VU University of Amsterdam
The economic impact of flooding
Flood risk assessment on the Westpoort harbour – Amsterdam
Lars de Ruig
Bachelor thesis Supervisor: E.E. Koks MSc.
Future Planet Studies 2
ndAssessor: Dr. P.J. Ward
Major Earth Science and Economics July 2014
2
Abstract
In this thesis, a flood risk assessment is made for the Westpoort harbour of Amsterdam. A breach of the Lekdijk or a breach of the floodgates at IJmuiden are two scenarios that could cause high economic damage for the Westpoort harbour. By using an integrative flood risk model, the direct and indirect economic damage is assessed and the effects of a shutdown of the electricity network in the study area is identified. For the Lekdijk scenario, total economic damage is assessed at 2.0 billion to 2.7 billion euros, and for the IJmuiden scenario at 8.4 billion to 10.3 billion euros. Furthermore, effects of damage to the electricity network could range up to 2 billion euros of losses. However, the model is sensitive to assumptions and parameter settings, and therefore caution should be taken when interpreting these results. Nonetheless, the results do give an indication of the magnitude of total economic damage in the Westpoort harbour as a result of a flood.
3
Acknowledgement
Firstly, I would like to express my special thanks to my supervisor and first assessor, E.E. Koks, for the pleasant collaboration during the whole project and the fast responses to any of my questions. I would also like to thank my second assessor, P.J. Ward, for his feedback in each phase of this thesis. Furthermore, I am grateful for the support and inspiring ideas of J.C.J.H. Aerts and H. de Moel.
Additionally, I would like to thank C. van Drimmelen of ‘Gemeente Amsterdam’, R. Koeze of ‘Waternet’ and R. de van der Schueren of the ‘Port of Amsterdam’ for their time and ideas regarding this thesis.
4
Table of Contents
1. INTRODUCTION ... 5
2. DIRECT AND INDIRECT ECONOMIC DAMAGE ... 8
3. METHODOLOGY ... 10
3.1 DISASTER SCENARIOS ... 10
3.1.1 Inundation modelling ... 10
3.1.2 Flood duration ... 12
3.1.3 Electricity network influence ... 12
3.2 INTEGRATIVE FLOOD RISK MODEL ... 14
3.2.1 Direct economic damage assessment ... 15
3.2.2 Translation to production losses ... 20
3.2.3 Indirect economic damage assessment ... 23
4. RESULTS ... 27
4.1 DIRECT ECONOMIC DAMAGE ... 27
4.2 INDIRECT ECONOMIC DAMAGE ... 29
4.3 TOTAL ECONOMIC DAMAGE ... 31
4.4 FLOOD DURATION AND ELECTRICITY NETWORK EFFECTS ... 31
5. VALIDITY ... 33
6. RECOMMENDATIONS ... 37
7. CONCLUSION... 38
5
1. Introduction
Harbour areas play a key role in a city’s economic structure by providing transport and trade networks (Koks et al., 2014). However, most harbours face high risks from flooding, caused by pressures such as sea-level rise and socioeconomic growth (Hanson et al., 2011). Most inner-dyke harbours have a low probability of flooding, though consequences are high when flood protection fails (Naulin et al., 2011). For example, flooding of a harbour area could temporarily shut down the transport and trade networks or other important key roles, causing significant economic impact on both local and national scale (Hallegatte et al., 2011; Merz et al., 2010).
The harbour of the city of Amsterdam, the Westpoort, has levees to reduce flood risks and should be able to hold a flood with a probability of occurrence of once every 1,250 years (Drimmelen et al., 2013). However, a recently published study by Jannink et al. (2013) on potential flood risk and damage in the Westpoort harbour shows that water levels can rise up to 1.40 NAP within 24 hours as a result of a breach of the Lekdijk, a potential weakness in the flood protection system (Drimmelen et al., 2013; Jannink et al., 2013). Furthermore, the IJmuiden floodgates connect the North Sea with the Westpoort harbour. Despite of the higher probability of occurrence of once every 10,000 years, a breach of the floodgates could result in a flood of the Westpoort causing water levels to rise up to 1.65 NAP within nine hours (Jannink et al., 2013).
Various studies, such as Klijn et al. (2008) and Messner et al. (2010), have shown the importance of flood risk assessments to effectively improve risk management strategies. However, flood risk assessments are often mainly focused on direct economic effects, defined as the physical damage of flood water to humans, property and the environment (Merz et al., 2004; Messner et al., 2010). The indirect economic damage, the disruption of physical and economic linkages of the economy (Merz et al., 2004), is often barely incorporated within the risk assessment framework. As a result,
6 knowledge gaps exist in the use of indirect economic impact models (Hallegatte, 2008; Messner et al., 2010; Koks et al., 2014). Koks et al., (2014) is one of the first that created an integrative flood risk model for port regions by incorporating both direct and indirect economic impact of floods. However, even though indirect economic impact due to damaged critical infrastructure has been identified as an important factor of a flood risk assessment, the effects of malfunctioning critical infrastructure have barely been used in full-scale flood risk assessments (Merz et al., 2010). As the full economic impact has not yet been identified in the Westpoort area, this thesis primarily aims to assess both the direct damages and indirect losses as a result of a flood in the Westpoort harbour. The primary aim of the research is achieved by answering the following research question:
‘What is the total economic impact as a result of flooding in the Westpoort - Amsterdam?’ Hereby, flooding refers to the inundation scenarios that simulate a breach of the Lekdijk or a breach of the IJmuiden floodgates. Furthermore, total economic impact entails both direct and indirect economic impact. As part of this assessment, the effects of a shutdown of the electricity network in the study area will be identified. Several studies have shown that critical electricity infrastructure proves to be highly vulnerable to floods (Whitehead, 2008; Espada et al., 2013; Jannink et al., 2013). Therefore, it proves to be valuable to identify the indirect economic impact as a result of damage to the electricity network. Finally, the results of the Westpoort harbour are used to fill the knowledge gaps on indirect economic impact modelling and the effects of critical infrastructure shutdown.
This paper will firstly debate on the definition of direct and indirect economic damage. Subsequently, the methodology will describe the model that is used for the food risk assessment and the used inundation scenarios. The results section will then give an overview of the outcomes
7 of the model and the discussion section will compare the results with other studies on the Westpoort and indirect damage models. Furthermore, the discussion will validate the outcomes of the model by comparing the outcomes of this thesis with other studies. The recommendation section formulates recommendations for further research and policy makers. And finally, the conclusion will give an answer to the main research question and the other research aims of this thesis.
8
2. Direct and indirect economic damage
According to Boĉkarjova et al. (2009) it is fundamental to define the concepts of indirect economic damage and narrowly related concepts such as direct economic damage, stocks and flows. The definitions of these concepts are often debated and cause conceptual issues when they are incorrectly used (Hallegatte & Przyluski, 2010; Merz et al., 2010). Therefore, this section will describe the different insights on indirect economic damage and related terms to be able to define how the concept of indirect economic is used in this paper.
An area struck by flooding suffers from economic losses, such as physical damage, interruption of production or even loss of life (Okuyama & Santos, 2014; Hallegatte, 2008). The latter, loss of life, is a difficult issue to measure in monetary values. It is recognized that emotional or psychological damages as a result of loss of life or human injuries have impact on the economy, though the discussion on how to value this impact is not taken into account in this paper as it requires special attention (Boĉkarjova et al., 2009).
