Assessment of flood risks in polders along the Dutch lakes F. den Heijer* & A.P. de LoofP WL\delft hydraulics ^Ministry of Transport, Public Works and Water Management, Directorate General of Public Works and Water Management (Rijkswaterstaat), Road and Hydraulic Engineering Division Abstract In 1996 the so-called Flood Protection Act came into force in the Netherlands. When discussing the Act in Parliament the question arose whether Lake Markermeer was a source of great flood hazard. The Ministry of Transport, Public Works and Water Management commissioned WLJDELFT HYDRAULICS to study this subject. The objective of the study is to submit arguments with which the Dutch Parliament is able to decide whether the Lake Markermeer is a source of great flood hazards, or not. Therefore a comparison has been made with the greatest lake in the Netherlands, Lake Usselmeer. The objective of this paper is to show an application of efficient and effective interactive use of computer simulations in risk analysis. The term 'risk profile' has been introduced as the total risk due to flooding of the polders around or along a water system. For the purpose of comparing the hazards due to flooding by different lakes the 'risk profile' is a good measurement tool, because of the integration of the most important aspects of flood hazards. 1 Introduction A great part of the Netherlands lays below sea level. As long as the Dutch people have lived in these low laying areas they have had to defend themselves against flooding. They knew flooding could cause
114 Risk Analysis great disasters, but they didn't know how to resist this threat sufficiently. Several times river floods, or storm surges flooded the low laying areas and caused great damage and claimed many lives. The last flood disaster occured in 1953. A heavy storm overtopped the sea defences in the southern part of the Netherlands. This resulted in about 2000 casualties and a damage of about 60 million Dutch guilders (nowadays several billions of Dutch guilders). After this disaster a special committee, the Delta Committee, was established to propose measures to defend the Netherlands against the sea, and to propose a strategy for flood protection. Since 1953 much knowledge has been generated about flood protection. The flood protection concept proposed by the Delta Committee is applied in a more and more enhanced way. However, the strengthening of the water defences is a process which shows slow progress. In 1993 and 1995 two river floods occured. The river Meuse flooded twice a large area. Many people had to be evacuated as a precautionary measure. A strong political force resulted in an acceleration of the reinforcement programme of the Dutch river dikes. The dikes along the lakes also were involved in this programme. In 1996 the so-called Flood Protection Act came into force in the Netherlands. In this Act the way of application of the flood protection concept is appointed. The safety standards of the primary water defences in the Netherlands have also been set in the Act. The primary water defences are defined as the water defence systems along the Dutch coast, the greater rivers, and Lake Ijsselmeer, which is the largest lake in the Netherlands. These water bodies were considered as sources of great hazard. In the Act the Lake Markermeer was not considered as a source of great hazard. The most important reason for this was the assumption that the mean lake level was better controlable compared with that of Lake Ijsselmeer. In contrast to Lake Ijsselmeer, Lake Markermeer is not directly influenced by the sea or the greater rivers. In Figure 1 a map of the lakes is shown The question arose whether Lake Markermeer was a source of great hazard as well. In 1995, while discussing the flood protection Act, the Dutch Parliament asked the Minister of Transport, Public Works and Water Management to analyse the hazard of the Lake Markermeer, related to the hazard of Lake Ijsselmeer. The Ministry of Transport, Public Works and Water Management commisioned WLJDELFT HYDRAULICS to study this subject. Parts of this study have been highlighted in this paper, especially the computer simulations and the risk analysis.
Risk Analysis 115 Figure 1 Map of the study area. 2 Objective and approach The objective of the study was to submit arguments with which the Dutch Parliament can decide whether or not Lake Markermeer is a source of great flood hazards, like Lake Usselmeer. The objective of this paper is to show an application of efficient and effective interactive use of computer simulations in risk analysis.
