk plugin components (Albano et al., 2017 b)

FLEMO model
The FLEMO model (Flood Loos Es timation MOdel) was developed by the German
Research Centre for Geosciences and it calculates the damages for the private sectors (FLEMOps)
as well as the commercial sectors (FLEMOcs) . The FLEMOps estimates direct tangible damage
to residential buildings, while FLE MOcs is used for the estimation of direct tangible damage to
buildings, equipment and goods of the companies. The model s were developed using empirical
damage data from the German regions affected b y floods in 2002, 2005 and 2006. T he damages
are calculated considering the water level, building type and quality of the buildings. Furthermore,
the effect of private precaution measures and the flood water contamination are taken into
consideration offering in this way more accurate results (Thieken et al., 2008; Jongman et al.,
2012; H. Kreibich et al., 2010 ).
The model can be applied at micro -scale (building level) as well as to meso -scale (land use
class level), showing good results for large areas, especially for the economic sector. However, it
tends to overestimate the damages and large uncertainties are registered for the residential sector

51
at very high water level s. It also present s uncertainties r elated to the transferability to other regions.
(Stena, 2012; Thieken et al., 2008 ; Seifert et al., 2010).

HAZUS MH
The HAZUS Multi -Hazard software is estimating the potential damages caused by natural
hazards such as riverine and coastal flooding, hurricane winds and earthquakes. It can be applied
at micro and meso -scale. It has a high level of detail, in this way the user can choose to use an
individual object approach or a surface area approach; also, regarding t he vulnerability analysis
two options can be used: replacement values or depreciated values. The model uses relative
function s and both empirical and synthetic data. The hydrological characteristics are: depth,
duration, velocity , debris, rate of rise, and timing (Jongman et al., 2012 ; Ding et al., 2008 ).
The model can be applied at three levels : Level 1 re quires minimal data and the analyse s
are quick, however the results are less accurate compared to the other levels. It can be use d
however for a preliminary analysis in order to determine the flood prone areas. Using Level 2 more
detailed analyse s can be performed, offering more reliable damage results . The Level 3 analysis is
the most accurate, howe ver more resources and information are needed (Ding et al., 2008). The
uncertainties of this model are related to the spatial distribution of the asse t values, the model
considering a uniform distribution (Sterna, 2012).

HIS-SSM (the Standard Method)
The Standard Method was developed i n Netherlands for the estimation of flood damages
(economic damage and the loss of human lives ). The software developed based on this method is
called HIS -SSM (HIS = High -water Information System; SSM = Damage and Casualty Module)
(Egorova et al., 2008). The damage categories considered in the Standard Method are the direct
tangible and intangible damages as well as the indirect tangible damages (Kok et al., 2004) . The
following hazard parameters are taken into account: velocity, rise rate and water depth. The main
inputs for this model reffer to flood characteristics, land use data, maximum damage value and
damage function s. The maximum damage value of the assets is determined using statistics and
insurance values. The damage functions are based on the water depth, however , for the residential
land use class water velocity is also considered . This model can be applied at meso and macro –
scale, providing good results. The uncertainties of this model are represented by the fact that the

52
direct and indirect damages are calculated using the same depth -damage functions (Meyer and
Messner, 2005 ; Sterna, 2012 ).

Damage scanner (DSM)
The Damage Scanner was developed ba sed on the HIS -SSM module and can also be
applied at meso -scale. Given the limited availability of land use data for large areas, one advantage
of the model is the fact that it doesn’t require detailed data regarding land use, these data being
aggregated. The model is using synthetic data, replacement values and relative function s for the
calculation of direct tangible damages (Jongman et al., 2012; Kellermann et al., 2015) . The general
input data are referring to water depth and land use data. The limitations of the model are
represented by the fact that it doesn’t have empirical validation and it can’t be use for loss of lives
estimation (te Linde et al., 2011).

Flamish model
The model was developed in Belgium and i s used for the estimation of flood damages at
meso and macro -scale . The model is calculating the direct and indirect tangible damages and the
loos of lives using relative damage functions, synthetic data and replacem ent values . The input
data refers to water depth and land use data, however for the estimation of loss of lives, the water
velocity is also taken into consideration . It is using homog eneous land use areas extracted from
the CLC database (Jongman et al., 20 12; Vanneuville et al., 2006).

Multi -Colo ured Manual (MCM)
The MCM is a method for flood damage calculation that was developed in UK and it can
be applied for different spatial scale analyse s. It calculates the direct and indirect economic
damages as well as the loss of lives, having a high level of detail. A large number of damage
categories a re considered, but there are no damage functions developed for infras tructure damage
estimation, the se damages being calculated based on observed traffic volumes. Regarding the
hazard characteristics, beside the water depth the model also takes into consideration the duration
of the flood. The MCM estimates damages using absolute depth –damage functions and synthetic
data. By u sing the absolute approach for the development of depth -damage functions, a frequent

53
update of functions must be done due to the change in the value of properties over time (Jongman
et al., 2012 ; Bub eck and Kreibich, 2011; Meyer and Messner, 2005 ).

Rhine Atlas (RAM)
The RAM model is used for the direct flood damage estimation and can be applied at micro
and meso -scale. It doesn’t have a very detailed land use classification , containing only five land
use classes . The model is using both empirical and synthetic data for the development of depth –
damage curves. The required input data are: water depth, assets characteristics, land use data,
warning time, season (Jongman et al., 2012; Bubeck and Kreibich, 2011; de Moel and Aerts,
2010). The model offers low damage values for different land use classes, underestimating the
potential flood damage, as demonstrated in several surveys ( de Moel and Aerts, 2010 ; Jongman et
al., 2012).

JRC (Joint Research Centre)
This model is a pan -European model developed by the European Commission’s Joint
Research Centre . The model can be applied at meso and macro -scale and uses both empirical and
synthetic data. Depth -damage functions for nine counties were developed in order to calculate th e
damages for f ive land use classes. Using these functions an average function was developed, for
application in countries with limited data availability. The method is also providing maximum
damage values of each Europea n country. In this way a harmonized approach for large scale
application s was developed (Jongman et al., 2012 ; Huizinga, 2007 ).
Some of the described models (FLEMO, DSM, Flemish model, HAZUS -MH, MCM, RAM
and JRC model) were used by Jongman et al., 2012 to estimate flood damage, taking into
consideration two case studies . The results show that JRC, MCM and the Hazaus MH models had
the best performance, while the RAM model underestimated the damages and DSM and Flemish
models tend to overestimate the residential damages (Jongman et al., 2012).
In another study by Scorzini and Frank, 2015 a comparative analysis regarding the
applicability of depth -damage functions was conducted. In this study, each function of JRC was
applied separately. In this case the greatest relative errors were represent ed by the results of the
JRC-UK and JRC -Danmark functions, while the Standard Method and JRC -Switzerland showed
a good performance.

54
De Moel and Aerts, 2011 analyzed the flood damage modeling uncertainties, comparing
three models: RAM, the Flemish model and DSM. The results showed that while the Flemish
model and DSM have similar results, the RAM tend s to underestimate the results.
The difference seen in the results when using different models may also be attributed to
the different approach es that are used to develop these models. For instance, some factors that can
influence the results are: the shapes of the damage functions, the maximum damage values that are
used, the type of damages tha t are taken into consideration, and the scale of application.

55

Table 6. Comparative description of the damage models

FloodRisk
FLEMO
Hazus -MH HIS-SSM
(Standard
Method)
DSM Flemish
model
MCM
RAM
JRC

Country
Italy
Germany
USA
Netherlands
Netherlands
Belgium
UK
Australia
Europe
Damage
function
Relative
Relative
Relative
Relative
Relative
Relative
Absolute
Relative
Relative
Data
development
approach
Synthetic
Empirical Empirical –
synthetic
Empirical –
Synthetic
Synthetic
Synthetic
Synthetic Empirical –
synthetic Empirical –
synthetic
Empirical
validation
No
Yes
Yes
No
No
No
Limited
No
No
Maximum
damage value
Replacement
values
Replacement
values Replacement
values
Depreciated
values Replacement
values
Depreciated
values
Replacement
values
Replacement
values
Depreciated
values
Depreciated
values Replacement
values
Depreciated
values

Scale of
application
All scales Micro -scale
(for each
building)
Meso -scale
(land -use
units)
Micro -scale
Meso -scale
Meso -scale
Macro -scale
Meso -scale
Meso -scale
Macro -scale
All scales
Miscro -scale
Meso -scale

Meso -scale
Macro -scale
Units of
analysis Individual
objects
Surface area
Surface area Individual
objects
Surface area
Individual
objects

Surface area
Surface area Individual
objects
Surface area
Surface area

Input data
Hazard
characteristics
(water depth,
flood
velocity), land
use data,
building type,
warning time,
population
census, Hazard
characteristics
(water depth,
contamination
of flood
water), value
of exposed
assets,
building type
and quality, Hazard
characteristics
(water depth,
flood velocity,
intensity and
timing of the
flood,
duration, rate
of rise,
debris), object Hazard
characteristics
(water depth,
flood duration
and v elocity),
building type,
age, social
class,
occupants. Hazard
characteristics
(water depth),
land use data Hazard
characteristics
(water depth ),
Land use data Hazard
characteristics
(water depth,
flood
duration),
value of
exposed
assets,
building type,
age, social Hazard
characteristics
(water depth),
land use data,
object
characteristic,
season Hazard
characteristics
(water depth)

56
private
precaution type, land use
data. class of the
occupants
Type of
damage
calculated Direct
tangible and
intangible

Direct
tangible Direct
tangible
Indirect
tangible Direct tangible
and intangible
Indirect
tangible Direct tangible
Direct
tangible and
intangible
Indirect
tangible
Direct
tangible and
intangible
Indirect
tangible Direct tangible Direct tangible

Uncertainties Does not
consider the
protection
measures in
the analysis Limited
transferability
in other
regions; tend
to
overestimate
the damages. The model
assumes a
uniform
distribution of
buildings. Requires
highly
detailed data
on
individual
buildings,
industr ies and
infrastructure.
The model
consider just
the depth
water factor.
There is only
one
infrastructure
and one
“industry”
(industry plus
commerce)
class The omission
of direct
damages
regarding
traffic
infrastructure
and cars. The
classification
system is not
very detailed;
Has only one
residential class;
The model
strongly
underestimates
the damages. Aggregated and
generalized data
Advantages The tool is
flexible and
can process
data of
different types
depending
of those that
are actually
available It provides
good results in
large areas for
the economic
sector
High level of
detail of the
model Easy to apply
and provides
comparable
results. It can be used
for the
estimation of
future flood
risk under
climate and
land use
changes. Flexib ility of
the model; fast
processing Very
advanced
method for
flood damage
estimation Doesn’t require
detail data,
being easy to
apply in scarce –
data areas Can be applied for
pan-European
flood risk
assessments;
harmonized
approach.
References L. Mancusi et
al., 2015;
Albano et al.,
2017 b Thieken et al.,
2008;
Jongman et
al., 2012;
Kreibich et
al., 2011;
Seifert et al.,
2010;
Stena, 2012 Jongman et
al., 2012
Bubeck,
Kreibich,
2011
Sterna, 2012
Ding et al.,
2008 Jongman et al.,
2012 , Meyer
and
Messner, 2005 Jongman et al.,
2012
te Linde et al.,
2011 Jongman et
al., 2012
Vanneuville et
al., 2006
de Moel, and
Aerts, 2010 Meyer and
Messner, 2005
Jongman et
al., 2012
Bubeck and
Kreibich,
2011 de Moel, and.
Aerts, 2011
Jongman et al.,
2012
te Linde et al.,
2011
Bubeck and
Kreibich, 2011 Jongman et al.,
2012
Huizinga, 2007.

57
2.6. Flood damage uncertainties
The results of the flood damage assessment are affected by different uncertainties which
can be induced by the input data, the modelling process as well as the spatial and temporal changes
in the information that are used (de Moel and Ae rts, 2011). The analysis of these uncertaint ies is
important for a better understanding of the flood risk process, highlighting the parameters that
induce the greatest errors in the results. In this way the data and the methods that are used can be
improved, making the results more reliable and accurate (de Moel et al., 2012).
In the stu dy by de Moel et al., 2015 the uncertainties related to different spatial scales are
described , showing the gaps and the needs for improvement at each scale. The uncertainties related
to macro -scale analyse s are represented by the fact that the flood protection and defence measures
may not be considered or by the fact that this information has a low accuracy. Furthermore, unitary
input data may not be available for very large areas and therefore the available databases must be
harmonised in order to be used in the analysis. At mes o-scale the uncertainties refer to the
probability of the hazard and the damage functions that are used, while at micro -scale they refer
to the modelling process and lack of detailed data (de Moel et al., 2015).
Another way to approach the estimation of the general uncertainties is to analyse the
uncertainties induced by each component of the flood damage assessment process. For example in
the study by de Moel and Aerts, 2011 the uncertainties related to the flo od depth, land use and
damage models are analysed. This study highlighted the fact that the greatest sources of
uncertainties are represented by the assets value and by the depth -damage functions.
In general the uncertainties related to the hydrological component are caused by the input
data and the choice of the hydraulic model. The use of a complex hydraulic model will need
detailed data for the study area which may not be available and therefore assumption s are made or
general data are used in order to perform the analysis (de Moel and Aerts, 2011). Moreover, if a
simplified method or model is used, less accurate results are obtained due to the fact that not all of
the geomorphologic characteristics of the area will be considered, the input data being less detailed.
In the exposure analysis the uncertainties can be attributed to the land use data resolution
and level of detail. In m any cases the land use data has a low resolution and a reduced number of
land use classes , the data being aggregated (de Moel and Aerts, 2011). The continuous growth of
economy and human settlements lead to changes in the land use, therefore attention must be paid
when using this input data, the use of outdated information leading to less accurate and reliable

58
results. Furthermore, as mentioned above, the value of the assets is an importa nt source of
uncertainties. These values are variable in time but also depend on their location, type and
characteristics of the asset, being therefore hard to assess and maintain an updated database. The
use of general average d values may not reflect the reality i n the field for certain areas.
The depth -damage functions represent another important source of uncertainties, this
problem being appr oached by many studies in the pas t years (Scorzini and Frank, 2015; Cammerer
et al., 2013; Egorova et al ., 2008; de Moel and Aerts, 2011 ; Jongman et al., 2012). The biggest
issue regarding the depth -damage functions is the ir transferability in space. The se functions have
been developed for a particular region or country, and their use in other areas is limited, leading to
uncertainties and unreliable results (Jongman et al., 2012). Such a study was perform by Scorzini
and Frank, 2015 in order to analyse the uncertainties related to different depth -damage functions.
The results showed that for the meso -scale analysis the relative difference factor for function
uncertainty range s between 1.1 and 14.8 and for micro -scale analyses this factor ranges between
1.1 and 2.2. Another study by de Moel and Aerts, 2011 in which the contributions of different
components to the general uncertainties are analysed, showed that the depth -damage functions
together with the value of elements at risk have a variation factor of 4, while the land use and
hydraulic components have a small effect on the uncertainties. Cammerer et al., 2013 highlighted
the importance of model validation, showing that the use of depth -damage functions without
validation in regions other than those for w hich they were created can cause the results to differ
by a factor of 18, compared with the results of validated functions which will only differ by a
factor of 2.3. This study also suggests that the selection of functions from similar geographical
regions will improve the accuracy of the results.
Other uncertainties related to the modelling process can be related to the fact that the
existing method s are mainly focusing on direct damages, while the indirect and intangible damages
are not taken into consi deration even though they can be accountable for a great part of the total
flood damage (Meyer et al., 2013 ; Albano et al., 2014 ; Penning -Rowsell et al., 2010 ). Also, even
though the water depth is the most significant factor that influence s the amount of damage that can
occur, other factors such as the water velocity, flood duration and the presence of dangerous
substance s in the water can contribute and increase the damages caused by floods. However, few
models take also this factors into consideration in their analysis, the majority of them using just
the water depth fa ctor (Messner and Meyer, 2005).