There are two main methods of defining direct and indirect loss according to Boĉkarjova et al. (2009). First, the spatial criterion state that all losses attributable to the affected area are classified as direct economic losses, while losses that occurred outside of the disaster area are indirect losses. The second method is based on the stock-flow differential criterion, which defines direct economic losses as stock losses and indirect economic losses as flow interruptions (Boĉkarjova et al., 2009; Rose & Lim, 2002). Herebym stocks refer to a quantity in a single point in time, while flows are services or outputs of stock over time (Rose & Lim, 2002). Following the definitions of the stock-flow differential criterion, indirect economic losses can occur within the disaster area (Boĉkarjova et al., 2009).
9 Even though both methods can be used in flood modelling, the stock-flow differential criterion will be used in this paper, because flows are closely linked to indirect economic losses (Rose & Lim, 2002; Okuyama & Santos, 2014). However, when using stock and flow as main concepts, caution must be taken to avoid double-counting (Rose & Lim, 2002; Messner et al., 2007; Merz et al., 2010). A good is represented by its stock value. Though, the good can also be used as a capital good to generate a flow of income to the owner, whereby the sum of this income over the life span of the capital good can represent the stock value of the good. Hence, in a healthy competitive market, the stock value of the good will equal the sum of flow values over the life span of the capital good (Messner et al., 2007; Rose & Lim, 2002). Consequently, when using both stock and flow values in the flood model, individual components of losses should not include both stock and flow values, as it will result in double-counting (Boĉkarjova et al., 2009; Merz et al., 2010). However, these definitions of stock and flow in flood modelling can be expanded to take the production process into consideration. The production process is the transformation of various inputs into the final products and is influenced by direct damage to capital and the loss of life that is part of lost labour (Hallegatte, 2008; Koks et al., 2014). The economic shock as a result of flooding, causes the loss in production in terms of value-added which can be used to calculate the indirect losses over time without the problem of double-counting stocks and flows (Hallegatte, 2008; Koks et al., 2014). The full implementation of the production process within the model will be described in detail in the methodology section (see page 13).
10
3. Methodology
This section will describe the methodology that is used in this paper and is divided into two parts. Firstly, the disaster scenarios that are used in this flood risk assessment are described. Subsequently, a full description of the integrative flood risk model is given.
3.1 Disaster scenarios
The use of scenarios, to identify drivers of economic losses and to gain more insights on the potential impacts in the study area, is commonly used in flood risk assessment (Hall et al., 2003). In this thesis, the scenarios will be based on the different inundations as a result of a breach of the Lekdijk or the floodgates at IJmuiden combined with variable flood durations to simulate a potential shutdown of the electricity network in the study area.
3.1.1 Inundation modelling
The inundation maps, as used in Deltares & De Urbanisten (2011), are constructed by using a model called Sobek 1D-2D (Deltares, 2012). The model consists of two components, a one dimensional module (1D) that simulates the channel flow and a two dimensional module (2D) that simulates overland flows (Alemseged & Rientjes, 2007; Lomulder, 2004). The 1D module is based on the Saint Venant Equations and uses a basic layer with water depths and flow velocities of the specific river or channel (Vanderkimpen et al., 2009). However, after a breach of a dyke, the 1D assumption of the first module no longer holds. Therefore, the second module models the overland flows based on the shallow water equations as a result of the breach. Breaches are modelled using a time dependent river weir and includes potential growth of the breach (Vanderkimpen et al., 2009; Lomulder, 2004; Alemseged & Rientjes, 2007; Deltares, 2012).
11 Figure 1, inundation maps of the Westpoort harbour that are used for the direct damage assessment, with the upper image the scenario of a breach of the floodgates at IJmuiden and the lower image represents the scenario of a breach of the Lekdijk. (Deltares & De Urbanisten, 2011- modified)
12 In the case of the Westpoort harbour, two inundation maps are made using the Sobek 1D-2D model. Figure 1 presents two inundation maps for the Westpoort harbour, the upper image as a result of a breach the floodgates at IJmuiden and the lower image as a result of a breach of the Lekdijk (Deltares & De Urbanisten, 2011). Each map forms the basis of one of the two groups of scenarios.
3.1.2 Flood duration
Flood duration is the time after the flood until the water recedes and reconstruction activities can commence (Wagenaar, 2012). However, the estimation of flood durations is difficult as it is dependent on specific areas and flood characteristics (Kundu, 2010). Still, both scenarios can be categorised as long lasting floods (>12 hours) (Messner et al., 2007; Penning-Rowsell et al., 2005), as the probability of occurrence is 1/10,000 for the IJmuiden scenario and 1/1,250 for the Lekdijk scenario (Deltares & De Urbanisten, 2011). Furthermore, Wagenaar (2012) and Nicholas et al. (2001) found flood durations in the range of days to weeks for similar cases as the Westpoort. However, because of the uncertainty of the flood duration, several scenarios with different flood durations are used, as shown in table 1(Messner et al., 2007; Wagenaar, 2012).
3.1.3 Electricity network influence
Besides flood duration, reconstruction can be delayed because of a shutdown of the electricity network in the area (Kates et al., 2006; Merz et al., 2010). Electricity networks provide the necessary power to start the reconstruction of an area but power is also required for production of goods by industries that are still able to produce after the flood (Balducci et al., 2002).
According to Erkelens (2014), the electricity network in the Westpoort is distributed with hubs on several voltage levels that are connected with spokes. Even though the spokes are protected from
13 water damage, the hubs are vulnerable when inundated. Inundation depths of above 30 cm are potential treats for 10 kV hubs while 50kV hubs are vulnerable to inundation depths of above 50 cm. Large hubs of 150kV might withstand inundation depths, though it has not been tested (Erkelens, 2014).
To mitigate flood damage, the 50kV and 10 kV hubs are disabled in advance to avoid short circuit. However, it is likely that exceedance of the inundation depth-thresholds of the hubs will cause critical damage, especially in the IJmuiden scenario, as the salt sea water from the North Sea will cause corrosion in the hubs (Stringer, 1998; Kelman & Spence, 2004; Pederson et al., 2006; Zhu et al., 2007). Consequently, each damaged hub must be repaired before the associated area has access to electricity. Furthermore, if 50kV or 150kV hubs are damaged, all lower voltage level hubs are disconnected from the network.
To implement damage to the electricity hubs in the integrative flood model, it is assumed that flood duration is extended due to damage in the electricity network, as the electricity shutdown distribution in the Westpoort show similar patterns as inundated areas (Erkelens, 2014). The amount of damaged hubs, the supply duration for new components and if the 150kV hubs remain functional, are all factors that contribute to what extent the reconstruction is delayed. For instance, according to Jannink et al. (2013) supply duration could be shorted by ‘cannibalising’ other parts
Flood duration length Breach of the Lekdijk Breach of the floodgates at IJmuiden
10 days Lek10 IJ10
20 days Lek20 IJ20
30 days Lek30 IJ30
40 days Lek40 IJ40
50 days Lek50 IJ50
100 days Lek100 IJ100
150 days Lek150 IJ150
200 days Lek200 IJ200
14 of the Dutch electricity network, to accelerate reconstruction activities. Therefore, combined with the uncertainties of actual flood duration, the model will be run with variable flood, as shown in table 1. The last three scenarios are extreme cases in which the inundation depths take longer to recede than expected. Additionally, the 150kV hub might get damaged and delays reconstruction as rare components with long supply durations are required in order to repair it.
3.2 Integrative Flood Risk model
This section will describe the integrative flood risk model that is used to assess the total economic impact as a result of flooding in the Westpoort harbour. The integrative flood risk model can be subdivided into several steps. Firstly, the direct economic impact will be assessed and subsequently the direct economic losses will be translated into loss of production. Lastly, the recovery and the reconstruction duration and costs are calculated to assess the indirect economic impact of the flood (Koks et al., 2014). The latter is based on the Adaptive Regional Input-Output (ARIO) model, developed by Hallegatte (2008, 2014). The complete methodology framework of the integrative flood risk model is presented in figure 2.