116 Risk Analysis The approach that has been followed, was to make a system analysis with which a comparison could be made between the effects and risks of flooding of dike ring area's (polders) bordering both Lake Usselmeer and Lake Markermeer. This system analysis has been carried out with the help of computer simulations. The comparison has been made based on statistical analysis of computer simulations of the system response due to the wind speed and wind direction, and the volume of water in the Lake at the time of the high wind speeds. The comparison consisted of The loads on the water defence systems, depending on several types of control of the lake water levels, which are simulated based on about 20 years of measurements of the lake water level. The amount of damage and casualties after failure of the dikes. The economic risk due to flooding of areas around the lake. In this paper the last subject is adressed. The term 'risk profile' has been introduced as the total risk due to flooding of the polders around or along a water system. The economic risk profile forms the foundation of the comparison of Lake Usselmeer and Lake Markermeer. 3 System analysis 3.1 The hydraulic regime The lake is schematised as a basin with two natural threats to the surrounding dikes: wind (speed and direction) and high mean lake levels (Figure 2). The wind causes surges up to 1 meter above the mean lake level, and waves up to 1.5m high. The precipitation, discharges of incoming rivers etc. cause high incoming volumes of water leading to high mean lake levels. Depending on the geometry of the lake, the water depth and the wind direction, the hydraulic circumstances are different around the lake.
Risk Analysis 117 wind water level mean lake level daily level Figure 2. System responses due to wind and high mean lake levels. 3.2 The water defence system The lake is surrounded by several polders. Each polder is seperated from the lake by a dike. To protect the polder from flooding the dikes have to withstand the hydraulic loads, water level and waves. Dike failure occurs due to several failure mechanisms: overtopping due to high water levels and waves, geotechnical instability (sliding, piping), instability of the revetments, etc. The length of dikes defending the whole polder can be up to 100 kilometres long. The dike can be subdivided into several individual dike sections. Each dike section has to withstand the hydraulic loads. The polder is flooding if at least one of the dike sections failes. 3.3 The inundation proces When the hydraulic loads cause failure anywhere along the dike, a breach is formed and the polder inundates. The water level in the inundated area depends mainly on the lake level at the moment of the breach, and the depth and compartmentation of the polder. The inundation can cause casualties and damage to buildings, agriculture, livestock, etc.
118 Risk Analysis 4 The risk analysis The risk of flooding is defined as the probability of flooding multiplied by the effects of flooding. Since we considered economic risk we took only the financial damage into account with respect to the risk analysis. To give insight in the risk due to a water system the term risk profile has been introduced as the total risk due to flooding of the polders around or along a water system: ) (1) in which: RP(W) : the risk profile of a water system W R(G,W) : the risk in polder G due to flooding by water system W In principle the risk analysis seems to be very simple. In practise investigations in risk analysis of water defence systems are complicated and time consuming. The calculation of the probability of flooding has serial and parallel components, correlations etc. The effects of many combinations of (in this case) wind speed, mean lake levels and dike breach scenario's have had to be simulated. To carry out an effective (accurate) and efficient (few simulations) risk analysis it is necessary to make some well chosen simplifications. The most important simplifications are described below: The probability of flooding of a polder has to be equal to or smaller than the safety standard, but varies in time because of deterioration of the dike system. In the study we assumed the probability of flooding equal to the safety standard. The inundation level in the polder depends on the mean lake level at the time when a failure mechanism occurs. In the great lakes in the Netherlands wave overtopping is the most important failure mechanism because of the severe wave attack. In principle many hydraulic circumstances are present which the dike cannot withstand. This is explained in Figure 3, where the curve gives all combinations of wind speed and mean lake level which lead to overtopping by waves. So many computer simulations seems to be needed to simulate the effects in the failure area in the whole range of mean lake levels. However, in this study the inundation level has been chosen which occurs due to flooding at circumstances in the design point. The
Risk Analysis 119 design point is that combination of wind speed and mean lake level with the highest probability density (Figure 4). Since the design point analysis results in mean lake levels a little higher than the daily level, a simple analysis has been made with the daily level as well. This made a comparison possible of the risk profiles of Lake Markermeer and Lake Usselmeer, with the risk profile of water systems from which no computer simulations of the hydraulic regime (see paragraph 3) were available (and so no design point), but from which no doubts are present with respect to the title 'source of great hazard'. (unacceptable) wave overtopping no (or acceptable) wave overtopping» mean lake level relative to MSL (m) Figure 3 The curve of all combinations of wind speed and mean lake level which lead to overtopping by waves. To carry out the risk analysis 3 classes of computer simulations were necessary: simulation of the water levels and waves near all dikes around the lakes simulation of the occurrence of high mean lake levels simulation of the inundation proces and corresponding damage These classes are described in the following paragraph.
