8th Iiilenmtional Symposium on Lowland Technolog)' September 11-13, 2012 in Bali, Indonesia

ACT OF DESPAIR OR FULL-FLEDGED EXPERIMENT: RETROSPECTIVE RESEARCH ON THE 1945 WIEREVGERMEER FLOOD

O.A.C. Hoes R.W. Hut' and M. Boomgaard^

ABSTRACT: The present state-of-the-art in flood risk assessment focuses on breach models, flood propagation models, and economic modelling of flood damage. However, models need to be validated with real data to avoid erroneous conclusions. This reference data can either be historic data, or can be obtained fi'om controlled experiments. The inundation of the Wieringermeer polder in the in April 1945 is one of the few examples for which sufficient historical information is available. The objective ofthis article is to compare the flood simulation behaviour using flood data from 1945. The context, the breach gi'owth process and the flood propagation are explained. Key fmdings for current flood risk management addresses the importance of the drainage canal network during the inundation of a polder, and the uncertainty that follows from not knowing the breach gi'owth parameters. This case study shows that historical floods provide valuable data for the validation of models and reveals lessons that are applicable in current day flood risk management.

Keywords: Floods, Simulation, Polder, Breach Growth.

INTRODUCTION geological records to reconstruct surge heights of pre­ historic storms. Gottschalk (1977) used wiitten records Predicting how an embankment fails, how the and artwork to consfruct an overview of all floods from adjacent area fills up with fiood water and estimating the the 14th until 18th century in the Netherlands. More damage involved are essential parts of undertaking a recently. Baart et al. (2011) used 18th century paintings flood risk assessment. The importance of understanding and drawings to estimate storm surge levels. Lumbroso these three issues can be illustrated by the recent fioods & Vinet (2011) compared the February 1953 flood in in Pakistan in 2010 and 2011. The present state-of-the- England with coastal flooding in France in 2010. art focuses on breach models, flood propagation models, However, none of these reconstructed floods were and economic modelling of flood damage (Apel et al. sufficient to validate a flood model. 2004), (Manen and Brinldiuis 2004), (Apel et al. 2006), In our search for a proper dataset to validate a polder (Bouwer et al. 2009), (Apel et al. 2009). However, flood, we found one case for which the necessary models need to be validated with real data to avoid information is as complete as one would like. In 1945 erroneous conclusions and the implementation of the Wieringermeer polder area was flooded deliberately unsatisfactory measures. Such real data can either be by the German armed forces about three weeks before historical data, or can be obtained from controlled the end of World War II. The 118000 German occupying experiments. Full-scale experiments are difficult and troops in the Netherlands controlled the western expensive to accomplish. Experiments on a small scale provinces in 'Festung Holland', but were enclosed by have the problem that results camiot be extrapolated to Allied Forces. To prevent the Allied Forces to enter large areas. Accurate historical data for validation are Festung Holland through the fi-om the north, hard to get, even in the Netherlands with its long history the Germans blew up the embanlonent at two locations of floods. During large disasters - as the February 1953 that separated the Wieringermeer polder from Lake flood in England, Belgium and the Netherlands - IJssel. Actually, as the Allied Forces did not even everyone is busy saving his or her loved ones and attempt taking this route, one may quafify the belongings, instead of collecting data. Wieringermeer inundation as an act of despair instead of Sometimes, historical data can be obtained by a military-strategic necessity. Anyway, the water level at retrospective analysis. Jelgerma et al. (1995) used Lake IJssel was 0.12 meter above mean sea level (MSL),

' lALT member, Delfl University of Technology, PO box 5048, 2600 GA, Delft, THE NETHERLANDS 2 Delft University of Technology, PO box 5048, 2600 GA, Delft, THE NETHERLANDS ^ Hoogheemraadschap Hollands Noorderkwartier, PO box 250, 1700 AG, Heerhugowaard,THE NETHERLANDS