Figure 2, the methodology framework used in this paper. The inputs are displayed as drak grey boxes, while the light grey boxes are the outputs of the individual steps (Koks et al., 2014 - modified)
15
3.2.1 Direct economic damage assessment
The first part of the integrative flood model uses depth-damage curves, a common methodology for direct damage assessments (Smith, 1994; Koks et al., 2014). Inundation maps of flood scenarios and a land-use map of the study area are combined with depth-damage curves and maximum damages per land-use class (Koks et al., 2012). The spatial information of the land-use map selects the appropriate damage curve and maximum damage, while the damage factor is determined by the inundation level. Multiplying the maximum damage with the determined damage factor results in the direct economic damage for each land-use (Koks et al., 2012; Moel et al., 2013).
The inputs of the direct damage assessment are the inundation maps displayed in figure 1 and the land-use map as shown in figure 3. The land-use map is made according to the methods presented in Moel et al. (2013). Three datasets (as seen in table 2) are combined to create the land-use map with a resolution of 5 × 5 m. The CBS (‘Centraal Bureau voor Statistiek’) land-use dataset is firstly used to identify different urban uses and differentiation in green land uses, such as parks, sport areas and recreation. Subsequently, the Top10 dataset adds infrastructure and agricultural areas. Then lastly, the building footprints of the BAG (‘Basisregistraties Adressen en Gebouwen’) dataset replaces CBS land-use classes for specific building types, such as residential, industrial or healthcare (Moel et al., 2013).
Merz et al. (2010) has shown the importance of high precision assessments, as these estimations give a more realistic view on the potential economic damage. Furthermore, it appears that the accuracy of the land-use map highly relates to the precision of the direct damage assessment. Therefore, this thesis expanded the land-use classification with 16 specialized classes, mainly focused on industrial sectors based on European Commission (2014), as displayed in table 3. After
16 the aggregation procedure of Moel et al. (2013), the additional sectors are manually classified to the BAG dataset using the Geographical Information System (GIS) of the Port of Amsterdam (Port of Amsterdam, 2014). Consequently, the high resolution and precision land-use map should result in a very specific direct damage assessment.
However, the inundation maps have a resolution of 100 × 100 m instead of the high resolution land-use map of 5 × 5 m. Hence, some areas will have an overestimation while others will have an underestimation of direct damage because of the lower resolution (Messner et al., 2007; Merz et al., 2010). Higher resolution inundation maps will resolve a majority of the uncertainty, though no high resolution inundation maps are available for the Westpoort harbour. However, even with 5 × 5 m inundation maps, some extent of uncertainty will exist in the assessment (Messner et al., 2007; Merz et al., 2010).
17 Figure 3, the land-use map of the Westpoort Amsterdam
18 The depth-damage curves for the corresponding land uses are based on Moel et al. (2013) with the expansion of the land-use classification using Tebodin (1998 & 2000) and Admiraal (2011). The maximum damage values are divided in the values of area and structure damage and content damage (Moel et al., 2013). Table 3 shows the added maximum values for the specific industry classes, other values can be found in Moel et al. (2013). An example of various depth-damage curves is displayed in figure 4.
Even though the depth-damage curves and maximum damages are commonly used, they hold uncertainties (Moel et al., 2013). The estimation of the actual curves and maximum damage values is based on generalisations and methodological differences in estimating the exposed structure or content (Moel et al., 2013; Admiraal, 2011; Merz et al., 2004). Specific research on depth-damage curves in the target area could decrease the uncertainty but is labour intensive and time consuming (Merz et al., 2004). 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0.00 1.00 2.00 3.00 4.00 5.00 Dam ag e fac to r Inundation depths (m)
Depth-damage curves
Industrial Recreation Infrastructure
19 Table 2, datasets that were used in the aggregation procedure (Moel et al., 2013 - modified)
Name Source Type
BAG Kadaster (2011) Points (function) and Polygon (building footprint) Top10 vector Kadaster (2005) Polygon (roads, buildings, etc.)
CBS land use CBS (2008) Polygon (use)
Table 3, maximum damage values for the additional industry classes (Moel et al., 2013; European Commission, 2014; Tebodin, 1998 & 2000)
Class Value area/building (€/m2) Value content (€/m2)
Industrial area 40 -
Industry 1,800 1,200
Shed (industrial) 1,200 1,000
Agriculture 40 -
Mining and quarrying 0.04 -
Food, beverages and tobacco 1000 1200
Textiles and leather 1000 1200
Coke, refined petroleum, nuclear fuel and chemicals
2000 1300
Electrical and optical equipment 1000 1300
Other manufacturing 1000 1200
Construction 1200 1000
Distribution and retail 1400 1200
Hotels and restaurants 1000 1000
Transport, storage and communications
1000 1400
Financial intermediation 1400 1200
Real estate, renting and business activities
1400 1200
Non-market services 1200 1000
Power plant 2500 2000
20
3.2.2 Translation to production losses
In order to determine the economic shock and the pre-recovery period, the principle behind the translation of the direct losses in capital and labour to production losses must be clear (Koks et al., 2014). An Input-Output framework (I-O framework) expresses capital and labour for each sector in the value-added part of the table, as shown in table 4. The value-added of capital and labour combined with intermediate inputs give, column-wise, the total production per sector (Koks et al., 2014). A Cobb-Douglas production function expresses capital and labour in total production (Cobb & Douglas, 1928). So, if labour and capital are known, the value-added per sector can be calculated with the Cobb-Douglas function (Koks et al., 2014). Consequently, changes of value-added can be calculated if the changes of capital and labour are known.
The direct damage, assessed in the direct damage model, is expressed in loss in capital goods and loss in labour. For the latter, the assumption is made that labour is evenly distributed among the
Demand Supply Processing sectors (purchases) Final demand Total output 1 … n Processing sectors (sales) 1 … n A B A+B
Value added Capital C D C+D
Labour E F E+F
Imports G H G+H
Total outlays A+C+E+G Total output
21 sectors. Therefore, the share of flooded industries is approximated to a loss in labour per sector. Hence, using the loss of capital and labour as inputs for the Cobb-Douglas production function yields the loss in value-added as a result of a flood (Koks et al., 2014).
For the calibration of parameters corresponding to the Cobb-Douglas function, the assumption is made that firms produce in an optimal state in the pre-disaster situation. Therefore, the I-O framework represents the pre-disaster values of value-added, capital and labour (Koks et al., 2014). The relationship between the Cobb-Douglas function and the I-O framework is used to determine the economic shock as a result of a flood. Here, the economic shock can be interpreted as the inoperability of a specific sector, or the lack of a sector to fulfill the planned level of production (Barker & Santos, 2010; Santos, 2006). However, the results of the Cobb-Douglas function are expressed in value-added instead of total production loss per sector. The conversion to total production loss per sector is made by assuming the sectoral value-added and the total production output are stably related. Consequently, the economic shock is expressed as the relative sectoral loss for each sector and will act as an input for the I-O framework (Koks et al., 2014).
The next phase of the integrative model is to assess the production loss in the period directly after the flood (Koks et al., 2014). After the actual flood, the struck area remains inundated for a variable amount of time before all the water recedes (see section 3.1.2 Flood duration, page 11; Nicholas et al., 2001; Kundu, 2010; Wagenaar, 2012). Additionally, shutdown of the electricity network could also delay the reconstruction activities (see section 3.1.3 Electricity network influence, page 11; Kates et al., 2006; Merz et al., 2010). It is assumed that the shutdown of the electricity network is only an extension of the pre-recovery period. However, it appears that one of the 150kV hubs at risk, also distributes electricity to about 60% of Amsterdam (Jannink et al., 2013; Erkelens, 2014).