59
Conclusions
Floods are natural hazards with a great negative impact on societies, in the past years
triggering an increased interest among the scientific community and stakeholders regarding the
assessment, mitigation and management of this type of hazard. The flood r isk management is
focusing on a more comprehensive approach, including in the analysis all the components of the
flood risk: hazard, exposure, vulnerability . The consequences are represented by the economic
damages and the loss of life. The flood damage as sessment methodologies are focussing on the
economic damage estimation, the se providing important knowledge on the flood risk. The flood
damage estimation includes information regarding: the hazard (water depth and velocity, flooded
area), the exposure (la nd use data) and the vulnerability (depth -damage functions and maximum
damage value).
The acknowledgement of the importance of flood damage assessment increased the need
for more effective and adequate tools and methods. Particularly the use of GIS tools has a cru cial
role in spatial analysis. Furthermore, with the development and availability of new tools and
datasets more attention is given to the development of standardised and harmonised methods that
can be applied over large areas.
However, the flood damage assessment process is affected by uncertainties induced by the
lack of data, the resolution and accuracy of the input data and the modelling process. The most
important factor in the modelling process that still presents shortcomings are the depth -damage
functions, this factor inducing the greatest uncertainties in the results. Another factor that can
influence the results is the land use data that most of the time are available in aggregated sectors.
Moreover, the large scales analyses are struggli ng with a lack of consistent data and it’s
low resolution , which induce large uncertainties in the results. Therefore, the current scientific
research field is focusing on developing simplistic approaches for macro -scale flood damage
assessment using high -resolution data, improving the accuracy of the results.
The flood damage assessment and subsequently the flood risk assessment process play an
important role in the development of flood management strategies and policies, in the
implementation of adequate protection and reduction measures, and in the cost -benefit analysis.

60
Part II . Flood damage assessment and its uncertainties in data -scarce environments .
Applications and results

In this part of the thesis two applications regarding the estimation of flood damages and
the related uncertainties using free and open source GIS tools and data that are publicly available
are presented . The aim is to provide a framework methodology that can be applied for different
scale analysis and in di fferent area, and particularly in data -scarec environments. Two case stu dy
were selected for the analyse s. In the first case study the damages for the 2006 flood event from
Ilișua basin, Romania were estimated and an uncertainty analysis was conducted . This flood event
represents one of the greates flood disaster in the country, regarding the loss of lives in material
losses in a reduced area. Moreover, in the Ilișua basin the implementation of protection measures
and the land use planning is poor, this aspects increasing the flood damages that may occur.
The second study presents a national flood damage assessment methodology that was
applied for the entire Romanian territory. Romania is one of the most affected country in Europe
regarding flood, 97% o f its terri tory being situated in the Danub e River Basin (Zaharia and
Toroimac, 2018) . Due to t he location in the continental region, many areas of the Romania territory
are exposed to floods (Ozunu et al., 2011) . In this context the flood risk mitigation become a major
concern, the national flood damage analyse s being necessary for an efficient flood management.

61
3. Case study 1. Flood damage assessment and uncertainties analysis for the 2006 flood
event in Ilișua basin in Romania
3.1. Introduction
In the past years the flood risk assessment approaches started to give more attention to the
consequences produced by floods, since they represent an important factor in the flood risk
management. This trend has been supported and encouraged b y the implementation of the Flood
Directive that requires the development of comprehensive approaches for flood risk management.
This approaches must include all the flood risk factors, such as hazard, exposure and vulnerability.
Furthermore, the Flood Dir ective highlighted the importance of developing tools that can bring
improvements in the flood risk quantification process.
In order to quantify the damages, different damage models were developed across Europe,
most of them using depth -damage functions i n order to correlate the water depth with the amount
of damages that may occur.
In this chapter a quantitative flood damage assessment approach is implemented. For this
purpose, the FloodRisk plugin, developed in QGIS software was used . In particular, the depth –
damage functions collected and harmonized by the European Joint Research Centre (JRC),
(Huizinga, 2007) have been used for a comparative assessment showing that the outcomes are
strongly influenced by the shape of the depth -damag e functions. Furthermore, an uncertainty
analysis was performed comparing the assessed damage obtained through the use of JRC damage
functions and real, surveyed damage of the proposed case study in North -Western Romania, i.e.
Ilișua Basin, regarding the 2 006 flood event.
The proposed methodology has been applied at meso -scale, considering the Ilișua basin in
North -Western Romania. The meso -scale analyses are done at regional or basin scale having the
aim to support flood protection and mitigation strategie s, flood risk management as well as to
provide hazard and risk maps. This analysis can also be helpful to insurance companies and
decision -makers (de Moel et al., 2015; Albano et al., 2015). The land use data are aggregated in
classes that represent econom ic sectors. The asset values are specific, referring to the region or
basin that is analysed and they can be obtained from official statistics (Messner et al., 2007).
Complex models are used that need more detailed data, such as 2D hydraulic models, althou gh
they require more time and effort. Even though the meso -scale analysis offers accurate results and

62
detailed information, it still presents uncertainties regarding the accuracy of absolute damage
results (de Moel et al., 2015).
3.2. Study area
For this c ase study the flood that occurred in the Ilișua Basin in June, 2006 was con sidered.
This catchment, (Fig. 14 ), is located in North -Western Romania, has a surface of 353 km2 and a
mean altitude of 493 m. The main river is Ilișua with a total length of 52 km which is the right
tributary of the Someșul Mare River. The average slope of the Basin has a value of 15%, varying
from 28% up -stream to 4% down -stream. The flood that occurred on June 21st 2006 was
characterized by a peak flow of Qmax = 280 m3/s (calcula ted in a section located in the middle of
the basin). This corresponds to an occurrence probability of 0.7 – 0.8% (125 – 140 years return
period). The main cause of this flood was the extreme rainfall that exceeded 100 l/m2 (Sofronie et
al., 2013; Albano e t al., 2017 a).

Fig. 14 . Localization of the study area; a. Romania – georgaphical map; b. Ilișua
Catchment

The effects were severe, with 13 deaths and significant structural damages (Sofronie et al.,
2013). For this analysis six main affected villages were considered: Cristeștii Ciceului, Ilișua,

63
Căian, Lunca Borlesei, Spermezeu, and Borleasa. The real, surveyed damage data was presented
in the Coverage and Risk Assessment Plan of Bistrița -Năsăud County, (2006) of the County
Committee for Emergency Situations Bistrița -Năsăud. The total registered damages for the
analysed area was 1.1 million euro, as following: for buildings/urban, 194.000 euro; for roads,
687.000 euro; for agriculture, 127.000 euro (Albano et al., 2017 a).

3.3. Methodology
In order to quantify the potential damages that may occur in the study area, the approach
is following the next steps: hazard analysis, exposure and vulnerability analysis, and damage
analysis. The main steps of this analysis are presented i n Fig. 15 and described in the following
sub-sections.

Fig. 15 . Schematic representation of performed flood damage assessment (Albano et al., 2017 a)

64
In the hazard analysis, the water depth map is developed using the HEC RAS hydraulic
model and a DEM with a resolution of 5 m. The exposure is represented by the assets at risk, which
are classified usually based on economic sectors (i.e. buildings, infrastructure and agriculture). For
this study, the CLC datab ase was used. The susceptibility of the assets at risk relates the damages
of the assets at risk to flood characteristics, using damage functions (Albano et al., 2017 a). For the
present study the JRC depth -damage functions (Huizinga, 2007) were introduced in the FloodRisk
model in order to calculate the damages.

3.3.1. Hazard analysis
In the hazard analysis process, the flood extent and water depth are calculated. For this
purpose, the QGIS and HEC RAS software were used . The QGIS applications are ideal for the
processing and preparation of data that are subsequently used for hydraulic simulations (i.e. HEC
RAS modelling). The QGIS contains specific tools that are able to process and analyse the DEM
of an area in order to extract the information that are needed for the development of the hazard
maps. The input data (i.e. the topographic map, the cross -section of the river, the land use map
etc.) were provided by the Romanian Waters National Administration of Someș -Tisa Wat er
Branch. By using the 1:5000 topographic map as a background, the contour lines, the elevation
points and the river centreline were digitized. The elevations of the river thalweg were calculated
by interpolating the data between two subsequent contour li nes along the river centreline. Next,
using the "Raster – Interpolation" QGIS plugin, these data were used to create a DEM through the
triangular interpolation method (Cellsize X and Y of 5 m). Given the reduced number of river cross
sections available (9) , synthetic cross -sections were created based on the available DEM. The
resulting data were exported in HEC RAS using the Q -RAS tool of QGIS. Giving the fact that the
preparation of geometry data in HEC RAS is difficult and time -consuming, QGIS could be us ed
for this process, Q -RAS being a useful tool that can export the results in the hydraulic model.
The hydraulic modelling process is essential for the development of the hazard maps. The
hazard maps are important tools, constituting the base of flood con trol, land -use policies,
emergency situation management, etc. For the hydraulic simulations, the 1D hydraulic model HEC
RAS ( Fig. 16 ) was used and the flow profile was developed. The initial and boundary conditions
that were used are: Q = 280 m3/s, flow co nditions – critical depth upstream and normal depth

65
downstream, mixed flow regime. The discharge (flow) of 280 m3/s represents the estimated value
for the June 21st 2006 event at a stream gauge in the middle of the basin.

Fig. 16 . Representation of data analysis in HEC RAS

Fig. 17 . Flood hazard map for Iliș ua Basin developed in this study

By using the RAS Mapper tool from HEC RAS, the outputted data were used to construct
a 2D grid and a raster file containing gridded values of the maximum water dept h for every flooded

66
area. The RAS Mapper tool can produce a raster through spatial interpolation of the water level
outputs from the 1D -RAS simulations. By subtracting the elevations of the terrain in raster file
format from the spatial interpolation of wa ter level, it is possible to get the water depth raster. This
raster ( Fig. 17 ) represents the hazard map and is one of the input data used by the damage model.

3.3.2. Exposure and Vulnerability analysis
The exposure analysis refers to the elements at risk, which are obtained by overlapping the
information about assets present in the area (i.e. land -use map and roads map) with hazard
information (i.e. flood extent and water depth) (Fig. 18 ). In this way th e elements that are affected
by floods are identified. The vulnerability analysis shows how these elements will be affected,
depending on their characteristics (i.e. type of element, asset value).

Figure 18 . Identification of the exposed elements for th e city of Căianu Mic
(adapted after Albano et al., 2017 a)

At meso -scale, the assets with similar characteristics are usually in an aggregated form (i.e.
land use maps); in this case study the CLC database were used. For a more accurate analysis of
the u rban category, the buildings were digitized and introduced in the CORINE land use map ( Fig.

67
19). The urban category is the most affected class in the damage assessment, being responsible of
around 80% of the total damage (de Moel and Aerts, 2011). Therefor e, the accurate representation
of the buildings is essential in order to obtain reliable results.

Fig. 19 . The improved CORINE land use map developed for Ilișua basin

Afterwards, the asset v alue was determined for each CORINE land use class. The data
regarding the value of the exposed assets were provided by the local authorities of the respective
communities. Due to the fact that the event took place in 2006, and the documentation regarding
this event was made available in 2012, the value of the asset s needs to be corrected with inflation.
In order to achieve this, the GDP deflator was used (GDP -Deflator, 2018 ), which is a demonstrated
tool used to quantify price inflation. For the selected case study, the inflation adjusted values
(euros/m2) for each CORINE land use class exposed to flooding, are presented in Table 7 .

68
Table 7. Land use classes reclassification and site -specific assets value
for the selected case study (Albano et al., 2017 a)
Code Description CORINE land use classes Assets value (euro/m²)
112 Discontinuous Urban Fabric 108
122 Road and rail networks and associated
land 16
211 Non-irrigated Arable Land 0.08
231 Pastures 0.06
242 Complex Cultivation Patterns 0.16
243 Land Principally Occupied by Agriculture
with significant areas of natural
vegetation 0.13
311 Broad -Leaved Forest 0.04
312 Coniferous Forest 0.04
313 Mixed Forest 0.04
321 Natural Grassland 0.06
324 Transitional Woodland -shrub 0.06
331 Beaches Dunes and Sand Plains 0.06

For each land use class affected by flooding a corresponding depth -damage function is
associated. For the selected case study, the JRC depth -damage functions (Huizinga, 2007)
developed for different European Countries were used. The JRC developed this functions with th e
aim of offering a harmonised approach across EU for the estimation of direct damages produced
by floods. For each of the considered countries, depth -damage functions were p rovided for 5 land
use classes: residential buildings, industry, commerce, infrast ructure, and agriculture.
In order to use these functions with the CORINE land use map for this study, the functions
were adjusted by associating the depth -damage curves of Huizinga 2007 to the land use class that
was considered to be most comparable, as s hown in Table 8 . For example, for the land use class
"discontinuous urban fabric", the "residential building" class from JRC was attributed, along with
its corresponding depth -damage functions; and so on.
Major infrastructure elements such as roads, were i nputted manually, i.e. by digitizing the
roads from the topographic map provided by the Romanian Waters National Administration of
Someș -Tisa Water Branch, and were assigned to the “Road and rail networks and associated land”
class and their corresponding asset value provided by the Romanian Waters National
Administration of Someș -Tisa Water Branch.

69
Table 8. Correspondences between JRC depth damage curves and
CORINE land use classes (Albano et al., 2017 a)
CORINE land use classes JRC depth -damage curves
classification on the basis of
economic sectors
Discontinuous Urban Fabric Residential buildings
Road and rail networks and associated land Infrastructure
Non-irrigated Arable Land Agricultural
Pastures Agricultural
Complex Cultivation Patterns Agricultural
Land Principally Occupied by Agriculture
with significant areas of natural vegetation Agricultural
Broad -Leaved Forest Agricultural
Coniferous Forest Agricultural
Mixed Forest Agricultural
Natural Grassland Agricultural
Transitional Woodland -shrub Agricultural
Beaches Dunes and Sand Plains Agricultural

3.3.3. Damage analysis
The quantitative estimation of damages is usually done by using damage models. This
models commonly use depth -damage functions that combine hazard with vulnerability
information. For the flood event considered in this study, the damages were calculated using the
free and open source tool FloodRisk (Mancusi et al., 2015) developed in QGIS platfo rm. The tool
calculates both direct tangible damage (damage to structure) and direct intangible damage (loss of
life) caused by floods and visualizes them in form of tables, graphs and maps. Maps give a direct
and strong impression of the spatial distribut ion of the flood risk, providing essential information
to stakeholders (Albano et al., 2015b). For the estimation of direct damages in this study, the
following input data were used: the water depth map developed in the hazard analysis step, the
improved C orine land use map containing the asset value for each class and the JRC depth -damage
functions.

70
The FloodRisk plugin provides a dataset containing default functions of different models
collected from literature (e.g. Hazus Model, RAM, DSM, etc.). These fu nctions provided by the
plugin are associated with the Urban Atlas land use classification. However, in this study the
damages were calculated using the JRC functions developed for various European Countries (i.e.,
Belgium, Czech Republic, Germany, Netherl ands, Norway, Switzerland, UK) and the CLC
classification. For this purpose, the data provided in the JRC report (Huizinga, 2007) were
collected and processed in order to substitute the standard input data available in the FloodRisk
plugin. Therefore, the damages were calculated using seven different functions introduced in the
plugin ( Fig. 20 ).

Fig. 20 . The JRC functions for different countries, introduced in the FloodRisk plugin

In this way an uncertainty analysis regarding the functions’ transferabi lity in space was
possible. The aim of this analysis is to determine the effect on the results when applying different
damage functions from other counties in another context than the one for which they were
developed.
For each land use class described in Table 8 , the damage was calculated using the JRC
depth -damage functions and the asset value described above. These data were subsequently
compared with the real flood event of June 2006 in Ilișua Basin. Given the fact that the surveyed
data were collected for only 3 classes (urban, roads and agriculture) the results of the simulations
have been grouped according to these classes in order to ensure a consistent comparison between
the calculated damage assessment and real damage surveyed by authorities after the June 2006
flood.

71
In Table 9 , the damages calculated and the real surveyed damages for the Ilișua Basin,
reported in the Coverage and Risk Assessment Plan, 2006 of the County Committee for Emergency
Situations Bistrița -Năsăud, are presented. The simula tions with the JRC functions exhibit higher
urban damages than flood losses in agriculture. For infrastructure however, which represents a
significant percentage of the total damage, the JRC functions strongly underestimate the
corresponding real, surveyed losses. Overall, the model systematically overestimates the losses for
the urban and agricultural classes, highlighting the fact the estimations of infrastructure damage is
less developed and can be a large source of uncertainty in the modelling process.