31 Hoes, el al. and the surface level adjacent to the location of the blast was 3.5 meter below MSL. At the start of the prospective 48 hours, the water seeped thi-ough the two grooves formed by the explosions, but these gi'ooves developed flu'ther and the polder raged full with a wall of water. Historic information fi-om this inundation is properly recorded and still available in different archives in the Netherlands, as the archive fi-om the Dutch Ministry of Infi-asti-ucture and the Environment, the Noord Hollands Ai-chive in Haarlem, Regional Ai-chive in Alkmaar, the Nieuwland Erfgoed Centre in Lelystad, and the NIOD Institute in Amsterdam. Available are the water levels from the adjacent Lake Ussel during World War II, the time of the blast, the size of the initial breach, the final extent of the breach, and information about times and locations where the water arrived and how fast the water level rose in the polder. These water levels were collected by those who refused to evacuate (e.g. Fig.1). By collecting and combining the information fi-om all Fig. 1 Office building in during the these sources to a full-fledged dataset, we could inundation. Ai-ound 30 people sheltered 3 weeks in this building until the capitulation on the 5th of May 1945 accurately reconstruct the Wieringerrneer flood. Such reconstructions are valuable for current flood enlarge the agi-icultural land in the Netherlands. It risk management. Especially in the Netherlands, it is measures 13 km fi-om west to east and 16 lan fi-om south inevitable for water authorities to be prepared for floods, to north (20.000 hectares). The surface level in the because of our low elevation compared to mean sea level. Wieringermeer polder descends fi-om 1 meter below All authorities practice how they will act in case of a MSL in the West towards 5 meters below MSL in the calamity and formulate evacuation plans with the East. The interface between Pleistocene sand and intention that these will lunit the damage and victims in Holocene clay slopes fi-om 4 meter below MSL in the case of a real flood. The starting point is that during this North to 13 meter below MSL in the South. The preparation all necessary knowledge is available thi-ough combination of these two slopes results in a thin clay flood simulations with detailed elevation data. However, layer in the Northeast, causing brackish seepage in the these models are run with default parameters based on canals in this area. theoretical exercises, and the results camiot be validated Two large pumping stations were built to discharge with data fi-om a real situation. The 1945 Wieringermeer water: pumping station Leemans (52°55'16"N, flood is the exception to the rule. 5°02'18"E) with two diesel engine driven pumps of 250 This paper presents the results of a retrospective m3/min at a head of 5.5 m and pumping station Lely analysis of this flood. We have collected available (52°46'35"N, 5°06'12"E) with three electric engine historical material, and built a combined 1D-2D flood driven pumps of 400 m3/min at a head of 6 m. propagation model that accurately reproduces the flood. The construction of the Wieringermeer was totally First we provide a description on the history of this event. organized by the national government, who created an In the second part we discuss the breach growth process ultramodern polder with - for those days - large parcels and inundation of the Wieringermeer polder, hi the last of 20 hectares and an extensive network of navigable part we discuss the reconstruction works and canals and sluices to transport agricultural products by reconstruction costs. In each part we will try to link the ship. Three villages were build: Wieringerwerf, flood in the Wieringermeer polder to the present flood Middenmeer and Slootdorp, of which the village modelling approach used in flood risk analysis. Wieringerwerf (located in the centre of the polder) was planned to become the main village. In retrospect, the three villages were built too close to each other and too WIERINGERMEERPOLDER far away fi-om the so called old land. Next, Middenmeer was build before Wieringerwerf and was also closer to the old land then Wieringerwerf As a resufi The Wieringermeer polder was reclaimed fi-om the Middenmeer developed naturally as a main village next sea between 1927 and 1930. This polder was the first of four large polders reclaimed between 1927 and 1980 to