22 Hence, the indirect damage is underestimated by the model if the specific 150kV hub is critically damaged by the flood (Jannink et al., 2013).
The production loss in the period directly after the flood is assessed using the Basic Equation (BE), as formulated by Steenge & Boĉkarjova (2007). The BE adjusts the original open I-O framework as defined by Leontief (1951), to an adapted closed I-O model (Smith, 1951; Boĉkarjova et al., 2004). The adapted closed I-O model represents a balanced pre-disaster economy in equilibrium with variables such as total production input, final demand and total employment (Steenge & Boĉkarjova, 2007). Subsequently, the assessed economic shock is integrated within the BE to describe the post-disaster system. However, the post-disaster economy is no longer in equilibrium due to the economic shock. Hence, the output of the BE represents the disturbed proportions between inputs and outputs in the pre-recovery phase of a flood (Koks et al., 2014; Steenge & Boĉkarjova, 2007). The disproportions are caused by the process of inputs of producers that become scarce as suppliers suffer from flood losses, while inputs of sectors without flood losses become redundant. Consequently, producers are forced to reduce or temporarily cease production (Koks et al., 2014).
It is assumed that the recovery aims to achieve pre-disaster production capacities and inter-industrial relations, the technological coefficients, are not changed after the economic shock. Thus, after the recovery phase, final demand should be met by the system (Koks et al., 2014).
In the pre-recovery period, the daily production loss (DPL) as a result of the inundation of the flooded area and a potential shutdown of the electricity network is assessed. The DPL per sector is defined in terms of the change in production during the pre-recovery phase when the area is inundated or unable to start reconstruction (Koks et al., 2014).
23
3.2.3 Indirect economic damage assessment
In the recovery period, the area is no longer inundated and the reconstruction starts. To assess the indirect damage in this period, the Adaptive Regional Input-Output (ARIO) model is used, developed by Hallegatte (2008 & 2014). The model uses an I-O framework to assess the indirect economic effects in response to a natural disaster. The ARIO model is dynamic as it overcomes two limitations of static I-O models; the inability to assess the consequence of a shock on the supply side of the economy and the lack of flexibility in the economic system (Hallegatte, 2008; Ranger et al., 2011; Wu et al., 2012).
The first limitation is resolved by assessing the productive capacity of each sector in the disaster area. The induced damage, as a result of a flood, reduces the production capacity of an industry which makes it unable to produce. Additionally, undamaged industries in and outside the disaster area that do not receive enough resources from damaged industries are also not capable of producing. Furthermore, some sectors, such as the construction sector and public utilities, have an increase in demand due to reconstruction. Consequently, ARIO model uses this reduction in production due to the economic shock, to change the supply side of the economic system (Hallegatte, 2008, 2014; Ranger et al., 2011).
The second limitation is a lack of flexibility in the economic system, defined as the inability of producers and consumers to respond to a lack of input. The ARIO model uses inventories and I-O table dynamics to model the flexibility in the production system (Hallegatte, 2014). For example, producers can delay maintenance when inventory stocks are present. Furthermore, the use of inventories allows for the modelling of the use of imports more accurately and the possibility of rescheduling production (Hallegatte, 2014).
24 To assess the indirect damage, the ARIO model calculates the maximum possible production capacity, based on the output of the Basic Equation (see section 3.2.2 Translation to production loss, page 19). Simultaneously, the reconstruction costs are determined from the direct economic damage to assess the final demand and the required production to complete the reconstruction (Koks et al., 2014; Wu et al., 2012). Subsequently, the maximum possible production capacity and the required total production are compared to identify the production available for reconstruction, other demand and export and how many intermediate orders between sectors are satisfied. Finally, the remaining reconstruction costs and the remaining damage in capital and labour can be identified (Koks et al., 2014; Wu et al., 2012; Hallegatte, 2008). The latter is re-assessed according to the Cobb-Douglas production function (see section 3.2.2. Translation to production loss, page 19).
The output of the model, remaining reconstruction costs and remaining damage in capital and labour, can be reused as inputs to create two loops that simulate time steps until the final demand is met and reconstruction is successful. The last step of the model is to calculate the loss in value-added for each time step, based on the reduction of the maximum production capacity (Wu et al., 2012). Consequently, the indirect damage is calculated as the difference between the total value-added without flooding and the total value-value-added with flooding for each time step (Koks et al., 2014; Wu et al., 2012). A schematic framework of the ARIO model is displayed in figure 5, and for a more extensive description of the model, see Hallegatte (2008 & 2014).
25 As the ARIO model is dependent on a large number of parameters, assumptions have to be made for the economic structure of the study area (Koks et al., 2014). Expert knowledge of the study area (Drimmelen et al., 2013; Jannink et al., 2013), and disaster modelling literature will be used to support the made assumptions (Koks et al., 2014; Halltegatte, 2008, 2014; Okuyama, 2004 Penning-Rowsell et al., 2005; Nicholas; 2001).
Firstly, according to Okuyama (2004), three different production modes can be distinguish for the amount of inventory per sector, namely; anticipatory, responsive and just-in-time. The anticipatory mode describes production that is anticipated on future orders, with an assumed inventory stock of 90 days. Agriculture and mining sectors are included in this mode. Responsive is the production that occurs after receipt of customer order, in sectors such as construction, services industries and Figure 5, a schematic display of the main steps of the ARIO model (Koks et al., 2014)
26 manufacturing industries. It is assumed that the responsive mode has an inventory stock of 60 days. The just-in-time mode is assumed to have an inventory stock of three days as it entails production that takes place as the order is placed. Public utilities is an example sector with a just-in-time production (Okuyama, 2004; Koks et al., 2014).
Furthermore, it is assumed that the study area has a high heterogeneity in goods and services between and within sectors and that labour fully recovers in three months. Finally, the assumption is made that companies in the area do not leave the area because of the flood and fully reconstruct their labour and capital to their initial state (Hallegatte, 2008). However, companies might lose trust in the location and leave the area (Jannink et al., 2013). Additionally, companies might decide to rebuild their company differently to increase efficiency that will result in a higher total value-added than in the initial state (Hallegatte, 2008). Other assumptions and parameter settings for the ARIO model are based on Koks et al. (2014) and Hallegatte (2014).
These assumptions bare some amount of uncertainty with them, though they are unavoidable (Merz et al., 2010). The assumptions were thoroughly tested and reviewed in Koks et al. (2014) and Hallegatte (2008, 2014), which showed that the used parameters are highly sensitive for changes. Therefore, caution is required when interpretation the results, as the outcomes might resemble concrete results, they are only indications of the magnitude of a possible flood (Merz et al., 2010; Koks et al., 2014).
Even though there is a degree of uncertainty within the model, the results should be well suited to gain insight on the study area and an indication of the costs of a potential flood in the Westpoort harbour (Koks et al., 2014; Merz et al., 2010). Still, more research is required to implement more macro-economic mechanisms and decrease the amount of assumptions within flood models to achieve more reliable and accurate results (Hallegatte, 2008; Koks et al., 2014).
27
4. Results
This section will firstly present the results for the direct damage assessment and subsequently, the indirect damage results are shown. Combining the direct and indirect damage assessments, the total economic damage is given. Lastly, the effects of different flood durations and shutdown of the electricity network are presented.
4.1 Direct economic damage
The total direct economic is assessed at 1.54 billion euros for the Lekdijk, as shown in table 6 (see page 28). For the IJmuiden scenario, the direct damage is assessed at 6.28 billion euros, as shown in table 7 (see page 28). The five sectors that are affected by most direct damage from flooding is displayed in table 5 for the Lekdijk and IJmuiden scenarios. For example, the distribution and retail sector and transport, storage and communications sector are affected most, as both are the two largest industrial sectors of the Westpoort (Port of Amsterdam, 2013).