Table 9. Results of damage calculation using different JRC depth -damage functions
and the reported damages (Albano et al., 2017 a)
Damages (MEuro)

urban roads agriculture Total
Belgium JRC 8.3 0.09 0.85 9.2
Czech Republic
JRC 5 0.14 0.69 5.8
Germany JRC 3.8 0.16 0.31 4.3
Netherlands JRC 3.2 0.10 0.81 4.2
Norway JRC 18.4 0.26 0.69 19.4
Switzerland JRC 13.8 0.13 0.58 14.6
UK JRC 32.4 0.13 1 33.6
Surveyed Ilișua
Basin 0.29 0.6 0.12 1.1

72

Fig. 21 . Graphic representation of the results of damage calculation using different JRC
depth -damage functions and the reported damages (adapted after Albano et al., 2017 a)

3.4. Uncertainty analysis
The results are characterized by a large variability ( Table 10 and Fig. 21 ), therefore an
uncertainty analysis regarding the damage functions’ transferability was conducted. The damage
values that were simulated using the FloodRisk tool and the reported damages of the 2006 flood
of Ilișua Catchment were compared. The reported dama ges were provided by the County
Committee for Emergency Situations Bistrița -Năsăud for Ilișua Basin for the 2006 flood.
Each damage value estimated for each of the JRC depth -damage functions was divided by
the reported damage value, resulting a ratio that was considered as the relative error. The relative
error, indicates error rate of the simulated damage respect to the surveyed damage. The results
showed that the Netherlands -JRC functions offers the lowest relative error (37%), while large
errors are indu ced by the Norway -JRC, Switzerland -JRC and UK -JRC functions ( Table 10 ).
However, a general overestimation of the damage results when using the JRC functions
can be observed, these findings being in agreement to other studies from literature (e.g. Merz et
al., 2004; Jongman et al., 2012). The large variation in the results and the large relative errors are
mostly related to the shape of depth -damage functions (Jongman et al., 2012, Messner at al., 2007,
de Moel and Aerts, 2011). For example, a great variati on in the results can be observed between 0510152025303540Damages for different land use classes
Urban Roads AgricultureMEuro

73
UK-JRC and Netherlands -JRC functions. This can be partially explained by analysing the shape
of this functions. In the case of the UK functions for the urban land -use class the damage factor
reaches 100% for a wate r depth of 3 m, while in the case of Netherlands functions, the damage
factor is around 20% for a water depth of 3 m ( Fig. 22 ).

Fig. 22 . Differences in the shape of damage functions (Netherlands -JRC and UK -JRC)

Table 10. Relative error and function uncertainty determination (Albano et al., 2017 a)
Relative
error Function
uncertainty
Total reported (MEuro) 1.1
Total
calculated
(MEuro) Belgium -JRC 9.2 8.37 2.21
Czech Republic –
JRC 5.8 5.28 1.39
Germany -JRC 4.3 3.92 1.03
Netherlands -JRC 4.2 3.79 –
Norway -JRC 19.4 17.53 4.63
Switzerland -JRC 14.6 13.18 3.48
UK-JRC 33.6 30.35 8.01

In order to identify the effect of the depth -damage functions and to quantify uncertainties
associated with the modelling of flood damage, the results calculated with the FloodRisk model

74
were analysed and compared. For this purpose, each calculated damage value was divided by the
lowest calculated value, the results being expressed in term of "function uncertainty" ( Table 10 ).
In the present case the lowest value is the one calculated through the JRC functions for
Netherlands, therefore, all the other estim ated values were divided by this one. The values of this
factor varies between 1.03 and 4.63 for all the functions, excepting the outlier value for the UK
JRC function which is equal to 8.01, i.e. the difference between the highest and lowest damage
estima tes.
These findings are consistent with others in the research field, suc h as the study done by
de Moel and Aerts, 2011, a function uncertainty factor of 5.6, the study by Cammerer et al., 2013 ,
which reported a factor of 2.3 for the cases when the models are thoroughly validated , the study
by Jongman et al., 2012 , based on two study areas reporting a factor of 3.7 and one of 10.5, and
finally, Scorzini and Frank, 2015, for which this factor varies quite extensively between 1.1 and
14.8.
The results also sh owed that the transferability of damage functions in space represents a
major problem in flood risk assessment, lea ding to great uncertainties which need to be
characterized and reduced when possible. This can be partially achieved by establishing a cohere nt
framework for data collection and evaluation.

3.5. C onclusions
In this case study a quantitative approach for flood damage assessment was presented. The
aim was to create a more comprehensive methodology that includes all flood risk factors in the
analysis and in this way to offer a better understanding of the risks that can occu r and help in the
decision -making process.
Furthermore, the quantitative economic damages estimation can be used in flood risk
estimation, if more flood hazard scenarios are used. This study is guided by the need for the
development of new methods and tool s that can efficiently support stakeholders in their
compliance with the Flood Directive2007/60/EC. In particular, this approach could be the initial
step to develop a rationale procedure to achieve objective results on which stakeholders can base
their de cisions for prioritizing investments, and performing cost -benefit analyses of mitigation
alternatives.

75
Moreover, the attempts in the past years to develop pan -Europeans damage models requires
more knowledge and more attention to the uncertainties regarding these models. Damage estimates
are influenced by many factors, first in the process being the input data. Therefore, a homogeneous
database across Europe is needed. Another important factor is the shape of depth -damage functions
which can lead to great un certainties.
Another objective of this case study was to perform an uncertainty analysis using the
damage functions of Huizinga, 2007 . It was found that more attention should be given to the shape
of the depth -damage functions used at EU level; these ones having a large effect on the damage
results. The calculated damages for the Ilișua catchment have a relative error which ranges from
37% to 300%, a ll the values being larger than the reported ones. The function uncertainty factor
obtained varies between 1.03 and 4.63. This variation on the results can be attributed to the shape
of the functions, the data used for the construction of these functions l ike synthetic or empirical
data, and weather or not the emergency service or stock costs are included.
The results show that overall applicability and transferability of depth -damage curves to
other geographical regions is still a major gap in current flo od damage modelling, but the
quantification of the uncertainties and its communication to stakeholders is the first step for the
maximization of quantitative risk approach effectiveness, towards flood risk management
objectives of the Flood Directive, ensu ring that risk information is robust, credible and transparent.

76
4. Case study 2. GIS tools for large -scale analy sis of direct economic flood damage. The case
study of Romania
4.1. Introduction
This increased growth and development of tools and models that can be used for flood
hazard and flood consequence analysis proved to be particularly useful for large -scale analysis that
has gained more attention in the past years. The flood risk assessment at large scale offers support
for national and global policies and are used in the prioritisation of investment at national level,
the principal stakeholders being the governments and the reinsurance companies. The governments
is using the hazard informat ion in order to implement flood damage reduction measures, for
emergency operations and for land use policies. On the other hand this information is useful for
insurance companies in order to establish the price for flood risk insurance.
The macro -scale a nalysis refers to large areas such as national or international scale.
Usually there are require d less detail data with a low resolution. The land use data are aggregated
and one general asset value is used for each class. C LC is a good example of database that can be
used at macro -scale analysis (Messner et al., 2007; de Moel et al., 2015).
However, the approaches at large scales present gaps given by the lack of consistent data
over large areas and the lack of high resolution data. The inconsistency of data is given by the use
of different information, models and methods in order to assess the flood risk for different areas.
Therefore, in order to have consistency in the results one meth od must be applied for the entire
area. This can limit the choice of the methods that can be used, simplified approached being
needed. Furthermore the level of detail of this analysis is different from one area to another,
depending on the availability of data . (de Moel et al., 2015; Alfieri et al.,2014; De Roo et al.,
2006). In many areas there are ungauged basins that provide little information and therefore the
hydraulic modelling is rather challenging and time -consu ming. In this context more simplistic
approaches were developed, using availab le data, such as the DEM, in order to extract the
necessary informations for the analysis of flood hazard . Moreover , the land use data available for
large areas have a course resolution , which can induce uncertainties in the results. However, in the
past y ears, many studies focused on using high -resolution EO datasets in order to extract land use
information.

77
The variability in space and time of the risk components (hazard, exposure and
vulnerability) present s another problem that must be considered when speaking of large scale
analysis.
In this work, a quantitative flood damage assessment methodology for data -scarce
environmnets is proposed, using data that are easily available . The flood damage analysis was done
for the entire Romanian territory, for a return period of 100 years, combining EO high -resolution
data (30 m resolution) and free and open -source GIS tools . Around 98% of the Romanian territory
is situated in the Lower Danube River Basin , therefore the study was focused on the Lower Danube
River and its tributaries. To the authors’ knowledge it represents the first large scale flood damage
analysis with a resolution of 30 m.
The main goal of this chapter is to offer a preliminary simplified approach and to provide
solutions for the current principals gaps in the flood risk assessme nt filed : limited availability of
data and quality/resolution of the data. The analysis is done at the grid resolution of 30 m using
innovative tools and methods.
4.2. Study area
The Danube river (Fig. 23 ), the second longest in Europe (approximately 2.860 km),
crosses on its way 10 countries (Germany, Austria, Slovakia, Hungary, Croatia, Romania,
Bulgaria, Republic of Moldova and Ukraine), from its headwaters in the Black Forest mountains
(Germany) and flowing into the Black Sea. The Danube river is divaded into three sections (FRMP ,
2015 ):
– The Upper Danube, from its headwaters to Bratislava city (Devin gate);
– The Middle Danube, from Bratislava to Bazias;
– The Lower Danube, from the entrance in Romania, at Bazias, to the Black Sea.
This study focuses on the Lower Danube section, which is situated on the Romanian
territory and has a length of 1.075 km. The Lower Danube is the most important regarding the
discharge (flow) and the navigation in this area and the hydrographic network includes most of
Romania’s rivers.
Adopting an integrated flood risk management is important particularly when speaking
about large rivers such as the Danube, the potential of damage occurrence being higher due to the
large number of population and economic assets situated in the flood prone areas. The extreme
events which took place in the Danube basin highlighted the importance of Flood Directive

78
(2007/60/CE) implementation as well as the importance of adopting a unitary and harmonized
methodology which includes the entire Danube basin.

Fig. 23 . Study area; a. Danube R iver – Europe location;
b. The Lower Danube – Romania

Several extreme events that took place in the Danube basin are described in detail by
Mikhailova et al., 2012 in a study regarding the impact of climate change in the Danube basin. The
floods from August 2002, caused by extreme precipitation, have had major consequences in the
Upper and Middle Danube, the consequences in the L ower Danube being reduced due to the
regulation of water capacity at Iron Gate I dam. The floods from 2006 have been caused by extreme
precipitation and the melt of the large snow accumulation, having major consequences on the entire

79
Danube basin and parti cularly on the Lower Danube, Romania being one of the most affected
countries (Mikhailova et al., 2012).
These events highlighted the fact that the existing protection measures have limitations,
their improvement being necessary as well as the improvement of population awareness policies
regarding the flood risk ( European Union, 2018 ).
A first attempt to realize a harmonized methodology on the entire Danube basin was done
by the Danube FLOODRISK cross -border project, initiated by the Ministry of Environment and
Waters Management in Romania (Danube -Floodrisk, 2018) . The main objectives of this project
were: the awareness of the population situated in the Danube basin regarding the flood exposure,
rising the population and assets protection level, priori tizati on of protection measures and decision
support. The main results are represented by the hazard and risk maps as well as the potential flood
damage calculation. The maps are represented on a scale of 1: 100.000. The hazard maps contain
the flood area and th e water depth (divided in 4 classes) and the risk maps contain the flood risk
areas (high, medium and low risk), for each risk class the potential damage being presented.
Furthermore, these damages are differentiated on land use types, the database that wa s used being
CLC . For the damage calculation, the damage functions were applied for each land use class
(European Union, 2018 ; FRMP , 2015).

4.3. Methodology
The proposed approach for the flood damage assessment is following three steps: hazard
analysis, exposure analysis and damage analysis (Fig. 24 ).
Regarding the hazard analysis, a simplified method ology for water depth estimation based
on the basin’s geomorphology was proposed. This method is u sing the DEM as the main input in
order to extract the ne cessary geomorphic features for the determination of the flood area and water
depth. This approach offer rapid and accurate results and can be applied in data -scarce areas. A
GIS-base model was applied in order to obtain the water depth using the Geomorphi c Flood Index
(GFI) (Samela et al., 2017).
For the exposu re analysis, the land use map for the study area was developed using the
Landsat 8 satellite images with a resolution of 30 m. A machine -learning algorithm was used for
the identification of 7 land use classes from the multi -spectral satellite images. In order to train the
algorithm, the Urban Atlas land use database was used as training data. The aim is to obtain high

80
resolution land use data over large areas, which can improve the accuracy of the flood damage
assessment.
For the damage analysis the FloodRisk model and the JRC depth -damage functions were
used. The damages were calculated for four scenarios using the existing hazard map and land use
data (JRC water depth map and CLC ) and the hazard and land use maps developed in this study.

Fig. 24 . Schematic representation of the applied methodology

Giving the fact that the hazard analysis method must be applied to a certain hydrographic
basin, the study area was divided into 5 sub -basins in order to be analysed :
– The Tisa sub -basin
– The Danube sector between upstream of the confluence with Timis river and
upstream of the confluence with Siret river, including the tributary sub -basins
– The Siret sub -basin
– The Danube sector between ups tream of the confluence with Prut river and the
Black See, including the tributary sub -basins
– The Danube Delta

81
4.3.1. Hazard analysis
In the hazard analysis process the flood prone areas are identified and the water depth for
this areas is calculated. For this purpose, a GIS -based model that is using the geomorphology of
the basin was applied (Samela et al., 2017) . This model allows the flood -prone area delineation
and wa ter depth calculation over large areas by using limited data. This is particularly useful for
the analysis in areas where the necessary data for hydraulic simulations are not available or this
simulation would be time consuming and expensive. The geomorpho logical features of the area of
interest are extracted using a DEM and based on this information the flood -prone areas are
delineated using a linear binary classifier . The DEM is the most important input data, the entire
analysis being based on the information extracted from it , therefore its quality and resolution will
have a great influence on the results.
The linear binary classifiers are basically separating the map in two classes: flooded areas
(assigning the code 1) and not flooded ar eas (assigning the code 0). For this purpose, a threshold
value for each of the morphological features is applied in order to separate the two classes. The
classifier must be trained in order to calibrate the threshold value. This step is done by using an
existing standard detailed hazard map that can be developed for small areas using hydraulic
simulations.
Some of the morphologic features that can characterize the flood hazard exposure are
presented below (Manfreda et al., 2015; Samela et al., 2017) :
“Single features
– Upslope contributing area, A(m2); upslope portion of the watershed that contributes to
water runoff to the point under exam.
– Surface curvature ∇2H [-]; Laplacian of the elevation.
– Local slope , S [-]; maximum slope among the eight possi ble flow directions that
connect the cell under exam to the adjacent cells.
– Flow distance to the nearest stream , D [m]; hydrologic distance from the location under
exam to the nearest element of the reference drainage network.
– Elevation difference to the nearest stream , H [m]; difference between the elevation of
the cell under exam and the elevation of the final point of the above -identified path.
Composite indices

82
– Ln (hl/H); this index compares in each point of the basin a variable water depth
hl with the elevation difference H. hl is calculated for each basin locationas a function
of the local contributing area, A l.
– Geomorphic Flood Index (GFI), ln(h r/H); this index compares in each point of the basin
a variable water depth hr with t he elevation difference H. hr is computed as a function
of the contributing area Ar in the nearest point of the drainage network hydrologically
conn ected to the point under exam”.
Based on the literature findings, the Geomorphic Flood Index (GFI) provides high accuracy
results, being the most suitable fo r flood -prone area delineation. In the study by Samela et al.,
2017, the GFI showed a good performance in the correct identification of flooded areas (the rate
of true positive RTP results being up to 89 %). It also has a limited variability, not being influenced
by changes in the input data (e.g. resolution, topography of the area), therefore it offers a good
transferability in other basins. In order to offer a simple way of implementation of this method a
new plugin, named Geomorphic Flood Area tool (GFA tool), was developed in QGIS (Samela et
al., 2018).
The GFA tool which is based on the GFI methodology was used in this study for water –
depth mapping for a return period of 100 years, over the entire territo ry of Romania. Samela et al.,
2018 performed a flood extend analysis for the territory of Romania using the GFI methodology,
therefore in this study this information and methodology will be used in order to further perform
a water depth analysis for this area.