32 Act of despair or full-fledged experiment : Retrospective research on the 1945 Wieringermeer Flood

and September the German army inundated around 120 000 hectares by blocking the discharge of excess water (See also Fig. 2). A telegram fi-om the resistance on the 16th of February 1944 for the Dutch government in exile in England stated that the German army wanted to raise the water level of Lake IJssel (Waalewijn 1985). A high water level was required to have a necessary buffer of water for inundations. Starting fi-om the 27th of February the maintained water level became -0.13 m MSL. The pre-war target level was -0.2 m MSL in the summer and -0.4 m MSL in the winter. Starting from October the water level in Lake IJssel was further raised tin it reached +0.36 m MSL on the 18th of November 1944 (Fig. 3). At the end of February 1945 the activities at the Wieringermeer dike commenced (N 52°53'26", E 5°04'32"). The plan was to dig ten holes of 7 m deep and 1.5 m in diameter at two locations; four holes in the inner slope, two at the crest, and four in the outer slope. Each hole contained a masonry lining to increase the impact of the blast. At the 14th of April, all 20 holes were dug, but the brickwork was not finished yet. Nevertheless, on the 15th and 16th of April, 1000 kg of Fig. 2 Ai-ound 220 000 liectares were inundated in May explosives was lowered in each hole. All the explosives 1945 (Jong 1982) were connected by electrical wire and connected to one to Wieringerwerf (Cock and Willemse 1955), (Terpstra master clock (Romburg 1987). and Vogt 1980). The first young pioneers who colonized the polder came fi'om all over the Netherlands and were aware that TUESDAY THE 17TH OF APRIL 1945 they had to build a new community on an unique piece of land. This can be illustrated by the existence of more At four o'clock in the morning the Mayor of the than 30 different societies and clubs within three months Wieringermeer was informed by the Germans that they after the first settlers arrived, providing the necessary would blow up the dike around noon. There was an contacts for a respectable social life (Kamp 1937). The evacuation plan for such situations, but that did not core binding factor was however the absence of a worked out at all as foreseen. In the plan the polder was municipality board. In 1930 the new polder area was divided into five sections with each a commanding divided in five sections and added to the adjacent officer. These five officers were assisted by 68 gi-oup existing municipalities on surrounding the old land. leaders. Every resident belonged to a group of around However, the contradictions between the old and new land were so large, that it brought the Wieringermeer 50 40 . Water leve Lake IJssel (c m residents unintentionally closer to each other. This has 30 - ^SLl changed in 1938 by the founding of the public institution of the Wieringermeer. On 1 July in 1941 this became the 10 - municipality of the Wieringermeer. r -30 -N PREPARATORY ACTIVITIES I- >:t L •) ir r ^ 1 c- ) a X ; J i _ The Inundation of the Wieringermeer in 1945 did not % Jov-194 5 - lan-194 6 - occur unexpectedly, but why the Germans fmally Fig. 3 Water level at lake IJssel during 1944 and 1945. decided to go along is unknown. The first reports fi'om The usually maintained water levels are -40 cm MSL in winter and -20 cm MSL in sunnner) inundated polder areas as a German war strategy in the Netherlands date fi'om February 1944. Between February

33 Hoes, et al.

100 people and a fixed meeting point was assigned to BREACH GROWTH every group (Bijdorp 2007). AU people were asked to show up at the assigned Research on breach gi-owth is incidentally conducted meeting point at 10:00 AM where they would be in the last decades, especially by Japanese, American, informed about their group's destination. The and Dutch researchers. The research in the US focuses destinations were not known in advance to avoid that on so called fiise plugs. In the US situation sand dams people would switch to groups heading towards more are used in spillways of reservoirs to prevent any popular destinations. In reality no one showed up at his uncontrolled overflow. In the Netherlands research or her meeting point but left the area individually. The started around 1990 with breach experiments of sand farmers had the advantage of their own horse carts, and dikes (TAW 1994). Later, this research extended took some of their belongings with them. The villagers towards cohesive soils, like clay dikes (Zhu 2005). In in the middle of the polder without a cart, had to flee general dike breach development can be divided in 5 almost 12 kilometres on foot to reach higher ground. phases (Costa 1985), (Visser 1998), (Fujisawa 2009), Only a limited nuinber of lone wolves stayed at home to (Chang 2010), (Vorogushyn 2011). Phase wise details prevent plundering. are provided hereby: The blast was at 12:15 PM and was heard up to 25 km away! Farmers who wanted to approach the location Phase 1 gully erosion at the iimer slope were stopped by the German army, but fi-om a distance they saw that there was some water seeping through one With just a small amount of water seeping through, shallow groove, that the other groove failed to cause any the flow velocity is rather slow through the groove fiow. The German army would have shovelled both (somewhere between 0 and 2 m/s), but increases over the grooves wider and deeper to fiirther initiate the water inner slope of the dike. At this stage the inner slope flow (Pladdet-Dees 1946). erodes causing the slope of the gully to become steeper The villagers who stayed at home during the time until a maximum slope. See fig. 4.1 to fig 4.2. that the polder was flooded, left the area several days later by boat after the waves caused by several spring Phase 2 head cut migi-ation storms started to destruct the first houses. None of the 7000 official residents and 2000 persons in hiding In this phase the gully develops backward - in our drowned during the flood, and also the majority of the case in the direction of Lake IJssel. After a while the cattle survived. During the weeks that followed only the gully reached the outer slope of the dike and nothing was corpses of 1 horse, 26 cows and 2 pigs were collected left fi-om the original crest. Up until this moment, the fiom the water (Directie van de Wieringermeer, 1955). groove was wider at the Wieringermeer side of the dike