Lekdijk Direct damage
(Millions)
Percentage of direct damage
Distribution and retail 741 48%
Transport, storage and communications 227 15%
Houses 92 6%
Real estate, renting and business activities 70 5%
Railroads 64 4%
IJmuiden Direct damage
(Millions)
Percentage of direct damage
Distribution and retail 2,666 42%
Transport, storage and communications 1,147 18%
Railroads 421 7%
Real estate, renting and business activities 330 5% Coke, refined petroleum, nuclear fuel and
chemicals
306 5%
Table 5, direct damage results for the Lekdijk and IJmuiden scenarios. The five most damaged sectors are displayed in absolute numbers and expresses as a percentage of direct damage
28 The significant difference in the assessed direct damages between the Lekdijk and the IJmuiden scenarios is notable. A breach of the floodgates at IJmuiden will connect the Westpoort harbour directly to the North Sea, causing higher inundation depths than the Lekdijk scenario. Consequently, the direct damage is higher for the IJmuiden scenario. However, four areas are not or barely inundated in the Lekdijk scenario but are inundated in the IJmuiden scenario. These four areas are the north embankment area of the Westpoort, the sewage waste treatment plant located in the south-west, oil terminals adjacent to the Petroleum, Amerika, Mauritius and Usselincx harbours and the Alfa Triangle located in the south-east.
Firstly, the north embankment area of the Westpoort harbour is densely filled with distribution and retail activity. The inundation of this area strongly contributes to the difference of almost two billion in direct damage in the distribution and retail sector between the Lekdijk and the IJmuiden scenario, as shown in table 5. Secondly, the associated direct damage due to the flooding of the sewage waste treatment plant in the IJmuiden scenario is assessed at 143 million euros or 2% of the total direct damage. Thirdly, most of the oil storage and blending is struck by a flood of the IJmuiden scenario that accounts for 306 million euros of direct damage. Lastly, the Alpha Triangle, an area located in the south-east of the Westpoort, has a dense business sector. The inundation of this area contributes to the increase of 260 million in the real estate, renting and business activities sector.
It should be noted that the power plant in the Westpoort does suffer from about 16 million euros in direct damage for the Lekdijk scenario and about 70 million euros for the IJmuiden scenario. However, this also includes all buildings related to the power plant and does not give an indication for the main building.
29 Table 6, direct, indirect and total damage results for the Lekdijk scenarios. Additionally the time in days for the total value-added to reach 99% of the initial value is shown
Table 7, direct, indirect and total damage results for the IJmuiden scenarios. Additionally the time in days for the total value-added to reach 99% of the initial value is shown
4.2 Indirect economic damage
In the Lekdijk scenario, indirect damage is assessed at 0.5 to 1.1 billion euros for the different scenarios, as shown in table 6. The IJmuiden scenarios are assessed at 2.1 billion to 4.0 billion, as shown in table 7. The higher values of the IJmuiden scenarios originate from the high economic shock due to the higher direct damage associated with the higher inundation depths. Furthermore, it is assessed that for the Lekdijk scenarios the total value-added for the area will be restored at 99% of the initial values in 80 to 200 days, depending on the flood duration time. In the case of IJmuiden, it is assessed that the total value-added will reach 99% of the initial value after 700 to 900 days. Both of these time ranges reflect the reconstruction time if all damaged companies decide
Scenario Direct damage (in billion euros)
Indirect damage (in billion euros)
Total damage (in billion euros)
At 99% of the initial value-added Lek10 1.54 0.47 2.02 84 days Lek20 1.54 0.48 2.03 85 days Lek30 1.54 0.50 2.04 85 days Lek40 1.54 0.52 2.06 86 days Lek50 1.54 0.55 2.09 86 days Lek100 1.54 0.73 2.27 107 days Lek150 1.54 0.92 2.46 157 days Lek200 1.54 1.11 2.65 206 days
Scenario Direct damage (in billion euros)
Indirect damage (in billion euros)
Total damage (in billion euros)
At 99% of the initial value-added IJ10 6.28 2.09 8.37 714 days IJ20 6.28 2.14 8.42 724 days IJ30 6.28 2.19 8.47 735 days IJ40 6.28 2.26 8.54 746 days IJ50 6.28 2.34 8.61 755 days IJ100 6.28 2.84 9.12 805 days IJ150 6.28 3.44 9.72 855 days IJ200 6.28 4.03 10.31 907 days
30 Figure 6, the total value-added after the flood event for the scenarios Lek20 and IJ20
to reconstruct instead of moving to a different region and that no reconstruction help is offered from outside the area (Koks et al., 2014).
The development of the total value-added is compared in figure 6 for the Lek20 and IJ20 scenarios. Even though figure 6 starts in the post-flood period, the economy is not directly at its lowest point, as most companies still have inventories of stock. Once these stocks run out and production falls behind, the value-added is at its minimum (Koks et al., 2014). The IJmuiden scenario has a lower minimum than the Lekdijk scenario, explained by the higher associated direct damage that causes stronger economic shocks per sector. In both scenarios labour recovers with a maximum of three months causing a rapid increase of value-added. In subsequent months, only remaining damaged capital has to be rebuild. Because of the similarity in labour recovery, the initial recovery of value-added is at the same steepness for both scenarios but after labour is fully recovered, the reconstruction of the IJmuiden scenario takes longer as more capital has to be rebuild, as seen in figure 6. 26.5 27.0 27.5 28.0 28.5 29.0 29.5 30.0 30.5 31.0 31.5 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 T o tal v alu e-ad d ed ( in b illi o n eu ro s)
Time (in days)
Total value-added after the flood event
31 Figure 7, the total value-added after the flood event for all flood duration scenarios for IJmuiden
4.3 Total economic damage
Combining both the direct and indirect damage as a result of the several flood scenarios results in the total economic damage, as shown in table 6 and 7 (see page 28), for the Lekdijk and IJmuiden floods respectively. Consequently, a breach of the Lekdijk could potentially yield in total economic damage of about 2 to 2.7 billion euros and about 8.4 to 10.3 billion euros as a result of a breach of the IJmuiden floodgates. As mentioned in the methodology section (see page 13), caution must be taken when interpreting these estimates, as these are indication of the magnitude of economic impact and not concrete measures.
4.4 Flood duration and electricity network effects
The effects of variation in flood duration, as a result of longer inundation duration or the shutdown of the electricity network, is shown in table 6 and 7 for the Lekdijk and IJmuiden scenarios respectively. Additionally, the development of the total value-added in the post-flood period is shown in figure 7 for the Lekdijk scenarios and in figure 8 for the IJmuiden scenarios.
29.0 29.2 29.4 29.6 29.8 30.0 30.2 30.4 30.6 30.8 31.0 31.2 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 T o tal v alu e-ad d ed ( in b illi o n eu ro s
Time (in days)
Total value-added after the flood event for Lekdijk
scenarios
Lek10 Lek20 Lek30 Lek40
32 Figure 8, the total value-added after the flood event for all flood duration scenarios for the Lekdijk
For the first five scenarios with flood durations from 10 to 50 days, the indirect damage only increases relatively little. In case of the Lekdijk scenarios, the difference between the Lek10 and Lek50 scenarios is only 70 million euros and for the IJmuiden scenarios the difference is 240 million. The scenarios with long flood durations of 100 to 200 days, differences in indirect damage are significant. For the Lekdijk scenarios it results in up to 0.5 billion euros and for IJmuiden up to two billion euros. In figure 8, it is seen that economy in the IJ100, IJ150 and IJ200 scenarios has more difficulty to recover. As these patterns are not notable in figure 7 for the Lekdijk scenarios, it appears that after a certain threshold of economic shock, the regional economy experiences difficulties in reaching the initial total value-added, which is also found in Koks et al. (2014). It should be noted that the IJ100, IJ150 and IJ200 are extreme scenarios and are less likely to occur. However, it does give an indication of the vulnerability of the economy after a flood event due to longer inundation times or because of inaccessibility of crucial repair components for the electricity network hubs.