4.3.1.1. The GFI method
The GFI method combines the geomorpho logical information extracted from a digital
description of the earth’s surface – a Digital Elevation Model (DEM) – with existing flood hazard
information from smaller areas in order to extend this information over large areas . The
Geomorphic Flood Index (GFI) is defined as the natural logarithm of the ratio between a variable
water depth hr and the elevation difference H (Samela et al., 2018).
𝐺𝐹𝐼 =𝑙𝑛(ℎ𝑟
𝐻) (1)
 hr represents the river stage in the point of the river network closest to the one under exam,
which is considered as the most p robable source of flood hazard;

83
 the parameter H represents the difference in elevation between the two above -mentioned
points.
The estimation of the river stage hr is derived as a function of the upslope contributing area
𝐴𝑟 using a hydraulic scaling relationship (see eq. 2) (Samela et al., 2018) :

ℎ𝑟≈𝑎𝐴𝑟𝑛 (2)

For th e study case of Romania , since paired values ( ℎ𝑟,𝐴𝑟) were not available to calibrate
the hydraulic scaling relationship, Samela et al.,2018 estimated the the exponent n using literature
values (Engelund and Hansen, 1967; Ibbi tt, 1997; Ibbitt et al., 1998; Knighton, 2014; Leopold et
al., 1965; Leopold and Maddock, 1953; Li, 1974; McKerchar et al., 1998; Park, 1977; Rodriguez –
Iturbe and Rinaldo, 1997; Smith, 1974; Whiting et al., 1999) . The mean value estimated is
n=0.3544. Thus, the hr was calculated by ‘neglecting’ a and assuming that the 𝑙𝑛 (𝑎) will be
included in the calibrated threshold.
In order to calibrate the classifier, the pan‐European flood hazard maps (Annex 1 ) derived
by Alfieri et al. 2014 for a return period of 100 years was used. The linear boundary between the
two classes of the binary classification (flood -prone areas and areas not prone to floods), must be
first calibrated for a training area that, according to Samela et al., 2017, must be equal or larger
than 2% of the basin of interest. In this way the optimat threshold τ is calibrated.
This parameter has to be calibrated iteratively by generating binary maps of potential flood –
prone areas that have been compared with the standard flood map assumed as “gold standard
truth”. From the comparison, the following performance measures can be assessed: the rates of
true positives (TP), false positives (FP), true negatives (TN), and false negatives (FN). The value
of τ that minimized the sum of the overestimation (R FP, false positive rate) and the underestimation
error (R FN, false negative rate), assigning equal weights to the two errors, can be considered as
optimal one and applied to the whole bas in.
The results of study done by Samela et al., 2018 are presented in Table 1 1, showing that
the GFI classifier has a good performance.

84
Table 11 . Values of the optimal thresholds and relative performance measures obtained in
calibrating the classifi er over the investigated basins (Samela et al., 2018)
Basin
number % training
area τ RTP RFN RTN RFP Obj.
function
= R FN+R FP AUC
1 14% -0.23 81.9% 18.1% 64.6% 35.4% 53.4% 0.81
2 12% -0.24 85.0% 15.0% 83.3% 16.7% 31.7% 0.92
3 14% -0.12 90.6% 9.4% 36.9% 63.1% 72.5% 0.63
4 6% -0.26 82.6% 17.4% 84.0% 16.0% 33.4% 0.92
5 5% -0.31 91.1% 8.9% 91.7% 8.3% 17.2% 0.97

4.3.1.2. Estimate of the water inundation depth using the Geomorphic Flood Index .
Methodology.
In this large -scale analysis, it was tried to exploit at the most the information derived from
the analysis of the basin morphology, in order to derive an instantaneous estimate of the maximum
inundation depth within the flood -prone areas previously identified in the Romanian territory by
Samela e t al., 2018 .
As mentioned in the previous section, the hr was calculated by ‘neglecting’ a and assuming
that the 𝑙𝑛 (𝑎) will be included in the calibrated threshold. In this sense, the GFI may be expressed
as:
𝐺𝐹𝐼 =𝑙𝑛(ℎ𝑟
𝑎𝐻) (2)
Nevertheless, after the boundary of decision between flood -prone areas and areas not prone
to floods that characterizes each specific basin was calibrated , along that boundary of decision the
below information were obtained :
𝑙𝑛(ℎ𝑟
𝑎𝐻)=𝜏 and ℎ𝑟
𝐻=1 (3)
(1
𝑎)=exp (𝜏) → 𝑎= (1
exp (𝜏)) (4)
In this way, the values of the initially estimated river stage ℎ𝑟 ′=𝐴𝑟𝑛 were corrected:
ℎ𝑟=ℎ𝑟′∗𝑎=ℎ𝑟′
exp (𝜏) (5)
Therefore , for each river network element resulted an estimate hr. From the geomorphic
analysis previously performed, there were obtained information about all the upslope basin
locations that drain into those elements and the difference in elevation that separates the locations

85
from the closest river. Therefore, within the delineated flood -prone areas, finally there were used
the hr values to estimate the water depth (WD) as follows:
WD = h r – H (6)
Based on this methodology and using the GFA tool the water depth map for the entire
territory of Romania was developed. The input data required by the GFA tool are: a. DEM,
depressionless DEM, flow direction and flow accumulation (for the GFI index determination); b.
a standard flood map . The framework of the methododlogy is presented in Fig. 25.

Fig. 25. Schematic representation of the hazard analysis methodology

86
The main input was represented by the the SRTM 1 Arc -Second Global elevation data
obtained from USGS. These DEMs provides worldwide coverage of void filled data at a resolution
of 1 arc -second (30 meters). The resolution of the DEM is an important factor, a low resolution
inducing error that may affect the analysis. Therefore, it is recommended the use of a filled DEM,
which was obtained using GIS tools. With this procedure, the sinks from the original DEM are
filled in order to obtain a proper river network topology.
Using the obtained filled DEM the flow direction and flow accumulation were calculated
(Annex 2 ). The flow direction shows the direction in which the water of one cell will flow and
subsequentl y the slop is calculated. Further, the flow accumulation is obtained, showing the flow
that is accumulated in each cell and in this way the stream can be determined.
The obtained hazard map ( Fig. 26 ) contains the extend of the flood and the water depth for
the entire territory including the second and minor rivers which usually are not considered in large
scale analysis. As is highlighte d in Fig. 27, the hazard map obtained using the GFA tool can
identify and analyse all the streams of the basin. Therefore, it is offering a more accurate hazard
map, comparing with the existing JRC pan -European map which is considering just the main
streams. The main advantage s of this approach are represented by the f act that it is a simple method
that does not requires detail data and costly resources to be applied, it can be used for areas with
limited availability of data and it can be applied for large scale analysis . Furthermore, the analysis
is done using a resol ution of 30 m which improve the accuracy of the results.
It must be mentioned that this kinds of procedures cannot replace a comprehensive
hydrological -hydraulic study, since they have several limitations, among them, it cannot take into
account anthropog enic modifications (e.g. soil sealing and presence of floods defence structures)
and neither can describe the reduction of the inundation depth once the water overflows the banks
and spreads over the adjacent land. Nevertheless, in case of unavailability o f data to perform a
detailed analysis, geomorphic procedures characterized by simple data requirement can be very
useful for local authorities and planners to realize management strategy.

87
Fig. 26 . Water depth map developed in this study (GFI hazard map)

88
Fig. 27 . Comparison betwee n the JRC hazard map (Alfieri et al., 2014)
and the GFI hazard map obtained in this study

4.3.2. Exposure analysis . Data and methods
The exposure analysis refers to the risk susceptible elements, which are obtained
overlapping the information regarding the economic assets from an area (land use map,
infrastructure map) and the information regarding the hazard (flooded area map, water depth map).
The current land use datasets available for large areas (such as Corine land cover) present
a course resolution inducing uncertainties in the flood damage analysis. Furthermore, due to the
rapid urbanisation that took place in the past years, up -to-data information become unavailable for
many regions. Even though detailed map s (such as the Urban Atlas datasets) do exist, they cover
certain areas or cities and therefore they cannot be used for large scale applications. In this context
a. JRC hazard map
(Alfieri et al., 2014)
b. GFI hazard map

89
the latest studies in the field (Biro et al., 2013; Liu et al., 2014; Wieland et Pitore, 2016) highlighted
the potential of satellite remote sensing data to provide information on the land use distribution .
For this study , a new land used map was developed from multi -spectral satellite images.
For this purpose the Landsat 8 data with a re solution of 30 m were used and machine -learning
classification algorithms were applied in order to identify the land use classes for the study area.
The method is using an object -based approach, which means that the analysis is done for
segments of an image and no t for each pixel. These segments are characterised by different features
that can be analysed using a machine learning classifier in order to detect a certain object, in this
case a land use class. This method is following three steps: satellite image proc essing, image
segmentation and feature -based description using training data and classification.

4.3.2.1. Satellite images processing
For this study 26 satellite images from the period June 2016 – September 2016 have been
acquired ( USGS, 2018a ), covering the entire territory of Romania . For data homogeneity, it is
important that all the images to be acquired from the same period .
Each image is composed of 11 bands, containing as well a Quality band (QA, Quality
Assessment). This band can be utilized to ind icate the pixels quality and to identify the pixels
covered by clouds. Each pixel has a value indicating the presence or the lack of the parameters
taken into consideration (clouds, snow/ice, water, etc.), as well as the level of confi dence of this
algorit hm (Table 12 .) (USGS, 2018 ).
The satellite images acquired for this study were selected to have a quality as good as
possible, containing a small area covered by clouds. This was done through image visual ization,
for example in Fig. 28 the lighter pixels indicated the fact that those surfaces are covered by clouds.
Thus, a number of images have been selected, which if they are overlapped provide a image with
a smaller cloud coverage , by selecting the pixel s with better quality . Further, each image (286
raster files – 26 image x 11 bands) has been processed using the QA band in order to exclude the
pixels covered by clouds from the images.

90
Pixel
value Cloud Cirrus
Conficence Snow /Ice
Confidence Water
Confidence Terrain
Occlusion Dropped
Frame Pixel description
2 Not
determined Not
determined No No No Yes Dropped pixel
20480 No No No No No No Clear Terrain
20484 No No No No Yes No Clear Terrain with Terrain Occlusion
20512 No No No Maybe No No Water terrain
23552 No No Yes No No No Snow/Ice terrain
28672 No Yes No No No No Cirrus clouds
31744 No Yes Yes No No No Cirrus clouds or Snow/Ice terrain
36864 Maybe No No No No No Mid-confidence cloud
36896 Maybe No No Maybe No No Mid-confidence cloud or Water terrain
39936 Maybe No Yes No No No Mid-confidence cloud or Snow/Ice
terrain
45056 Maybe Yes No No No No Mid-confidence cloud with Cirrus
clouds
48128 Maybe Yes Yes No No No Mid-confidence cloud with Cirrus
clouds or Snow/Ice terrain
53248 Yes No No No No No High -confidence clouds
56320 Yes No Yes No No No High -confidence clouds or Snow/Ice
terrain
61440 Yes Yes No No No No High -confidence clouds with Cirrus
clouds
64512 Yes Yes Yes No No No High -confidence clouds with Cirrus
clouds or Snow/Ice terrain
Table 12 . Examples of pixels values and their meaning for QA band (Landsat 8) ( USGS, 2018 )

91

Fig. 28. Landsat 8 satellite image, band 1
The first step was to process the QA bands ( Fig. 29 ) for each image, using the Raster
Calculator from Gdal plugin (QGIS software). Thus, the values below the value 53248 (which
represent the smallest value of the pixels covered by clouds – Table 6) have been replaced by 1
and the higher values have been re placed by 0. In this way a raster with two class of values was
obtained:
– 1 – for the surfaces that are not covered by clouds
– 0 – for the surfaces covered by clouds ( Fig. 29 , b)

Fig. 29. QA band processing

92
Therefore, multiplying each band of one imagine with the processed QA band (with values
of 1 and 0), the pixels covered by clouds will be excluded. The re sult is represented in Fig. 30 .

Fig. 3 0. Satellite image, band 1, with no clouds covered areas
Given the fact that the panchromatic band (band 8) has a resolution different (15 m) from
the rest of the bands, these bands were resampled to 30 m in order to be used for further analysis.
For the reclassification t he r.resamp.stats. function from GRASS , QGIS was used.
Moreover , the 286 raster files have been organised into categories of bands (band 1, band
2, etc.). Thus, each band contains 26 raster files which will be overlapped and united in order to
form an image of the entire surface of Romania. Due to th e extended surface of Romania, the raster
files have different coordinate reference system as following:
– 11 images have the EPSG: 32634 coordinate reference system
– 14 images have the EPSG: 32635 coordinate reference system
– 1 image has the EPSG: 32636 coordinate reference system
The images with the same coordinate reference system have been merged using the SAGA
(System for Automated Geoscientific Analyses) tool from QGIS software, obtaining three images

93
with three different coordinate reference system. This images have been re -projected in the EPSG:
3035 coordinate reference system and merged in o ne image. This process was repeated for each
band, resulting 11 images of the Romanian territory (Annex 3).

4.3.2.2. Image segmentation
In this process, basically the p ixels are grouped into segments that are further analysed. For
this purpose, an algorithm , called gr aph-based segmentation algorithm, is used in order to
determine the boundary betwe en two segments (Felzenszwalb and Huttenlocher , 2004). For this
purpose the algorithm compares the featur es values of the pixels within one segment with the
features values of pixels outside of the segment. The segmentation parameters have been adjusted
using reference data, selecting the param eters that had the best performance. This step is important
in ord er to define the scale of analysis. A good quality segmentation implies that the image
properties within a segment are homogeneous having a low variability, while the image properties
between segments are heterogeneous.

4.3.2.3. Feature -based description and classification
The image features are used to describe the segments, the classification process being done
based on these features. In this study 44 image features were used , categorised on three types :
spectral features, textural features and spectral band indexes .
The spectral features were represented by:
– mean spectral value in the image band (mean_b1, mean_b2, mean_b4, mean_b5,
mean_b6 and mean_b7);
– standard deviation of the brightness value in the image band (sdt_b1, sdt_b3);
– mean val ue of the Normalized Difference Vegetation Index (mean_NDVI);
– minimum brightness value in the image band (min_pixel_b1, min_pixel_b3,
min_pixel_b9 and min_pixel_b10);
– maximum brightness value in the image band (max_pixel_b1, max_pixel_b4,
max_pixel_b6, max_pixel_b9 and max_pixel_b10);
– Weighted brightness (WB).
The textural features were represented by:

94
– Homogeneity derived from the GLCM * in the band (homogeneity_b1,
homogeneity_b4 and homogeneity_b6) ;
– Contrast derived from the GLCM * in the band (contrast_b1, contrast_b2, contrast_b4
and contrast_b7) ;
– Dissimilarity derived from the GLCM * in the band (dissimilarity_b1, dissimilarity_b2
and dissimilarity_b6) ;
– Angular Second Moment derived from the GLCM * in the band (asm_b7) ;
– Mean derive d from the GLCM * in the band (glcm_meani_b1, glcm_meani_b3,
glcm_meani_b4, glcm_meani_b5, glcm_meani_b6, glcm_meani_b7, glcm_meani_b9
and glcm_meani_b10) ;
– Variance derived from the GLCM * in the band (glcm_variancei_b10) .
*GLCM (Gray -Level Co -occurrence M atrix)
The spectral band indexes are represented by:
– Mean value of soil adjusted vegetation index, SAVI (savi_mean);
– Mean value of normalised difference vegetation index, NDVI (ndvi_mean);
– Mean value of modified normalised difference water index, MNDWI (mndwi_mean);
– Mean value of normalised difference built -up index, NDBI (ndbi_mean);
– Standard deviation value of normalised difference built -up index, NDBI (ndbi_std)
(Wieland and Pittore, 2016) .
For the cl assification of land use classes using the Landsat 8 satellite imagery data, an
Artificial Neural Network (ANN) called Multi -layer perceptron (MLP) was applied, using as
training data the available Urban Atlas land use maps for Romania. The MLP is using th ree types
of layers, each of them containing processing nodes interconnected to each other but not within
the same layer. The MLP used in this study contains one input layer, two hidden layers and one
output layer. The input layer is represented by the ban ds of the imagery; the hidden layers are used
for computations purposes; the output layer contains different codes, representing the land use
classes. The ANN calculated the weights of each layer node . Each node contains the weighted
input from all of the nodes connected to it from the previous layer. The selection of the features is
based on the weights assigned by the algorithm to this features.
In order to be able to recognize the land use classes, the MLP algorithm must be trained.
For this purpose, the back -propagation learning algorithm is used. First, based on the features

95
weights, the training data are introduced in the algorithm in order to identify the pattern. In the
second step, the error, which is the difference between the known and the estimated value, is send
backwords through the network. In this way the weights of the nodes are adjusted, in order to
minimize the error . This process is repe ated until the error become acceptable/minimal ( Rumelhart
et al., 1986).
In this s tudy 8 neural networks were used, each of them receiving the 44 features . One
cascade binary classification is done:
– 1: built -up area – non built -up areas
– 2: urban – non urban
– 3: vegetation – water
– 4: agriculture – forest
– 5: industrial – infrastructure
– 6: continuous urban – discontinuous urban
– 7: water – non water
– 8: built -up area – vegetation

4.3.2.4. Training data. Land use data processing
In order to train and calibrate the algorithm for the recognition of land use classes, the
Urban Atlas land use data were processed and used as training data.
Urban Atlas database
For this study , the available data for 14 cities of Romania (Bucarest, Cluj -Napoca,
Timișoara, Craiova, Brăila, Oradea, Bacău, Arad, Sibiu, Târgul Mureș, Piatra Neamț, Călărași ,
Giurgiu, Alba -Iulia) were acquired form European Environment Agency site. The Urban Atlas
database has a large number of land use classes providing detailed data, however this data are
available just for large urban areas.
For the homogeneity of the dat a, the Urban Atlas classes have been reclassified , reducing
their number ( Table 13 ). Further, this vector files have been converted into raster files with a
resolutio n of 30 meters. In Fig. 3 1 the reclassified Urban Atlas land use for Cluj -Napoca city of
Romania is represented in vector and raster format.