120 Risk Analysis design point low probability high probability -0.4 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 mean lake level in the design point > mean lake level relative to MSL (m) Figure 4 Explanation of the design point 5 The computer simulations 5.1 Water levels and waves The hydraulic circumstances in the lakes are simulated with 2DH-water movement models. After schematisation (bottom profile, geometry, boundary conditions) and calibration, 216 runs were carried out with combinations of the natural conditions of wind speed (6), wind direction (12), and mean lake level (3). During an arbitrary storm situation the water levels and waves around the lake are known with the results of this simulations. These data were used to fill a database. This database formed the input to a model for dike safety calculations. 5.2 Simulation of high mean lake levels Lake Markermeer was separated from Lake Usselmeer in 1976, when the Houtribdike between the two lakes was constructed. To make a proper (frequency) analysis of the mean lake level of Lake Markermeer a longer
Risk Analysis 121 historical database was needed. Besides, to study alternatives of the control of Lake Markermeer a simulation model was needed because in these cases no data was available at all. With this objective a model was built to simulate the mean lake levels in time, based on a volume analysis. Incoming volumes were precipitation, river discharges and sluice discharges, outgoing volumes were evaporation and sluice discharges. Sluice discharges depend on the wind speeds and directions (via the local water levels). Wind data were available since 1951. Calibration of the model was possible with the available data of the last 20 years. In Figure 5 a comparison is made between the model and the measurements for 1993. months in 1993 Figure 5 The mean level of lake Markermeer in 1993. Measured versus calculated. In this way we were able to carry out the simulations from 1951 till 1996 This is twice as long as the existance of Lake Markermeer. Apart from that, simulations could be made of the mean lake level in different control situations, or other lake geometries. 5.3 Simulating the inundation and corresponding damage The flooding process is a strongly dynamic one. However, in this study we used a sort of 'bath tub' model for two reasons. At first, the polders in the considered area's consist of several small compartments ('bath tubs'), which reduces the dynamic character of the inundation. Secondly, the
122 Risk Analysis simulation of the dynamic flooding process only makes sense when you are able to simulate the response of people with respect to flooding. The damage is not much influenced by the dynamic character. The volume in the lake is finite. This implies that the inundation level at the end of the process depends on the volume of the lake and the volume of of the polder. The only problem are the dikes inside the polder which subdivide the polder into compartments. These dikes can possibly also withhold the water. The calculations were carried out controlling the height of the compartment dikes. Only when one compartmernt dike is overtopped, the following compartment can be inundated. The inundation depth in each part of the polder indicates the amount of casualties and damage, which also depends on the contents of the polder. To this extent a model is used based on effects of the 1953 disaster. 6 Results of the risk analysis The results, given in terms of risks of the total area exposed to flooding by a lake (the risk profile), were almost equal, comparing Lake Markermeer and Lake IJsselmeer. The economic risk of flooding are both about 4 million Dutch guilders per year. Set in a wider context we compared the risk of these lakes with risk profiles of other water systems in the Netherlands. The indicative results were shown in Figure 6.
Risk Analysis 123 Lake Lake Grevelingenmeer Markermeer Lake Ijsselmeer Eastern Scheldt North Sea Figure 6 The risk profiles of several water systems in the Netherlands 7 Conclusions For the purpose of comparing the hazards due to flooding by different lakes the 'risk profile' is a good measuring tool. The risk profile integrates the following aspects of flood hazards: the behavior of the lake due to storms or rainfall the behavior of the flood protection measures, such as dikes, the social and economic aspects in the inundated areas Along with the other arguments provided in the WLJDELFT HYDRAULICS study, the decision whether or not Lake Markermeer is a source of great flood hazards can be made based on the results of the risk profiles of Lake Markermeer and Lake Ijsselmeer. Furthermore, in the risk analysis it has been possible to use the computer simulations in an effective and efficient way.