26.5 27.0 27.5 28.0 28.5 29.0 29.5 30.0 30.5 31.0 31.5 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 T o tal v alu e-ad d ed ( in b illi o n eu ro s)
Time (in days)
Total value-added after the flood event for the IJmuiden
scenarios
33
5. Validity
This section compares the results of this thesis with the direct damage assessment made by by Deltares & De Urbanisten (2011) and the ‘Veiligheid Nederland in Kaart 2’ (VNK2) report (Bisschop & Huisman, 2011; Projectbureau VNK2, 2012) in order to validate the results for the study area. Subsequently, the results are also compared with Koks et al. (2014) and Hallegatte (2008) as they used the same models but for other cases.
A rough direct damage assessment was made by Deltares & De Urbanisten (2011) for similar scenarios of the Lekdijk and IJmuiden as used in this paper, to give an estimate for the potential losses as a result of a potential flood. The direct damage for the Lekdijk scenario, which also included other inundated areas of Amsterdam, was assessed at one to five billion euros. Given the assessment of Deltares & De Urbanisten (2011) also included the ‘IJ-oevers’, ‘Tuindorp-Oostzaan’, ‘Zuidoost’ and ‘Watergraafsmeer’ areas, the results of this thesis show similar results. For the IJmuiden scenario, Deltares & De Urbanisten (2011) also assessed the direct damage at one to five billion euros. Anew, other areas, ‘IJ-oevers’, ‘Tuindorp-Oostzaan’ and ‘Watergraafsmeer’, were also taken into consideration in their assessment, and therefore the results of this thesis, 6.28 billion for the Westpoort harbour alone, are relatively high.
Subsequently, the VNK2 report assessed direct damage for similar IJmuiden scenario at one billion euros (Bisschop & Huisman, 2011; Projectbureau VNK2, 2012). However, this assessment also include all flooded areas in the area between IJmuiden and the ‘IJmeer’, east of Amsterdam, as seen in figure 9. In comparison with the assessed direct damage of 6.3 billion for the Westpoort only, the difference is significant.
34 A possible explanation for the
difference in assessed direct damage is that the methodology used by the VNK2 that does not accurately include industry in the calculations. The model used by the VNK2, the HIS-SSM model, assess the direct damage by inundation maps and a land-use map of the study area. However, the land-use map is based on postal codes (Zethof et al., 2011).
For a postal code zone, the corresponding land-use is assigned to the centre of that area. For densely populated areas this is not a problem, though in outer-dyke areas or industrial areas the use of postal codes causes inaccurate outcomes, because firstly, it does not incorporate elevation differences in a single postal area. Secondly, the location of the centre of a postal zone could not represent the actual location of the building or even lie outside the postal code area. Lastly, the HIS-SSM model only counts a land-use as flooded if the centre of the postal zone is flooded. Therefore, if large postal code zones are present is the study area, a significant area could flood but not registered by the model because the centre points are outside the flooded area (Zethof et al., 2011). The latter is most likely the case for the Westpoort and thereby the model used in this thesis should give a more accurate indication for potential direct damage as a result of a flood than the model used in the VNK2 report. However, further research should specifically identify the difference in assessed damage between both models.
Figure 9, inundation map of the VNK2 report for the IJmuiden scenario. The assessed direct damage is accounting all inundated areas
35 Figure 10, inundation map of the VNK2 report for the Lekdijk scenario. The assessed direct damage is accounting all inundated areas
The VNK2 report also assessed the direct damage as a result of a breach of the Lekdijk with several maximum inundation depths, based on different probabilities of occurrence from 1/250 up to 1/10,000. In total the direct damage is assessed between 8 billion to 14 billion euros, including all areas that are flooded because of a breach, as displayed in figure 10(Bisschop & Huisman, 2011; Projectbureau VNK2, 2012). Because the Westpoort harbour is only a fraction of the total flooded area, it is difficult to compare the results. However, the assessed 1.5 billion euros direct damage for the Lekdijk scenario is relatively large compared to the 8 billion to 14 billion euros for the total flooded area. Therefore, it is likely that the assessed damage for the Westpoort is higher in this thesis than the direct damaged assessed by the VNK2 report.
The ratio between direct and indirect damage for the Lekdijk and IJmuiden is 75% to 25% for the Lek10 and IJ10 scenarios and 60% to 40% for the Lek200 and IJ200 scenarios. In Hallegatte (2008) a ratio of 60% to 40% was found, which is similar to the findings of this thesis for the more extreme scenarios of this thesis. However, Koks et al. (2014) found ratios of 40% to 60% for high probability floods (1/10,000) and ratios of 60% to 40% for low probability floods (1/10). The case study of Rotterdam is different as the port area only floods for about 10% in the scenario of 1/10,000 (Koks et al., 2014), while the Westpoort is inundated for about 90% for the IJmuiden scenario. Additionally, the Port of Rotterdam is about five times larger than the Westpoort harbour
36 (Port of Amsterdam, 2014; Port of Rotterdam, 2014). Because of the difference in sizes and inundated area, both cases are difficult to compare, though the different findings of the ratio between direct and indirect damage might be explained with bottlenecks of production.
As only a relatively small part of the port area of Rotterdam is flooded, other non-damaged companies might not be able to continue production because of a lack of input from the damaged sectors. Moreover, the most damaged sector appears to be the retail sector, that could cause bottlenecks in the distribution of products to consumers (Hallegatte, 2008). In case of the Westpoort, the majority of the sectors is damaged and therefore, only a small part of the indirect damage is a result of bottlenecks of production because only a small part of the sectors are undamaged and able to produce.
Additionally, even though the direct damage assessment is more accurate than the direct damage assessment of Koks et al. (2014), the indirect damage assessment is more extensive for the case of Rotterdam than commenced in this thesis for the Westpoort (Koks et al., 2014). Hence, the difference in accuracy might give different ratios (Messner et al., 2007; Merz et al., 2010). Consequently, the found ratios are differently distributed that the ratios found in Koks et al. (2014), but might be explained by the amount of production bottlenecks resulted from flooded sectors and difference in accuracy of the direct and indirect assessments. Further research is required to confirm these speculations.
37
6. Recommendations
Based on the results and the discussion, several recommendations are made for further research and policy making proposes. Firstly, it is recommended to more accurately specify the assumptions. Because of the time scale of this thesis, it was not possible to identify specific characteristics of the economy in the area, so assumptions were made based on scientific literature. However, in the processes, generalisations were made that could be prevented by specifically researching these characteristics. Mainly the inventory stocks, inundation map resolution and labour recovery parameters can become more accurate if extensive research is done on these topics specifically.
Secondly, it is urged to verify the results of the VNK2 model as it is used in the Delta Programme and the Flood Safety sub-programme (WV21) (VNK2, 2011). If the results of the VNK2 model underestimate the direct damage for industrial and outer-dyke areas, as shown in the discussion section (see page 32), the upcoming decisions of the Delta Programme might lead to incorrect flood management policies. Hence, a possible flood of the Westpoort or other similar areas might result in higher consequences than anticipated for.
Lastly, it is advised to incorporate the indirect economic damage in the flood damage assessments for both scientific and policy purposes. Even though the indirect economic damage as assessed by the integrative risk model is dependent of assumptions, it can still give a good indication for possible economic losses that are currently not accounted for in flood or water management policies.