96
Table 13 . Reclassification of Urban Atlas classes
Urban Atlas Code Description Reclassification
code Description
11100 Continuous Urban Fabric (S.L. >
80%) 111 Continuous urban
fabric 11210 Discontinuous Dense Urban
Fabric (S.L.: 50% – 80%)
11220 Discontinuous Medium Density
Urban Fabric (S.L.: 30% – 50%) 112 Discontinuous urban
fabric
11230 Discontinuous Low Density
Urban Fabric (S.L.: 10% – 30%)
11240 Discontinuous Very Low
Density Urban Fabric (S.L. <
10%)
11300 Isolated Structures
12100 Industrial commercial public
military and private units 121 Industrial,
commercial, public,
military, private units
12210 Fast transit roads and associated
land 122 Infrastructure
12220 Other roads and associated land
12230 Railways and associated land
12300 Port areas
12400 Airports
13100 Mineral extraction and dump
sites
13300 Construction sites
13400 Land without current use 200 Agricultural and
green areas 14100 Green urban areas
14200 Sports and leisure facilities
20000 Agricultural + Semi -natural
areas + Wetlands
30000 Forests 300 Forests
50000 Wetlands 500 Water

97

Fig. 31. Urban Atlas land use of Cluj -Napoca city ;
a. Vector file of Urban Atlas land use;
b. Raster file of Urban Atlas land use

98
4.3.2.5. Validation and c lassification results
In order to assess the performance of the algorithm, the overall accuracy was calculated.
The validation of performed for 10 cities outside outside Romania, for which the Urban Atlas data
were available (Bratislava, Budapest, Debrecen, Kosice, Presov, Misk olc, Plovdiv, Sofia, Szeged
and Varna). The overall accuracy was calculated by dividing the number of the correct
classifications to the total number of segments. The results showed a good accuracy of the
algorithm, the ac curacy ranging from 0.68 to 0.83 (Annex 4 ). An example of the input data used
for validation and output data of the algorithm is presented in Fig. 32.

Fig. 32. Input and output data for the validation of the algorithm.
Sofia city, Bulgaria (accuracy 0.76)

The land use classes were determined for the area of i nterest – the flood prone areas . In
this way , seven land use classes were obtained: continuous urban, discontinuous urban, industrial,
infrastructure, agricultural, forests and water (Fig. 33 ).

99

Fig. 33 . Land use map developed for the flood prone area using
Landsat 8 satellite imagery

4.3.3. Damage analysis
In order to calculate the possible direct tangible damages at national level, the free and
open source FloodRiks plugin was used. This plugin calcula tes the economic damages and the loss
of lives and can be used for different scales analysis allowing to the users to apply different depth –
damage functions available in its repository or to enter new functions. In this study the JRC depth –
damage functions were used as well as the maximum damage values provided by JRC (Huizinga
et al., 2017). The main input data that were used for this study are:
– hazard data containing the water depth ;
– exposure data that are represented by the land u se map that contain information regarding the
location of exposed elements and the maximum damage value of this elements.
The damages were calculated for four sets of input data:
– JRC water depth and C orine Land Cover (CLC)
– GFI water depth and Landsat 8 land use

100
– JRC water depth and Landsat 8 land use
– GFI water depth and CLC
The JRC water depth is the pan -European hazard map developed by Alfieri et al., 2014;
the GFI water depth represents the hazard map developed in this study using the GFI methodology;
the Landsat 8 land use represents the land use map developed in this study using Landsat 8 satellite
images.
The JRC depth -damage functions provided by the FloodRisk plugin are associated to the
land use classes of Urban Atlas database. In order to have a c orrect association between this
functions and the land use classes used in this analysis, a set of reclassification must be done. First,
to each of the CLC class a corresponding Urban Atlas code is associated . The land use map
obtained usi ng the Landsat 8 data (Landsat 8 land use map ) contains seven classes (continuous
urban, discontinuous urban, industrial, infrastructure, agricultural, forests and water ). Therefore ,
in order to have consistent comparable results regarding the damages of the four considered
scenarios , the Urban Atlas codes associated to the CLC classes were further grouped in seven
classes ( Table 14 ). The description of the CLC and Urban Atlas codes is presented in Table 3 and
Table 4 . This codes were further associated to the corresponding classes of Landsat 8 land use map
(Table 15 ).

Table 14 . Reclassification of CORINE land use classes
Corine code Urban Atlas
code Urban Atlas
reclassification
code Description of Urban Atlas codes
111 11100 11100 Continuous Urban Fabric (S.L. > 80%)
112 11220 11220 Discontinuous Medium Density Urban Fabric (S.L. :
30% – 50%)
121 12100 12100 Industrial, commercial, public, military and private
units 131 13100
132 13100
133 13300
122 12220 12220 Other roads and associated land/ Infrastructure
123 12300
124 12400
141 14100 20000 Agricultural + Semi -natural areas + Wetlands

101
142 14200
211 20000
212
213
221
222
231
242
243
321
322 13400
331
332
333
411
412
311 30000 30000 Forests
312
313
324
511 50000 50000 Water bodies
512

Table 15 . Correspondence between Landsat 8 land use classes and Urban Atlas codes
Description Urban Atlas code
Continuous urban 11100
Discontinuous urban 11220
Industrial 12100
Infrastructure 12220
Agricultural 20000
Forest 30000
Water 50000

102
For each land use class a value was determined, representing the asset value. For this
purpose the data provided by Huizinga et al., 2017 were used . In this study a consistent set of
maximum damage values was developed for different countries and for various damage classes
(Huizinga et al., 2017) This data are representative of the year 2010, therefore in order for the data
to be accurate f or the year 2017, the inflation was calculated and the data were adjusted ( Table
16). During the analysis it was observed that the continuous urban class of CLC is covering small
areas, not being present in the flood prone areas, therefore the same assets value was associated to
both urban classes. The CLC does not have an accurate distinction between continuous and
discontinuous urban, most of the urban areas be ing considered discontinuous. On the other hand
the Landsat 8 land use map differentiates the two classes for all the urban areas. Therefore, it is
not possible to used different values for the two classes, because in the case of CLC map the
damages for ur ban will be calculated with the smaller value (the one of discontinuous class) while
in the case of Landsat 8 map the damages will be calculated with different values for the two
classes.
Table 16 . Assets value for the study area
Code Land use class Assets value (euro/m2) Adjusted assets value (euro/ m2)
11100 Continuous urban 417 495
11220 Discontinuous urban 417 495
12100 Industrial 561 667
12220 Infrastructure 9.44 11.2
20000 Agricultural 0.06 0.07
30000 Forest 0.04 0.04
50000 Water 0 0

4.4. Results and discussions
For the first scenario of flood damage assessment, the next data were used: the water depth
map provided by JRC ( JRC water depth ) and the Corine land Cover (CLC) database from 2012 .
The JRC water depth map represents the flood events with 100 years return period for the entire
Europe and has a resoluti on of 100m (Alfieri et al., 2014 ). As demo nstrated by Alfieri et al., 2014 ,
the water depth map showed a good performance, it can be used for analysis at large scales. The
data pro vided by CLC are available at a scale of 1:100000 and include 44 land use classes. For the

103
second scenario the hazard data were represented by the national water depth map with a resolution
of 30 m developed in this study ( GFI water depth ). For the exposur e data , the map developed from
the Landsat 8 images with a resolution of 30 m ( Landsat 8 land use ) was used . The results are
presented in Table 17 and Fig. 34 .
Table 17 . Damage results for the next scenarios: a. JRC water -depth and CLC ;
b. GFI water -depth and Landsat 8 landuse
Land use class JRC – CLC damages (Meuro) GFI – Landsat 8 damages
(Meuro)
Urban 101155.47 283452.07
Industrial 37718.22 445927.42
Infrastructure 120.74 3300.41
Agricultural 749.87 214.02
Forests 21.56 25.76

Fig. 34 . Graphic representation of damage results for the next scenarios:
a. JRC water -depth and Corine land use; b. GFI water -depth and Landsat 8 landuse

For the first scenario (JRC – CLC ) the results showed that 72% of the total damages are
represented by the damages from urban areas, this sector being the most affected, while the
damages in the industrial areas represent 27% of the total damages. For the second scenario (GFI
– Landsat8) the greatest damages are represented by the industrial areas with a percen tage of 61%,
while the damages in the urban areas represent 39% of the total damages.
The second scenario shows an overestimation of the damages, particularly in the urban and
industrial areas. However , this can be explained by the use of a more detailed flood map that was
developed in this study, using the GFI method. This method provided flood maps for all the rivers,
72%
27%
0%
1%
0%JRC_CLC
Urban
Industrial
Roads
Agricultural
Forests
39%
61%
0%
0%
0%GFI_Landsat 8
Urban
Industrial
Roads
Agricultural
Forests

104
including the small stream, which are not considered in the flood analysis maps of JRC, therefore
in the case of GFI flood map more areas are affected by floods and in conclusion greatest damages
are reported .
Furthermore, in the second scenario the damage value registered for industrial area is
higher than the urban damage value. This fact can have two causes : (1) the GFI water depth map
is affecting a larger industrial area compared with the JRC map; (2) the Landsat 8 land use map is
overestimating the areas covered by industrial, comparing with CLC map.
In order to determine which of the input data is causing the greatest differences in the
results, another two scenarios were simulated: GFI and CLC; JRC and Landsat 8. The results are
presented in Table 18 and Fig. 35 .
Table 18 . Damage result s for the next scenarios: c . GFI water -depth and Corine land use;
d. JRC water -depth and Landsa t 8 landuse
Land use class GFI – CLC damages
(Meuro) JRC – Landsat 8 damages
(Meuro)
Urban 242954.451 74826.2565
Industrial 67540.5823 196963.496
Infrastructure 141.953 3546.7985
Agricultural 800.126 703.0057
Forests 31.1435 30.6149

Fig. 35 . Graphic representation of damage results for the next scenarios:
c. GFI water -depth and Corine land use; d. JRC water -depth and Landsat 8 landuse

In Fig. 36 , a comparison of the damage values calculated for the four scenario s is presented.
78%
22%
0%
0%
0%GFI_CLC
Urban
Industrial
Roads
Agricultural
Forests
27%
72%
1%
0%
0%JRC_Landsat 8
Urban
Industrial
Roads
Agricultural
Forests

105

Fig. 36 . Damage value comparison between the f our scenarios.

The results showed that when keeping a constant land use and apply different hazard map s
(GFI and JRC), there are no significant differences in the percentage of damage values . For instant,
when using th e Landsat 8 land use together with JRC and then with GFI, a 9% difference is
observed between the 2 scenarios for industrial areas. For the CLC simulations with JRC and GFI,
the difference is 5%. This results indicate that the choice of hazard map does not have a great
influence on the damage results. Moreover , the simulations done with CLC show a larger percent
of damages in the urban areas, while the simulations done with Landsat 8 have a larger percent of
damages in inductrial areas.
On the other hand, when a constant hazard map is used with different land use maps (CLC
and Landsat 8) a great difference in the results can be observed. For example , when using the JRC
hazard map with CLC land use map and then with Landsat 8 land use map, the difference in
percentage of industrial damages is 45%. For the simulations with GFI hazard map and CLC land
use and then with the Landsat8 land use map, the difference between industrial damages is 39%.
For both cases when using the Landsat 8 map (Lan dsat 8 -GFI; Landsat 8 -JRC) the industrial
damages are greater than the urban ones. This fact indicate that the Landsat 8 land use have a
greater influence on the industrial damage results, overestimating them.
Regarding the urban dama ges, the results show ed a good accuracy between scenarios.
When comparing same hazard map with different land use the results are similar; the difference

106
that can be observed in the urban damages is the fact that the scenarios using the GFI maps have
overestimated results; how ever this situation was previously explained by the extended flood area
of GFI map (Fig. 37 ).

Fig. 37. Comparison between urban damages
Rega rding the industrial areas , the use of CLC (in two scenarios) shows similar results,
with a small overestimation when it is combined with GFI hazard map. However, the use of
Landsat 8 land use show an overestimation in both cases, the use of GFI hazard map increasing
even more the damage value ( Fig. 38).

Fig.38. Comparison between industrial damages

107
The differences caused by the land use maps may be given by the resolution of the data.
The CLC has a resolution of 100 m, offering a coarse classification of the land use. The Landsat 8
land use map has a resolution of 30 m, offering therefore more accurate results. When the maps
are analyzes by simple visualization, it can be observed that the Landsat 8 contains a larger
industrial areas compared to CLC, this fact being a result of the higher resolution of Landsat 8 land
use map.
In order to prove this hypothesis, that larger industrial areas are identified when the
resolution of land use data is higher, the classification of CLC (100 m resolution) was compared
with the classification of Urban Atlas (10 m resolution) . A confusion matrix was apply for all areas
in Romania where Urban Atlas data were available (35 cities ). The results showed a general good
accuracy, however when just the industrial class is considered, the results showed a low accuracy
(Table 19 ). This demonstrates that the CLC fails to correctly identify the industrial area and
therefore the damages resulted using this land use map ma y present large uncertainties.
Table 19 . Results of confusion matrix between C LC and Urban Atlas
The accuracy between C orine land cover and Urban Atlas land use
Analyzed cities Total accuracy Urban accuracy Industrial accuracy
Alba Iulia 0.86 0.79 0.29
Arad 0.90 0.87 0.39
Bacău 0.81 0.89 0.44
Baia Mare 0.87 0.78 0.61
Bârlad 0.91 0.92 0.43
Bistrița 0.85 0.60 0.44
Botoșani 0.90 0.76 0.54
Brăila 0.84 0.89 0.51
Brașov 0.90 0.77 0.53
București 0.81 0.87 0.62
Buzău 0.85 0.79 0.43
Călărași 0.91 0.87 0.72
Cluj-Napoca 0.81 0.78 0.52
Constanța 0.81 0.87 0.62
Craiova 0.83 0.80 0.62
Drobeta -Turnu Severin 0.85 0.78 0.60
Focșani 0.88 0.83 0.30
Galați 0.90 0.87 0.63
Giurgiu 0.80 0.82 0.68
Iași 0.84 0.78 0.48
Oradea 0.77 0.76 0.50

108
Piatra Neamț 0.87 0.84 0.66
Pitești 0.79 0.83 0.51
Ploiești 0.85 0.78 0.64
Rîmnicu Vâlcea 0.81 0.76 0.67
Roman 0.88 0.89 0.57
Satu Mare 0.91 0.81 0.43
Sibiu 0.88 0.83 0.32
Slatina 0.80 0.83 0.68
Suceava 0.85 0.79 0.46
Târgoviște 0.80 0.77 0.52
Târgu Mureș 0.80 0.79 0.41
Timișoara 0.80 0.87 0.46
Tulcea 0.75 0.84 0.63

It is acknowledge the fact that the analysis at large scales have the tendency of
overestimating the damages due to the fact that the data are aggregated and the values of the assets
are generalized and uniform. This gap may be solved by improving the reso lution of the data that
are used (such as the DEM, water depth, land use) at large scales. The use of local detailed datasets
for the assets value may also improve the accuracy of the results, however this data are not
harmonized for large areas and they m ay not be available. In fact, the free availability of data
represent an important problem in this research field inducing great uncertainties in the process of
flood damage assessment.
In this study a methodology for flood damage assessment at national scale was proposed,
using free and open source data and tools with the aim of providing an improved and simplified
method of flood damage assessment in data -scarce environments . The advantage of this
methodology is the fact that it can be applied over large areas with limited availability of data. The
study proposed innovating and up to date methods for all three steps of flood damage assessment:
hazard, exposure and vulnerability analysis.
Furthermore, the resolution of the data (water depth and land use) used in the second
scenario may improve the accuracy of the results, showing that constructed area s may be more
susceptible to floods and therefore more attention must be given to the urban planning, f lood risk
management and risk reduction measures.