38
7. Conclusion
This thesis assessed the total economic damage for the Westpoort harbour in Amsterdam as a result of a potential flood. Two main scenarios were made, one based on a breach of the Lekdijk and the other based on a breach of the floodgates at IJmuiden. Additionally, the flood duration parameter was varied to show the effects of longer inundation times and a potential shutdown of the electricity network. An integrative flood risk model, developed by Koks et al. (2014), was used to assess the direct economic damage and indirect economic damage. Based on the results of the integrative flood risk model, the total economic damage is assessed to give an answer to the main research question, which can be formulated as follows;
For the Lekdijk scenario, total economic damage is assessed at 2.0 billion to 2.7 billion euros for the different flood duration scenarios. Hereby, 1.5 billion euros are accounted for direct damage and between 0.5 billion to 1.1 billion for indirect damage. Furthermore, it is estimated that reconstruction will last for 80 to 200 days to return to 99% of the initial total value-added. For the IJmuiden scenario, direct damage is assessed at 6.3 billion euros and indirect damage at 2.1 billion to 4.0 billion euros for the different flood duration scenarios. Consequently, a total of between 8.4 billion to 10.3 billion euros is assessed for the IJmuiden scenario and a recovery time of between 700 and 900 days to reach 99% of the initial value-added.
Besides the main research aim to identify the total economic damage as a result of a flood, it was also aimed to identify the effects of the shutdown of the electricity network. Variation of the flood durations parameter in the different scenarios gives an indication of the effects of a potential shutdown. It appears indirect damage could increase significantly, up to two billion for the IJmuiden scenarios, due to the associated delay of reconstruction.
39 Even though the model is highly dependent on many assumptions, the results give an indication of the magnitude of potential costs and losses as a result of a flood. Further research could reduce uncertainty within the model by identifying specific characteristics of the study area. Finally, indirect economic damage appears to be a significant part of the total damage, and therefore indirect economic damage should be taken into consideration when assessing flood risks or adjusting flood management policies.
40
8. Bibliography
Admiraal, J. (2011). Flood damage to port industry, Case study: vulnerability of the port of Rotterdam to climate change. Arcadis, VU University of Amsterdam.
Alemseged, T. H., & Rientjes, T. H. (2007). Uncertainty issues in hydrodynamic flood modeling. In Proceedings of the 5th international symposium on spatial data quality SDQ.
Balducci, P. J., Roop, J. M., Schienbein, L. A., DeSteese, J. G., & Weimar, M. R. (2002).
Electrical Power Interruption Cost Estimates for Individual Industries, Sectors, and U.S. Economy. Richland, Washington: Pacific Northwest National Laboratory.
Barker, K., & Santos, J. R. (2010). Measuring the efficacy of invertory with a Dynamic Input0Output Model. In International Journal of Production Economics (Vol. 126, pp. 130-143). doi:10.1016/j.ijpe.2009.08.011
Bisschop, C., & Huisman, C. (2011). Veiligheid Nederland in Kaart 2 Overstromingsrisico dijkring 44 Kromme Rijn. Rijkswaterstaat Waterdienst, Ministerie van Infrastructuur en Milieu, Unie van Waterschappen en Interprovinciaal Overleg.
Bočkarjova, M., Steenge, A. E., & Hoekstra, A. Y. (2009). Management of Flood Catastrophes An Emerging Paradigm Shift? In Water Policy in the Netherlands: Integrated
Management in a Densely Polulated Delta (pp. 70-90). Resiyres for the Future.
Bočkarjova, M., Steenge, A. E., & Veen, A. v. (2004). On direct estimation of initial damage in the case of a major catastrophe: derivation of the “basic equation”. In Disaster Prevention and Management (4 ed., Vol. 13, pp. 330 - 336). doi:10.1108/09653560410556555
41 Canadian Press. (2013, September 13). Water slowly receding from Calgary, flood threat moving
toward Medicine Hat region. Retrieved June 16, 2014, from National Post:
http://news.nationalpost.com/2013/06/22/water-slowly-receding-from-calgary-flood-threat-moving-toward-medicine-hat-region/
CBS. (2008). Bestand bodemgebruik. Den Haag, Centraal Bureau voor de Statistiek, Nederland. Retrieved from
http://www.cbs.nl/nl-NL/menu/methoden/dataverzameling/bodemgebruik.htm
Cobb, C. W., & Douglas, P. H. (1928). A Theory of Production. In American Economic Review (pp. 139-165).
Deltares & De Urbanisten. (2011). De Waterbestendige stad Achtergrondrapport. DHV B.V. Deltares. (2012). Sobek - D-Flow 2D Overland. Retrieved May 5, 2014, from Deltares Systems:
http://www.deltaressystems.com/hydro/product/108282/sobek-suite/1168718 Drimmelen, C. v., Koeze, R., & Monchen, E. (2013). Waterbestendig Amsterdam. PLAN
Amsterdam.
Erkelens, M. (2014 - unpublished). Bachelor thesis: Hydrologische kwetsbaarheidanalyse van de Amsterdamse Westpoort. Amsterdam.
European Commission. (2014, January 30). Business economy by sector - NACE Rev. 2. Retrieved June 4, 2014, from Eurostat:
http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/Business_economy_by_se ctor_-_NACE_Rev._2
42 Hall, J. W., E. P., Penning-Rowsell, E. C., Sayers, P. B., Thorne, C. R., & Saul, A. J. (2003).
Quantified scenarios analysis of drivers and impacts of changing flood risk in England and Wales: 2030–2100. In Environmental Hazards (Vol. 5, pp. 51-65). Elsevier Ltd. doi:10.1016/j.hazards.2004.04.002
Hallegatte, S. (2008). An Adaptive Regional Input-Output Model and its Application to the Assessment of the Economic Cost of Katrina. In Risk Analysis (Vol. 28, pp. 779-799). Society for Risk Analysis.
Hallegatte, S. (2014). Modeling the Role of Inventories and Heterogeneity in the Assessment of the Economic Costs of Natural Disasters. In Risk Analysis (Vol. 34, pp. 152-167). Society for Risk Analysis. doi:10.1111/risa.12090
Hallegatte, S., & Przyluski, V. (2010). The Economics of Natural Disasters - Concepts and Methods. The World Bank.
Hallegatte, S., Ranger, N., Mestre, O., Dumas, P., Corfee-Morlot, J., Herweijer, C., & Wood, R. M. (2011). Assessing climate change impacts, sea level rise and storm surge risk in port cities: a case study on Copenhagen. In Climatic Change (Vol. 104, pp. 133-137). Springer.
Hanson, S., Nicholls, R., Ranger, N., Hallegatte, S., Corfee-Morlot, J., Herweijer, C., & Chateau, J. (2011). A Global Ranking of Port Cities with High Exposure to Climate Extremes. In Climatic change (Vol. 104, pp. 89-111). Springer.
Jannink, P., Berkel, S. v., Rep, F., Buul, L. v., Berk, Z. v., Ravenstijn, P., . . . Valkenburg, L. (2013). Waterbestendige Westpoort. Amsterdam: MUST stedebouw & Witteveen + Bos.
43 Kadaster. (2005). Top10vector. Apeldoorn: Topografische Dienst Kadaster.
Kadaster. (2011). Basisregistraties adressen en gebouwen (BAG). Apeldoorn: Topografische Dienst Kadaster.
Kates, R. W., Colten, C. E., Laska, S., & Leatherman, S. P. (2006). Reconstruction of New Orleans after Hurricane Katrina:. In PNAS (Vol. 103, pp. 14653–14660).