109
4.5. Conclusions
In this paper, a large -scale flood damage assessment methodology for data -scarce
environments is proposed combining EO high -resolution data (30m resolution) and free and open –
source GIS tools. The main goal of this study is to offer a cost -effective and parsimonious approach
that can provide solutions for the current principal gaps in large -scale flood risk assessment and
could support different stakeholders in their compliance with risk map delineation and the
management of current and future flood risk.
Compared with the current approaches in the field, the proposed me thodology have some
advanta ges. First, a GIS-base model that is using the morphology of the basins in order to calculate
the water depth is applied. T his method can be applied in data -scarce environments and it consider
in the analysis all the streams of the basins, the existing large -scale hazard maps considering just
the main rivers. Second, high -resolution data (30 m resolution) were used for both hazard and
exposure analysis , while the current analysis in the field are ranging from 100 m to 1 km. Finally,
a simple and flexible GIS tool (FloodRisk) was used in order to quantitatively estimate the
damages. Furthermore, the pan -European harmonized damage functions (JRC depth -damage
funct ions) were applied . This tool can be used for different scale applications , allowing any type
of input data ( concerning the level of detail and availability), being a simple alternative for the
complex damage estimation approaches and models.
The results showed that the flood damage can be highly influenced by the quality of the
input data. Using the hazard map developed in this study, that contains more areas affected by
floods, the value of damages is increasing considerably particularly in urban areas. Furthermore,
it was demonstrated that the coarse resolution of the existing land use map s at large scale (e.g.
Corine land cover) induce large uncertainties in damage calculation, particularly for industrial
areas. It was demonstrated that the Corine land cover present errors regarding the identification of
industrial land use class, underest imating this areas.
The main limitation of the proposed approach is related to the fact that the flood defense
measures are not considered in the analysis and the simplified procedures may decrease the
accuracy of the results. However, the proposed metho dology, applied and tested for the entire
territory of Romania, can be applied over large areas with limited availability of data and where
hydraulic analysis cannot be conducted, offering high -resolution results. Furthermore, the
homogeneity of data offer comparable and consistent results.

110
In this study, the methodology was applied in order to estimate the potential damage for a
return period of 100 years. However, the methodology can be used for flood risk assessment by
applying it for different return periods.

111
5. General c onclusions
Floods are phenomena that occur annually worldwide, being the main natural hazard with
a devastating impact on societies (I. Escuder -Bueno et al., 2012; Schober et al., 2013). In recent
years the studies (te Linde et al., 2010; BUBECK and Kreibich, 2011; BUBECK P. et al., 2011;
Escuder -Bueno et al., 2012, etc.) have shown that the frequency but also the economic damage
caused by floods is increasing in many regions (Albano et al., 2017). The main causes are
associated with climate and socio -economic change . The first one refers to changes in
environmental conditions like rainfall patterns and intensities while the second one refers to the
growth of population, assets and economic activity in flood prone areas and land use change .
In this context , the flood risk management has faced an increasing development in order to
deal with the devastating impacts of floods. The focus on more base -risk approaches and the need
of comprehensive flood risk assessment methodologies was also supported by the impleme ntation
of the Flood Directive. The current flood risk methodologies are integrating in the analysis both
hazard and consequence information for the devel opment of hazard and risk maps. In this way , a
better understanding of what can be expected is achieved and therefore preparing and protection
measures in order to reduce the risk can be applied .
This thesis focused at improving the understanding and the knowledge of flood risk
assessment process by provideng and applying a quantitative framework methodology for damage
estimation and analysing the related uncertainties. The scope of the thesis was stated as follows:
to quantitatively analyze the flood damage and its related uncertainties in data -scarce
environments, using innovative tools and methods that can offer cost -effective (instead of
expensive and time -consuming approach) and repeatable (transparent) solution for the flood area
delineation and damage calculation .
The thesis achieved this scope through a series of objectives:
 by presenting a theoretical background regarding the flood risk and dama ge assessment,
the current tools and methods that are used in the literature;
 by proposing free and open source GIS tools for hazard, exposure and damage estimation ;
 by applying the selected tools for the quantitative estimation of flood damages in data –
scarce environments ;
 by using free available and high -resolution data ;

112
 by presenting an uncertainty and sensitivity analysis regarding the flood damage
assessment .
In order to provide a thorough understanding of flood risk assessment process, the main
concepts related to floods, flood risk and flood damage were presented and synthesized. The
terminology regarding flood hazard and risk is described , as well as the methodology of flood
damage assessment. The different types of damages and the factors that are influencing them are
presented. There are a variety of tools and models that are used for hazard, exposure and damage
analysis, therefore a review was done with the aim of underlining the advantages an limitations of
different tools.
In the case study Flood damage assessment and uncertainties analysis. The case study of
2006 flood in Ilișua basin in Romania , a quantitative flood damage assessment was presented,
combining hazard and exposure information and using depth -damage functions. In order to
perform the damage simulations a GIS tool, called FloodRisk was applied. An uncertainty and
sensitivity analysis was further performed in order to assess the impact of damage functions on the
results. It is acknowledge that the transferability of damage functions in different regions can
induce great uncertainties in the analysis. Therefore the estimations of this kind of uncertainties
and their communication to the stak eholders is an important part of flood risk management, this
information being useful in the decision -making process.
In the second study, GIS tools for large -scale analysis of economic flood damage in data –
scarce environments. The case study of Romania , a national quantitative flood damage assessment
was conducted (for a return period of 100 years) . The main outcomes of this study are represented
by:
– A national hazard map , that contains the flood area and water depth for all the streams in
Romania, inclu ding the secondary ones that usually are not considered in large -scale analysis.
– A national land use map containing seven land use classes that were extracted from
Landsat 8 multispectral satellite images
– Damage results at national level that were obt ained using the above mentione d maps;
furthermore, the damages were also calculated using standard data (Corine Land Cover and the
pan-European JRC hazard map), and a comparison analysis was conducted.
Therefore, in this study a preliminary methodology for the estimation of flood damages in
data-scarce environments was applied. In order to obtain the hazard map, a GIS tool (GFA tool)

113
that is using the geomorphology of the basin was used. Using the DEM of the study areas, the
geomorphological features of this area can be extracted and used for the flood delineation and
water depth calculation. This method is easy to apply and can provide extended flood maps that
includes all the streams from the basins tha t are analysed, offering in this way a complete analysis
over large -areas. Furthermore for the analysis it was used a high -resolution DEM (30 m), the
current studies in the field using data with a resolution ranging from 100 m to 1 km. Therefore,
more accu rate and reliable results are obtained.
A land use map for the flood affected areas was developed in this study. For this purpose
the Landsat 8 multi -spectral satellite images with a resolution of 30 m were used along with
machine -learning classification algorithms in order to detect the land use classes for the study area.
For the training of the algorithm, the Urban Atlas database was used. This method can be used for
the identification of land use classes over large areas with limited availability of data.
The damages were quantitatively estimated using the pan -European JRC depth -damage
curves along with the FloodRisk tool. Four scenarios were considered regarding the input data for
the damage estimation: JRC water depth and CLC ; GFI water depth and Landsat 8 land use; JRC
water depth and Landsat 8 land use and GFI water depth and CLC . It was demonstrated that when
the GFI map is used, the damages increase, particularly in urban and industrial areas. This is
explained by the fact that larger areas are affected by floods. Furthermore the CLC database induce
great uncertainties in the damage results, particularly the industrial areas , this areas being
underestimated by the Corine database.
The proposed methodology can be used for cost -benefit analysis, offering quantitative
information regarding the flood damages at national level. It can be used in the decision -making
process, ident ifying the high risk areas and therefore prioritizing the risk reduction measures and
strategies. Furthermore the use of free publicly available tools and data allow the further scientific
development of this approach, improving the knowledge in the field.

114
6. Potential applications for stakeholders and future recommendation

The devastating impact of floods in the past years triggered the awareness and interest
regarding the need of effective flood risk reduction measures and strategies . In this context, the
stakeholders are facing a difficult task needing up to data knowledge, methods and techniques in
order to be able to respond to the growing requirements for an improved and unitary flood risk
management . Moreover, the lack of data in ma ny areas represent a real issue, the traditional method
to obtain this data being expensive and time -consuming.
The flood risk assessment at meso -scale is used for regional hazard and risk mapping,
helping the authorities t o identify high -risk areas, to prioritize investments and to implement
effective flood risk reduction measures. It also play an important role in the decision -making
process and it can be used for planning strategies. ROrisk connection? to be foreseen
More over, stakeholders such as the governments and the insurance industry need
comparable and consisted data over large areas for national and global policies and for the
prioritization of investments at national level . The governments is using the hazard info rmation in
order to implement flood damage reduction measures, for emergency operations and for land use
policies. For this purpose, efficient methods and tools are needed for hazard and risk mapping. On
the other hand , this information are useful for insu rance companies in order to establish the price
for flood risk insurance. The frequency and the damages produced by flood are expected to
increase as well as the number of flood exposed assets; therefore, the number of insurances will
increase, the need of accurate an undated data regarding the flood risk being of upmost importance
for insurance companies.
However, even though the research in the field of flood risk management is constantly
improving, there is a gap between the scientific knowl edge and the implementation of this
knowledge for common practices. Therefore, the research findings should be delivered and
materialized in methods and tools that can be applied in a useful way be the stakeholders. On the
other hand, the research goals sh ould be relevant to the stakeholders’ needs. These facts can be
accomplished through a collaborative approach that promote the common involvement of scientist
and stakeholders in the flood risk management process.
In this study , the proposed methodology was used for the estimation of flood damages for
a return period of 100 years. As a future research aim, this methodology can be applied for different

115
return periods in order to assess the flood risk at national scale. This could offer a better
understandi ng of the high -risk area distribution on the territory of Romania.
Another aim for further researches is the improvement of the data regarding the asset value,
this data representing an important source of uncertainties in flood risk assessment process . For
large scale, generalised data are used, however they may not be representative for the entire
analysed territory. Therefore, there is a need for a detailed database containing this information for
different categories and type of assets (e.g. land use classes) as well as for different locations. The
damage functions represent another component of the analysis that must be improved, this
approach being still new in the research field. Few areas have developed specific da mage functions
and their transferability in space can induce high uncertainties, therefore a thorough validation
must be done before using them in other areas.
The flood damage assessment should focus more on the indirect damage estimation, since
they rep resent an important percentage of the total damages. Since now the estimation of this
damages has not been approached in many studies, however it would offer a complete assessment
useful in the decision -making process.
One of the limitation of the proposed methodology is the fact that the protection measures
are not considered in the analysis. T his is a general situation of the current studies in the field , few
research approaches analysing in effects of this measures. Future research should focus on the
integration of structural and non -structural mitigation measures in the flood risk assessment
process in order to reduce the uncertainties.
Another field of research that can be analysed more closely is the one referring to Natech
events (natural disasters which trigger technological disasters). Romania is a country with an
extended area exposed to floods, containing as well a large number of technological sites (which
contain hazardous substances) (Ozunu et al., 2011; Ozunu et al., 2017). As the industry wi ll
increase and the climate will change the occurrence probability of Natech events caused by floods
will increase as well. The Natech disasters are complex, being difficult to analyses due to the
limited information that is available regarding this type o f events. Moreover, the intervention is
more difficult since there are two disaster types taking place in the same time and the resources
(that may be limited) must be used for the management of both situations; the flood event would
make harder the interv ention in case of a technological disaster; dangerous substance could by
spilled in water, and therefore the flood will have a greater impact on the assets. All this factors

116
increase the consequences of the Natech events, comparing with the occurrence of a single event
(natural or technological). Therefore, there is a need for a better analysis and management of this
type of disasters.

117
References
Albano, R., Sole, A., Adamowski, J., Mancusi, L., 2014. A GIS -based model to estimate
flood consequences and the degree of accessibility and operability of strategic emergency response
structures in urban areas. Natural Hazards Earth System Sciences, 14, 2847 -2865.
Albano, R., Mancusi, L., Sole, A., Adamowski, J., 2015a. Collaborative Strategies for
Sustainable EU Flood Risk Management: FOSS and Geospatial Tools —Challenges and
Opportunities for Operative Risk Analysis. ISPRS Int. J. Geo -Inf.4, 2704 -2727.
Albano, R., Sole, A., Adamowski, J., 2015b. READY: a web -based geographical
information syste m for enhanced flood resilience through raising awareness in citizens. Natural
Hazards Earth System Sciences, 15, 1645 -1658.
Albano, R., Crăciun, I. , Mancusi, L., Sole, A., Ozunu, A., 2017a. Flood damage
assessment and uncertainty analysis: the case study of 2006 flood in Ilișua basin in Romania.
Carpathian Journal of Earth and Environmental Sciences, July 2017, Vol. 12, No. 2, p. 335 – 346
Albano , R., Mancusi, L., Sole, A., Adamowski, J., 2017b. FloodRisk: a collaborative, free
and open -source software for flood risk analysis. Geomatics, Natural Hazards and Risk, 8:2, 1812 –
1832. DOI: 10.1080/19475705.2017.1388854.
Albano, R., Mancusi, L., Abbate, A., 2017c. Improving flood risk analysis for effectively
supporting the implementation of flood risk management plans: The case study of “Serio” Valley.
Environmental Science and Policy 75, 158 –172. http://dx.doi.org/10.1016/j.envsci.2017.05.017.
Alfieri, L., Salamon, P., Bianchi, A., Neal, J., Bates, P., Feyen, L., 2014. Advances in pan –
European flood hazard mapping. Hydrol. Process. 28, 4067 –4077.
https://doi.org/10.1002/hyp.9947
Alifieri, L., Feye n, L., Salamon, P., Thielen, J., Bianchi, A., Dottori, F., 2016. Modelling
socio -economic impact of floods in Europe. Nat. Hazards E arth Syst. Sci., 16, 1401 –1411.
www.nat -hazards -earth -syst-sci.net/16/1401/2016/ doi:10.5194/nhess -16-1401 -2016
Arghiuș, V., 2008. Studiul viiturilor de pe cursurile de apa din estul muntilor Apuseni si
riscurile associate, Casa Cartii de Stiinta, Cluj -Napoca.
Arghiuș, V., Botezan, C., Gagiu, A. C., Samara, I., Senzaconi, F., Ozunu, A., 2011.
Normalized economical flood damages in Romania during 2000 -2009. Environmental
Engineering and Management Journal, 10, 17 -21.

118
Arghiuș, V., Ozunu, A., Samara, I., Roșian, G., 2014. Results of the post flash -flood
disaster investigations in the Transylvanian Depression (Romania) during the la st decade (2001 –
2010). Natural Hazards and Earth System Sciences, 14, 535 -544.
Banks, J., C., Camp, J., V., Abkowitz, M., D., 2014. Adaptation planning for floods: a
review of available tools. Nat Hazards 70:1327 –1337. DOI 10.1007/s11069 -013-0876 -7.
Barred o, J.I. and De Roo, A., 2007. Flood risk mapping at European scale, Water Science
& Technology 56(4):11 -7, DOI:10.2166/wst.2007.531
Barredo, J. I., 2006. Major flood disasters in Europe: 1950 –2005. Natural Hazards Earth
System Sciences, 42, 125 -148.
Biro, K., Pradhan, B., Sulieman, H., Buchroithner, M., 2013. Exploitation of TerraSAR -X
Data for Land use/Land Cover Analysis Using Object -Oriented Classification Approach in the
African Sahel Area, Sudan. J Indian Soc Remote Sens., 41(3):539 –553. DOI 10.1007/s1 2524 -012-
0230 -7.
Blöschl, G., Gaál, L., Hall, J., Kiss, A., Komma, J., Nester, T., Parajka, J., Perdigão, R. A.
P., Plavcová, L., Rogger, M., Salinas, J., L., and Viglione, A., 2015. Increasing river floods: fiction
or reality?, WIREs Water 2015, 2:329 –344. doi: 10.1002/wat2.1079
Boteza, C., Ozunu, A., Ștefănie, H., 2015. Vulnerability assessment: the case of the Arieș
river middle basin. Journal of Environmental Protection and Ecology 16, No 4, 1326 –1336.
Bubeck, P., Kreibich, H., 2011. Natural Hazards: di rect costs and losses due to the
disruption of production processes. Global Assessment Report on Disaster Risk Reduction. 20p.
Bubeck, P., de Moel, H., Bouwer, L.M., Aerts, J.C.J.H, 2011. How reliable are projections
of future flood damage?.Natural Hazard s Earth System Sciences, 11, 3293 –3306.
Büttner, G., 2014. CORINE land cover and land cover change products. Chapter: Land Use
and Land Cover Mapping in Europe, Volume 18 of the series Remote Sensing and Digital Image
Processing pp. 55 –74.
Cammerer, H., Th ieken, A. H., Lamme, J., 2013. Adaptability and transferability of flood
loss functions in residential areas. Natural Hazards and Earth System Sciences, 13, 3063 –3081.
Cantisani, A., Giosa, L., Mancusi, L., Sole, A., 2014. FLORA -2D: a new model to simulate
the inundation in areas covered by flexible and rigid vegetation. Int J Eng Innov Technol 3(8):179 –
186

119
Chingombe, W., Pedzisai, E., Manatsa, D., Mukwada, G., Taru, P., 2015. A participatory
approach in GIS data collection for flood risk management, Muzarabani district, Zimbabwe,
Arabian Journal of Geosciences, Volume 8, Issue 2, pp 1029 –1040.
City Hall Uriu , 2012.Assessment Report for buildings special constructions and lands.
City Hall Uriu, 2012. Assessment Report for streets, roads, bridges and footbridges.
Cirella, G.T., Semenzin, E., Critto, A., Marcomini, A., 2014. Natural Hazard Risk
Assessment and M anagement Methodologies. Review: Europe. In book: Sustainable Cities and
Military Installations, DOI10.1007/978 -94-007-7161 -1_16
County Committee for Emergency Situations Bistrița -Năsăud, 2006. Summary Report on
defense against floods, dangerous meteoro logical phenomena and hydraulic structures accidents
in the County of Bistrita -Nasaud of 20 June 2006.
Copernicus, Land Monitoring Service, Corine Land Cover, 2018. Available at:
https://land.copernicus.eu/pan -european/corine -land-cover , [Accessed on 2 February 2018].
Copernicus, Land Monitoring Service, Urban Atlas , 2018a. Available at:
https://land.copernicus.eu/local/u rban-atlas, [Accessed on 25 February 2018].
CRED EM -DAT, Centre for Research on the Epidemiology of Disas ters, 2018. The
International Disaster Database, Available at: www.emdat.be , [Accessed 06 February 2018] .
Danube -Floodrisk, European Union, 2018. Danube Floodrisk project, Available at:
http://www.danube -floodrisk.eu/ , [Accessed on 11 January 2018].
De Roo, A., Barredo, J.I., Lavalle, C., Bodis, K., Bonk, R., 20 06. Potential flood hazard
and risk mapping at pan -European scale. In: Digital elevation modelling – Development and
applications in a policy support environment, Peckham, R. and Jordan, G. (eds.), Joint Research
Centre, European Commission, Ispra (Italy).
Ding, A., White, J.F., Ullman, P.W. Fashokun, A.O., 2008. Evaluation of HAZUS -MH
Flood Model with Local Data and Other Program, Natural Hazards Review 9(1),
DOI:10.1061/(ASCE)1527 -6988(2008)9:1(20)
Directive 2000/60/EC of the European Parliament and of th e Council of 23 October 2000
establishing a framework for Community action in the field of water policy, Official Journal L 327
, 22/12/2000 P. 0001 – 0073
Directive 2007/60/EC of the European Parliament and the Council of 23 October 2007 on
the assessment and management of flood risks. Off. J. Eur. Union 2007, L288/27 -L288/34.

120
Dornaika, F., Moujahid, A., El Merabet, Y., Ruichek, Y., 2016. Building detection from
orthophotos using a machine learning approach: An empirical study on image segmentation and
descriptors. Expert Systems With Applications 58, 130 –142.
Eleuterio, J., 2012. Flood risk analysis: impact of uncertainty in hazard modelling and
vulnerability assessments on damage estimations, PhD thesis, Universite de Strasbourg, France
Engelund, F., Han sen, E., 1967. A monograph on sediment transport in alluvial streams.
Monografia 65. https://doi.org/10.1007/s13398 -014-0173 -7.2
Egorova, R., van Noortwijk, J.M., Holterman, S.R., 2008, Uncertainty in flood damage
estimation, International Journal of River Basin Management 6(2):139 -148,
DOI:10.1080/15715124.2008.9635343
Escuder -Bueno, I., Castillo -Rodríguez, J.T., Zechner, S., Jöbstl, C., Perales -Momparler, S.,
Petaccia, G., 2012. A quantitative flood risk analysis methodology for urban areas with integration
of social research data. Natural Hazards and Earth System Sciences, 12, 2843 –2863.
Felzenszwalb P. and Huttenlocher, D., 2004. Efficient graph -based image seg mentation.
Int. J. Comput. Vis., 59, 167 –181.
Feranec, J., Hazeu, G., Christensen, S., Jaffrain, G., 2007. Corine land cover change
detection in Europe (case studies of the Netherlands and Slovakia). Land Use Policy 24 (2007)
234–247. doi:10.1016/j.landusepol.2006.02.002.
Feyen, L., Dankers, R., Bódis, K., Salamon, P., Barredo, J.I., 2012. Fluvial flood risk in
Europe in present and future climates, Climatic Change 112:47 –62 DOI 10.1007/s10584 -011-
0339 -7
Freni, G., La Loggia, G., Notaro, V., 2010. Uncertainty in urban flood damage assessment
due to urban drainage modelling and depth -damage c urve estimation. Water Science and
Technology, 61 (12) 2979 -2993; DOI: 10.2166/wst.2010.177.
FRMP – Flood Risk management Plan, “Romanian Waters” National Administration,
2015.
Frohn, R., Reif, M., Lane, C., Autrey, B., 2009. Satellite remote sensing of isolated
wetlands using object­oriented classification of Landsat­7 data. Wetlands, Vol. 29, pp. 931 –941.
GDP -Deflator , TredEconomy, 2018. Available at:
http://data.trendeconomy.com/datavie wer/wb/wbd/wdi , [Accessed on 2 April 2018].

121
Gogoașe Nistoran, D., E., Gheorghe Popovici, D., A., Craia Savin, B., A., Armaș, I., 2016.
GIS for Dam -Break Flooding. Study Area: Bicaz -Izvorul Muntelui (Romania). Space and Time
Visualisation pp 253 -280. https ://doi.org/10.1007/978 -3-319-24942 -1_15
Gouldby, B. P., Klijn, F., Samuels, P.G., Sayers, P.B., Schanze, J., 2005. Language of
Risk—Discussion Document, FLOODsite report T32 -04-01, available from www.floodsite.net
G.O. 846/2010 , Government Order no. 846/2010 for the approval of the National Strategy
of flood risk management – published in the “Monitorul Oficial” (Official Gazette of Romania),
Part I, No. 626/September 6, 2010 .
HEC -RAS, US Army Corps of Engineers – Hydrologic Engineering Center , 2017.
Hydrologic Engineering Center's (CEIWR -HEC) River Analysis System , Available at:
http://www.hec.usace.army.mil/software/hec -ras/ , [Accessed on: 23 Septembe r 207].
Huizinga, J., 2007. Flood damage functions for EU member states. European Commission
– Joint Research Centre.
Huizinga, J., de Moel, H., Szewczyk, W., 2017. Global flood depth -damage functions.
Methodology and the database with guidelines. EUR 28552 EN. doi: 10.2760/16510.
Ibbitt, R.P., 1997. Evaluation of optimal channel network and river basin heterogeneity
conc epts using measured flow and channel properties. J. Hydrol. 196, 119 –138.
https://doi.org/10.1016/S0022 -1694(96)03293 -3.
Ibbitt, R.P., McKerchar, A.I., Duncan, M.J., 1998. Taieri River data to test channel network
and river basin heterogeneity concepts. Wa ter Resour. Res. 34, 2085 –2088.
https://doi.org/10.1029/98WR00483 .
Istomina, M., N., Kocharyan, A., G., Lebedeva, I. P., 2005. Floods: Genesis,
Socioeconomic and Environmental Impacts, Water Resources, Vol. 32, No. 4, 2005, pp. 349 –358
Johnson, F., White, C. J., van Dijk, A., Ekstrom, M., Evans, J. P., Jakob, D., Kiem, A. S.,
Leonard, M., Rouillard, A., Westra, S, 2016. Natural hazards in Australia : floods. Climatic
Change, 139 (1). pp. 21 -35. ISSN 0165 -0009 , http://dx.doi.org/10.1007/s10584 -016- 1689 -y.
Jongman, B., Kreibich, H., Apel, H., Barredo, J.I., Bates, P.D., Feyen, L., Gericke, A.,
Neal, J., Aerts, J.C.J.H., Ward, P.J., 2012. Comparative flood damage model assessment: Towards
a European approach. Natural Hazards and Earth System Sciences, 12, 373 3–3752.

122
Kabenge, M., Elaru, J., Wang, H., Li, F., 2017. Characterizing flood hazard risk in data –
scarce areas, using a remote sensing and GIS -based flood hazard index, Nat Hazards (2017)
89:1369 –1387, DOI 10.1007/s11069 -017-3024 -y.
Kaźmierczak, A., Cavan, G., 2011. Surface water flooding risk to urban communities:
Analysis of vulnerability, hazard and exposure. Landscape and Urban Planning 103, 185 -197.
Kellermann, P., Schöbel, A., Kundela, G., Thieken, A.H., 2015. Estimating flood damage
to railway infrast ructure – the case study of the March River flood in 2006 at the Austrian Northern
Railway, Nat. Hazards Earth Syst. Sci., 15, 2485 -2496, 2015
https://doi.org/10.5194/nhess -15-2485 -2015
Khattak, M., S., Anwar, F., Saeed, T., U., Sharif, M., Sheraz, K., Ahmed, A., 2016.
Floodplain Mapping Using HEC -RAS and ArcGIS: A Case Study of Kabul River. Arab J Sci Eng
(2016) 41:1375 –1390. DOI 10.1007/s13369 -015-1915 -3.
Klijn, F., 2009. Flood Risk Assessment and Flood Risk Management. An Introduction and
Guidance Bas ed on Experiences and Findings of FLOODsite; FLOODsite Project: Delft, the
Netherland.
Knighton, D., 2014. Fluvial forms and processes: a new perspective.
Kok, M., Huizinga, J., Vrouwenvelder, A., C., W., M., Barendregt, A., 2004. Standard
Method 2004. Dam age and Casualties caused by Flooding.
Kovacs, A., Ștefănie, H., Botezan, C., Crăciun, I. , Ozunu, A., 2017. Assessment of natural
hazards in european countries with impact on young people, 17th International Multidisciplinary
Scientific GeoConference (SGE M), Conference Proceedings, vol. 17(52), pp. 73 -81.
Kreibich, H., Seifert, I., Merz, B., Thieken, A.H., 2010. Development of FLEMOcs – a
new model for the estimation of flood losses in the commercial sector, Hydrological Sciences
Jurnal, Volume 55, 2010 – Issue 8, Pages 1302 -1314,
https://doi.org/10.1080/02626667.2010.529815
Kumar Sharma, V., Amminedu, E., Srinivasa Rao, G., Nagamani, P. V., Ram Mohan Rao,
K., Bhanumurthy, V., 2016. Assessing the potential of open -source libraries for managing satellite
data products – A case study on disaster management. Annals of GIS, 23:1, 55 -65, DOI:
10.1080/19475683.2016.1231718.
Kundzewicz, Z.W., Pińskwar, I., Brakenridge, G.R., 2013. Large floods in Europe, 1985 –
2009, Hydrological Sciences Journal, 58:1, 1 -7, DOI: 10 .1080/02626667.2012.745082

123
Kuntz, S., Schmeer, E., Jochum, M., Smith, G., 2014. Towards an European Land Cover
Service and High -Resolution Layers. In book: Land Use and Land Cover Mapping in Europe –
Practices & TrendsChapter: 4, Publisher: Springer.
Lastr a, J., Fernández, E., Díez -Herrero, A., Marquínez, J., 2008. Flood hazard delineation
combining geomorphological and hydrological methods: an example in the Northern Iberian
Peninsula. International Society for the Prevention and Mitigation of Natural Haza rds, vol. 45(2),
pages 277 -293. DOI: 10.1007/s11069 -007-9164 -8.
Leopold, L.B., Maddock, T.J., 1953. The Hydraulic Geometry of Stream Channels and
Some Physiographic Implications, Geological Survey Professional Paper 252.
https://doi.org/10.1016/S0169 -555X( 96)00028 -1
Leopold, L.B., Wolman, M.G., Miller, J.P., 1965. Fluvial processes in geomorphology. J.
Hydrol. https://doi.org/10.1016/0022 -1694(65)90101 -0
Li, R., 1974. Mathematical modeling of response from small watershed.
Liu, J., Wang, S., Li, D., 2014. h e Analysis of the Impact of Land -Use Changes on Flood
Exposure of Wuhan in Yangtze River Basin, China, Water Resour Manage (2014) 28:2507 -2522,
DOI 10.1007/s11269 -014-0623 -1
te Linde, A. H., Bubeck, P., Dekkers, J. E. C., De Moel, H., Aerts, J. C. J. H., 2011. Future
flood risk estimates along the river Rhine. Natural Hazards and Earth System Sciences, 11, 459 –
473, doi:10.5194/nhess -11-459-2011.
Lugeri, N., Kundzewicz, Z.W., Genovese, E., Hochrainer, S., Radziejewski, M., 2010.
River flood risk and adaptat ion in Europe —assessment of the present status, Mitig Adapt Strateg
Glob Change (2010) 15:621 –639 DOI 10.1007/s11027 -009-9211 -8
Mancusi, L., Albano, R., Sole, A., 2015. FloodRisk: a QGIS plugin for flood consequences
estimation, Geomatics Workbooks n°12 – FOSS4G Europe Como.
Manfreda, S., Samela, C., Gioia, A., Consoli, G., G., Iacobellis, V., Giuzio, L., Cantisani,
A., Sole, A., 2015. Flood -prone areas assessment using linear binaryclassifiers based on flood
maps obtained from 1D and 2D hydraulic models. N atural Hazards. Nat Hazards (2015) 79: 735.
https://doi.org/10.1007/s11069 -015-1869 -5
Manfreda, S., Samela, C., Troy, T., J., 2018. The Use of DEM -Based Approaches to Derive
a Priori Information on Flood -Prone Areas. Flood Monitoring through Remote Sensing, pp 61 -79.
https://doi.org/10.1007/978 -3-319-63959 -8_3.