Kelman, I., & Spence, R. (2004). An overview of flood actions on buildings. In Engineering Geology (pp. 297 – 309). Elsevier B.V. doi:10.1016/j.enggeo.2004.01.010
Klijn, F., Samuels, P., & Os, A. v. (2008). Towards flood risk management in the EU: State of affairs with examples from various European countries. In International Journal of River Basin Management (4 ed., Vol. 6, pp. 307-321). doi:10.1080/15715124.2008.9635358 Koks, E., Boĉkarjova, M., Moel, H. d., & Aerts, J. (2014). Development and sensitivity analysis
of an integrative flood risk model for port regions.
Koks, E., Moel, H. d., & Koomen, E. (2012). Comparing Extreme Rainfall and Large-Scale Flooding Induced Inundation Risk – Evidence from a Dutch Case-Study. In Studies on water management issues (pp. 3-26). Intech, Rijeka. doi:10.5772/30378
Kundu, S. (2010). Master Thesis: Flood Characterisation and Inundation Pattern Mapping using Radersat Imagery for Rice Vulnerability Assessment. Enschede, The Netherlands: International Institue for Geo-information Science and Earth Observation.
Leontief, W. W. (1951). Input–output economics & analysis. In Input-output Economics (pp. 15-27). Oxford University Press.
44 Li, J., Crawford-Brown, D., Syddall, M., & Guan, D. (2013). Modeling Imbalanced Economic
Recovery Following a Natural Disaster Using Input-Output Analysis. In Risk Analysis (10 ed., Vol. 33, pp. 1908-1923). Society for Risk Analysis. doi:10.1111/risa.12040 Lomulder, R. (2004). Appropriate modelling: Application of Sobek 1D2D for dike break and
overtopping at the Elbe. Universiteit Twente.
Merz, B., Kreibich, H., A., T., & Schmidtke, R. (2004). Estimation uncertainty of direct monetary flood damage to buildings. In Natural Hazards and Earth System Sciences (Vol. 4, pp. 153-163). European Geosciences Union.
Merz, B., Kreibich, H., Schwarze, R., & Thieken, A. (2010). Review article "Assessment of economic Flood damage". In Natural Hazards and Earth System Sciences (Vol. 10, pp. 1697-1724). Hamburg: Copernicus Publications.
Messner, F., Penning-Rowsell, E., Green, C., Meyer, V., Tunstall, S., & Veen, A. v. (2007). Evaluating flood damages: guidance and rcommendations on principles and methods. FLOODsite Consortium.
Moel, H. d., Vliet, M. v., & Aerts, J. C. (2013). Evaluating the effect of flood damage-reducing measures: a case study of the unembanked area of Rotterdam, the Netherlands. In Regional Environmental Change (pp. 895-908). Springer. doi:DOI 10.1007/s10113-013-0420-z
Moel, H. d., Vliet, M. v., & Aerts, J. C. (2014). Evaluating the effect of flood damage-reducing measures: a case study of the unembanked area of Rotterdam, the Netherlands. In Regional Environmental Change (Vol. 14, pp. 895-908). Springer. doi:10.1007/s10113-013-0420-z
45 Naulin, M., Kortenhaus, A., & Oumeraci, H. (2011). Failure Probability of Flood Defence
Structures/Systems in Risk Analysis for Extreme Storm Surges. In Coastal Engineering Proceedings (Vol. 32, pp. 1-15).
Nicholas, J., Holt, G., & Proverbs, D. (2001). Towards standardising the assessment of flood damaged properties in the UK. In Structural Survey (Vol. 4, pp. 163-172). MCB University Press.
Okuyama, Y. (2004). Modeling spatial economic impacts of an earthquake: input-output approaches. In Disaster Prevention and Management (4 ed., Vol. 13, pp. 297-306). Emerald Group Publishing Limited. doi:10.1108/09653560410556519
Okuyama, Y., & Santos, J. R. (2014). Disaster Impact and Input-Output analysis. In Economic Systems Research (Vol. 26, pp. 1-12). London: Routledge.
doi:http://dx.doi.org/10.1080/09535314.2013.871505
Pederson, P., Dudenhoeffer, D., Hartley, S., & Permann, M. (2006). Critical Infrastructure Interdependency Modeling: A Survey of U.S. and International Research. Idaho Falls: Idaho National Laboratory.
Penning-Rowsell, E., Johnson, C., Tunstall, S., Tapsell, S., Morri, J., J. C., & Green, C. (2005). The Benefits of Flood and Coastal Risk Management: A Handbook of Assessment Techniques. London: Middlesex University Press.
Port of Amsterdam. (2013). Statistics 2013. Amsterdam: Port of Amsterdam.
Port of Amsterdam. (2014). Cijfers, Factsheet uitgave 3. Amsterdam: Havenbedrijf Amsterdam N.V.
46 Port of Amsterdam. (2014, May). Port Map. Retrieved June 4, 2014, from
http://www.portofamsterdam.com/Eng/port-map
Port of Rotterdam. (2014). Haven in Cijfers. Rotterdam: Havenbedrijf Rotterdam N.V. Projectbureau VNK2. (2012). Veiligheid Nederland in Kaart Tussenresultaten fase 1b.
Ministerie van Infrastructuur, Interprovenicaal overleg, Unie van Waterschappen. R. Espada, J., Apan, A., & McDougall, K. (2013). Using spatial modelling to develop flood risk
and climate adaptation capacity metrics for assessing urban community and critical electricity infrastructure vulnerability. In 20th International Congress on Modelling and Simulation. Adelaide.
Ranger, N., Hallegatte, S., Bhattacharya, S., Bachu, M., Priya, S., Dhore, K., . . . Corfee-Morlot, J. (2011). An assessment of the potential impact of climate change on flood risk in Mumbai. In Climatic Change (Vol. 104, pp. 139-167). Springer Science.
doi:10.1007/s10584-010-9979-2
Rose, A., & Lim, D. (2002). Business interruption losses from natural hazards: conceptual and methodological issues in the case of the Northridge earthquake. In Environmental Hazards (Vol. 4, pp. 1-14). Elsevier Science Ltd.
Santos, J. R. (2006). Inoperability Input-Output Modeling of Disruptions to Interdependent Economic Systems. In Systems Engineering (Vol. 9). Wiley Periodicals, Inc. doi:10.1002/sys.20040
Smith, D. (1994). Flood damage estimation - A review of urban stage-damage curves and loss functions. In Water SA (Vol. 20, pp. 231-238).
47 Smith, H. M. (1951). Uses of Leontief's open input-output model.
Steenge, A. E., & Boĉkarjova, M. (2007). Thinking about Imbalances in Post-catastrophe Economies: An Input–Output based Proposition. In Economic Systems Research (2 ed., Vol. 19, pp. 205-223). Routledge. doi:10.1080/09535310701330308
Stringer, J. (1998). Coatings in the electricity supply industry: past, present, and opportunities for the future. In Surface and Coatings Technology (pp. 1-9). Elsevier Science S.A.
Tebodin. (1998). Schade bij inundatie van buitendijkse industrie. Document nr 334393. Tebodin consultants & engineers.
Tebodin. (2000). Schadecurves industrie ten gevolge van overstroming. Document nr 334666. Tebodin consultants & engineers.
Vanderkimpen, P., Melger, E., & Peeters, P. (2009). Flood modeling for risk evaluation – a MIKE FLOOD vs. SOBEK 1D2D benchmark study. In Flood Risk Management: Research and Practice (pp. 77-84). London: Taylor & Francis Group.
VNK2. (2011). Flood risk in the Netherlands. Ministerie van Infrastructuur en Milieu, Interprovinciaal Overleg & Unie van Waterschappen.
Wagenaar, D. (2012). Master Thesis: The significance of flood duration for food damage assessment. Delft University of Technology & Deltares.
Whitehead, J. (2008). Flood Risk and Electricity Companies. In Journal of Pratical Ecology and Conservation (Vol. 7, pp. 173-177). Wildtrack Publishing.