124
Marzocchi, R., Federici, B., Cannata, M., Cosso, T., Syriou, A., 2014. Comparison of one –
dimensional and two -dimensional GRASS -GIS models for flood mapping. Applied Geomatics,
Volume 6, Issue 4, pp 245 –254. DOI: https://doi.org/10.1007/s12518 -014-0140 -1
McKerchar, A.I., Ibbitt, R.P., Brown, S.L.R., Duncan, M.J., 1998. Data for Ashley River
to test channel network and river basin heterogeneity concepts. Water Resour. Res. 34, 139 –142.
https://do i.org/10.1029/97WR02573
Merz, B., Kreibich, H., Thieken, A., Schmidtke R., 2004. Estimation uncertainty of direct
monetary flood damage to buildings. Natural Hazards Earth System Sciences, 4, (1), 153 –163.
Merz, B., Kreibich, H., Schwarze, R., Thieken, A., 2010. Assessment of economic flood
damage. Natural Hazards and Earth System Sciences, 10, 1697 –1724.
Messner, F., and Meyer, V., 2005. Flood damage, vulnerability and risk perception –
challenges for flood damage research. UFZ discussion papers, published in Schanze, J., Zeman,
E., Marshalek, J., Flood risk management – Hazards vulnerability and Mitigation Measures.
Messner, F., Pennning -Rowsell, E.C., Green, C., Meyer, V., Tunstall, S.M., Van der Veen,
A., 2007. Evaluating flood damages: guidance and reco mmendations on principles and methods,
FLOODsite, Report No. T09 -06-01.
Meyer, V., Messner, F., 2005. National Flood Damage Evaluation Methods. A Review of
Applied Methods in England, the Netherlands, the Czech Republic and Germany.
Meyer, V., Becker, N ., Markantonis, V., Schwarze, R., van den Bergh, J.C.J.M., Bouwer,
L.M., Bubeck, P., Ciavola, P., Genovese, E., Green, C., Hallegatte, S., Kreibich, H., Lequeux, Q.,
Logar, I., Papyrakis, E., Pfurtscheller, C., Poussin, J., Przyluski, V., Thieken, A.H., Vi avattene, C,
2013. Assessing the costs of natural hazards – state of the art and knowledge gaps. Hazards and
Earth System Sciences,13, 1351 –1373.
MIKE FLOOD, 1D -2D Modelling, User Manual, 2017. Available at :
http://manuals.mikepoweredbydhi.help/2017/Water_Resources/MIKE_FLOOD_UserMa
nual.pdf , [Accesed on 25 March 2018 ].
Mikhailova, M. V., Mikhailov, V. N., Morozovc, V. N., 2012. Extreme Hydrological
Events in the Danube River Basin over the Last Decades, Water Resources, 2012, Vol. 39, No. 2,
pp. 161 –179, DOI: 10.1134/S0097807812010095
de Moel, H., Aerts, J.C.J.H, 2011. Effect of uncertainty in land use, damage models and
inundation depth on flood damage estim ates.Hazards and Earth System Sciences, 58, 407 –425.

125
de Moel, H., 2012. Uncertainty in Flood Risk, PhD Thesis, VU University of Amsterdam,
ISBN: 978 -94-6203 -055-8
de Moel, H., van Vliet, M., Aerts, J.C.J.H, 2013. Evaluating the effect of flood damage –
reduc ing measures: a case study of the unembanked area of Rotterdam, the Netherlands. Reg
Environ Change, 14,(3):895 -908.
de Moel, H., Jongman, B., Kreibich, H., Merz, B., Penning -Rowsell, E., Ward, P.J., 2015.
Flood risk assessment at different spatial scales. Mitig Adapt Strateg Glob Change, 20, 865 -890.
Molinari, D., Menoni, S., Aronica, G. T., Ballio, F., Berni, N., Pandolfo, C., Stelluti, M.,
Minucci, G., 2014. Ex post damage assessment: an Italian experience, Nat. Hazards Earth Syst.
Sci., 14, 901 –916, doi: 10.5194/nhess -14-901-2014, 2014a.
Montero, E., Van Wolvelaer, J., Garzón, A., 2014. The European Urban Atlas. Land Use
and Land Cover Mapping in Europe: Practices & Trends, Remote Sensing and Digital Image
Processing 18. DOI 10.1007/978 -94-007-7969 -3_8.
Mosquera -Machado, S., Ahmad, S., 2007. Flood hazard assessment of Atrato River in
Colombia. Water Resour Manage 21:591 –609. DOI 10.1007/s11269 -006-9032 -4.
Muller, U., 2013. Implementation of the Flood Risk Management Directive in Selected
European Countries, Int. J. Disaster Risk Sci. 2013, 4 (3): 115 –125 doi:10.1007/s13753 -013-0013 –
y
NASA – Landast 8 Overview, 2018. Available at: https://landsat.gsfc.nasa.gov/landsat –
8/landsat -8-overview/ , [Accessed on 24 May 2018]
N.I.H.W.M . – National Institute for Hydrology and Water Management, 2016. The
National Plan for prevention, protection and mitigation of flood effects , (in Romanian)
Olfert, A., Schanze, J., 2007. Methodology for ex -post evaluation of measures and
instruments in flood risk management (postEval), Leibniz Institute for Ecological and Regional
Development (IOER), FLOODsite Report T12 -07-01, Dresden.
Ok, A.O., 2013. Automated detection of buildings from single VHR multispectr al images
using shadow information and graph cuts. ISPRS Journal of Photogrammetry and Remote Sensing
86, 21 –40. http://dx.doi.org/10.1016/j.isprsjprs.2013.09.004 .
Ozunu, A., Anghel, C. I., 2007. Evaluarea riscului tehnologic și securitatea mediului,
Publishing house Accent, Cluj -Napoca.

126
Ozunu, A., Senzaconi, F., Botezan, C., Ștefănescu, L., Nour, E., Balcu, C., 2011.
Investigations on natural hazards which trigger technological disasters in Romania, Nat. Hazards
Earth Syst. Sci., 11, 1319 –1325. doi:10.5194/nhess -11-1319 -2011.
Ozunu, A., Mereuță, A., Török, Z., Literat, L., 2017. A national hazard analysis and
mapping for Seveso establishments. Journal of En gineering Sciences and Innovation, Volume 2,
Issue 3 / 2017, pp. 93 -102.
Park, C.C., 1977. World -wide variations in hydraulic geometry exponents of stream
channels: An analysis and some observations. J. Hydrol. 33, 133 –146.
https://doi.org/10.1016/0022 -1694(77)90103 -2
Patro, S., Chatterjee, C., Mohanty, S., Singh, R., Raghuwanshi, N., S., 2009. Flood
Inundation Modeling using MIKE FLOOD and Remote Sensing Data. J. Indian Soc. Remote Sens.
37:107 –118.
Pazúr, R., Feranec, J., Štych, P., Kopecká, M., Holman, L., 2017. Changes of urbanised
landscape identified and assessed by the Urban Atlas data: case study of Prague and Bratislava.
Land Use Policy, 61, 135 -146. http://doi.org/10.1016/j.landusepol.20 16.11.022
Penning -Rowsell, E.,Viavattene, C., Pardoe, J., Chatterton, J., Parker, D., Morris, J., 2010.
The Benefits of Flood and Coastal Risk Management: A Handbook of Assessment Techniques.
Flood HazardResearch Centre: London, UK.
Prastacos, P., Chrysoul akis, N., Kochilakis, G., 2011. Urban Atlas, land use modelling and
spatial metric techniques. 51st European Congress of the Regional Science Association
International. European Regional Science Accossiation. Barcelona, Spain, August 30 – September
3.
Priest, S.J., Suykens, C., Van Rijswick, H.F.M.W., Schellenberger, T., Goytia, S.,
Kundzewicz, Z.W., Van Doorn -Hoekveld, W.J., Beyers. J.C., Homewood, S., 2016. The European
union approach to flood risk management and improving societal resilience: lessons from the
implementation of the floods directive in six European countries. Ecol Soc 21(4):50 .
QGIS – A Free and Open Source Geographic Information System , 2017. Available at:
https://qgis.org/en/site/ , [Accessed on 15 January 2017]
RiverFlow 2D, 2D model, 2017. Available at: http://www.hydronia.com/riverflow2d/ ,
[Accessed on 13 June 2017].

127
Rodriguez -Iturbe, I., Rinaldo, A., 1997. Fractal river basins: chance an d self -organization,
Power. https://doi.org/10.1063/1.882305
Rumelhart, D. E., Hint on, G. E., and Williams, R. J., 1986. Learning representations by
back -propagating errors. Nature, 323, 533 –536.
Samela, C. , Albano, R., Sole, A., Manfreda, S., 2018. A GIS tool for cost -effective
delineation of flood -prone areas. Comput. Environ. Urban Syst. (accepted).
Samela, C., Troy, T.J., Manfreda, S., 2017. Geomorphic classifiers for flood -prone areas
delineation for da ta-scarce environments. Adv. Water Resour. 102, 13 –28.
https://doi.org/10.1016/j.advwatres.2017.01.007
Sampson, C., Smith, A., Bates, P., Neal, J., Alifiri, L., Freen, J., 2015. A high -resolut ion global
flood hazard model. Water Resour. Res., 51,7358 –7381, doi:10.1002/2015WR016954.
Samuels, P., Gouldby, B., 2009. Language of Risk -Project Definitions (Second edition);
FLOODsite Project:Delft, the Netherlands
Schober, B., Hauer, C., Habersack, H., 2013. A novel assessment of the role of Danube
floodplains in flood hazard reduction (FEM method), Nat Hazards, 75:S33 –S50.
Scorzini, A.R., Frank, E., 2015. Flood damage curves: new insights from the 2010 flood in
Veneto, Italy. FloodRis kManagement. https://doi.org/10.1111/jfr3.12163
Seifert, I., Kreibich, H., Merz, B., Thieken, A.H., 2010. Application and validation of
FLEMOcs – a flood -loss estimation model for the commercial sector, Hydrological Sciences
Jurnal, Volume 55, 2010 – Issue 8, Pages 1315 -1324,
https://doi.org/10.1080/02626667.2010.536440
Smith, T.R., 1974. A Derivation of the Hydraulic Geometry of Steady -State Channels from
Conservation Principles and Sediment Transport Laws. J. Geol. 82, 98 –104.
https://doi.org/10.1086/6279 39
Sole, A., Giosa, L., Albano, R., Cantisani, A., 2013. The laser scan data as a key element
in the hydraulic flood modelling in urban areas. Int. Arch. Photogramm. Remote Sens. Spatial Inf.
Sci., XL -4/W1, 65 -70, 2013.
Sofronie, C., Stoica, F.Ș., Selagea , H.I., Dulău, R.B., Sârb, M.T., Cocuț, M., Scuturici,
D.G., 2013. Bazinul hidrografic Someș -Tisa. Ed. U.T.PRESS.
Sterna, L., 2012. Pluvial flood damage modelling. Assessment of the flood damage model
HOWAD -PREVENT, MSc Thesis, TU Delft, the Netherlands.

128
Stancanelli, L.M., Foti, E., 2015. A comparative assessment of two different debris flow
propagation approaches – blind simulations on a real debris flow event. Nat. Hazards Earth Syst.
Sci., 15, 735 –746. doi:10.5194/nhess -15-735-2015
Steiniger, S., Bocher, E., 2009. An overview of current free and open source desktop GIS
developments. International Journal of Geographical Information Science, 23:10, 1345 -1370,
DOI: 10.1080/13658810802634956
Ștefănescu, L., Botezan, C., Crăciun, I ., 2018. Vulnerability analy sis for two accident
scenarios at an upper -tier Seveso establishment in Romania. Geographia Technica, Vol. 13, Issue
1, pp 109 to 118.
The Law No. 14/1995 – regarding the collaboration for the protection and sustainable usage
of the Danube River – publish ed in the “Monitorul Oficial” (Official Gazette of Romania), Part I,
No. 41/February 27, 1995.
The Law No. 107/September 25, 1996 – Water Law – published in the “Monitorul Oficial”
(Official Gazette of Romania), Part I, No. 244/October 8, 1996.
The Law No. 310/2004, amending The Law No. 107/September 25, 1996 – published in
the “Monitorul Oficial” (Official Gazette of Romania), Part I, No. 584/June 30, 2004
The Law No. 112/2006, amending The Law No. 107/September 25, 1996 – published in
the “Monitorul Ofi cial” (Official Gazette of Romania), Part I, No. 413/May 12, 2006
The Law No. 146/2010, amending The Law No. 107/September 25, 1996 – published in
the “Monitorul Oficial” (Official Gazette of Romania) , Part I, No. 497/July 19, 2010
Thieken, A. H., Muller, M., Kreibich, H., Merz, B., 2005. Flood damage and influencing
factors: New insights from the August 2002 flood in Germany, Water Resour. Res., 41, W12430,
doi:10.1029/2005WR004177
Thieken A.H., Ackermann, V., Elmer, F., Kreibich, H., Kuhlmann, B., Kunert , U.,
Maiwald, H., Merz, B., Müller, M., Piroth, K., Schwarz, J., Schwarze,R., Seifert, I., Seifert, I.,
2008. Methods for the evaluation of direct and indirect flood losses, Institute for Catastrophic Loss
Reduction. 4th International Symposium on Flood Defence: Managing Flood Risk, Reliability and
Vulnerability Toronto, Ontario, Canada, May 6 -8, 2008
TUFLOW, Aquaveo, 1D/2D Hydrodynamic Engine for Flood Modeling , Available at:
https://www.aquaveo.com/software/sms -tuflow , [Acessed on 18 June 2017]

129
Ullah, S., Farooq, M., Sarwar, T., Tareen, M., J., Wahid, M., A., 2016. Flood modeling and
simulations using hydrodynamic model and ASTER DEM —A case study of Kalpani River. Arab
J Geosc i 9: 439. DOI 10.1007/s12517 -016-2457 -z.
USGS, Landsat Missions , 2018. Landsat 8 Data Users Handbook , Available at:
https://landsat.usgs.gov/landsat -8-l8-data-users -handbook , [Accesed on 21 April 2018]
USGS, Earth Explorer tool, 2018. Available at: https://earthexplorer.usgs.gov/ , [Accesed
on 21 April 2018]
Vanneuville, W., Gamanya, R., De Rouck, K., Maeghe, K., De Maeyer, P., Mostaert, F.,
2005. Development of a Flood Risk Model and applications in the management of hydrographical
catchments. Proceedings of the Cartographic Cutting -Edge Technology for Natural Hazard
Management. p.169 -180.
Wieland, M., Pittore, M., 2014. Perfor mance Evaluation of Machine Learning Algorithms
for Urban Pattern Recognition from Multi -spectral Satellite Images. Remote Sensing, 6, 2912 –
2939; doi:10.3390/rs6042912.
Wieland, M., Pittore, M., 2016. Large -area settlement pattern recognition from Landsat -8
data. ISPRS Journal of Photogrammetry and Remote Sensing, Volume 119, Pages 294 -308.
https://doi.org/10.1016/j.isprsjprs.2016.06.010
Winsemius, H.C., Van Beek, L.P.H., Jongman, B., Ward, P.J., Bouwman, A., 2013. A
framework for global flood risk assessme nts, Hydrol. Earth Syst. Sci., 17, 1871 –1892,
doi:10.5194/hess -17-1871 -2013
Whiting, P.J., Stamm, J.F., Moog, D.B., Orndorff, R.L., 1999. Sediment -transporting flows
in headwater streams. Bull. Geol. Soc. Am. 111, 450 –466. https://doi.org/10.1130/0016 –
7606(1999)111<0450:STFIHS>2.3.CO;2
Yılmaz, R., 2010. Monitoring land use/land cover changes using CORINE land cover data:
a case study of Silivri coastal zone in Metropolitan Istanbul. Environ Monit Assess (2010)
165:603 –615. DOI 10.1007/s10661 -009-0972 -z.
Zaharia, L., Toroimac, G. I., 2018. Overview of River -Induced Hazards in Romania:
Impacts and Management. Water Management and the Environment: Case Studies, Water Science
and Technology Library 86. https://doi.org/10.1007/978 -3-319-79014 -5_9.

130
Zhang, J., Li, P., Wang, J., 2014. Urban Built -Up Area Extraction from Landsat TM/ETM+
Images Using Spectral Information and Multivariate Texture. Remote Sensing, 6, 733 9-7359;
doi:10.3390/rs6087339.

131
Annex 1. Hazard analysis. Flow direction and flow accumulation maps

Fig. A1 . Flow direction

132
Annex 2. The pan -European flood hazard map (JRC map) with a return period of 100 years

Fig. A3 . The pan -European flood hazard map (JRC map) (adapted after Alfieri et al., 2014)

133
Annex 3. Landsat 8 image processing results

Fig. A4 . Lansat8 Band 1 – Roma nia

Fig. A5 . Lansat8 Band 2 – Roma nia

134

Fig. A6 . Lansat8 Band 3 – Roma nia

Fig. A7 . Lansat8 Band 4 – Roma nia

135

Fig. A8 . Lansat8 Band 5 – Roma nia

Fig. A9 . Lansat8 Band 6 – Roma nia

136

Fig. A10 . Lansat8 Band 7 – Roma nia

Fig. A11 . Lansat8 Band 8 – Roma nia

137

Fig. A12 . Lansat8 Band 9 – Roma nia

Fig. A13 . Lansat8 Band 10 – Roma nia

138

Fig. A14 . Lansat8 Band 11 – Roma nia