34 Act of despair or full-fledged experiment: Retrospective research on the 1945 Wieringermeer Flood

compared to the Lake IJssel side of the dike, hi this process, that fiood simulations must be judged with a phase the first scratches of what would become two deep certain distrust. scour holes appeared, as the water had to change its direction after it hit the polder floor. This corresponds with fig 4.2 via fig. 4.3 to fig. 4.4. INUNDATION PATTERN

Phase 3 vertical deepening of the breach. Although the Netherlands are thought of being very flat, there are height differences in all polders. As The last remaining portion of the original dike was a mentioned, the Wieringermeer polder surface slopes triangular slice at the outer slope (lakeside). When this from -1 m MSL in the west to -5 m MSL in the east over slice started to decrease, the flow velocity increased. At a distance of about 12 km. This sloped surface has the end, the old dike disappeared and fi-om now on the braided through a pattern of canals and ditches, causing breach became wider at the lake-side compared to the flood water flowing differendy in each direction over the polder side of the breach. The flow velocity depends on surface, although water tends to flow to the lowest point the difference in water level over the breach. In this case, faster and spreads fi-om there on fijrther upward. The it was H = 3.5 in. Therefore, the velocity v increased (v lowest points, however, are the ditches and canals. As a = 1.7'H3/2) up to 11 m/s. See also fig. 4.4 to fig. 4.5. result, these canals fill up in the first hours, with the water being redistributed as shallow water waves (v = Phase 4 widening of the breach. sqi-t(g'd)) thi-oughout the polder rather quickly as compared to sheet water flow over the surface. As soon The breach width started to grow seriously as the as the canals are filled, water will rise out of them at the increased velocity eroded submerged sidewalls of the lowest fields. As a result, in the Wieringermeer polder breach. That caused the sand of the sidewalls to slide fiood water blocked the evacuation routes fi-om the tln-ee into the raging water and being fiushed out of the breach. In the prior phases, the side waUs did also slide into the water, but at that time there was more time needed to transport/flush the sand away as the velocity was lower. This corresponds with fig 4.5 to fig. 4.6.

Phase 5 Transition super to sub critical flow.

After a while the water level in the Wieringermeer polder started to rise seriously decreasing the speed with which the water flowed into the polder. That reduced the turbulence and the growth of the scour holes. After the flow of water became subcritical, the breach was only growing slowly. This growth stopped when the shear stresses were low enough for the material of the dike side walls and bottom surface to stay in place. See 4.6 and 4.7. The final result was 692 million m^ of water in the Wieringermeer and two large scour holes, which are still present today. The two scour holes are different in size, though the starting point for both breaches was almost the same: 10 wells with 1000 kg of explosives that exploded at the same time. At present, the southern scour hole is 31 meters deep and 5.4 hectares in area. The northern scour hole is 23 meters deep and 3.7 hectares in area. Apparently, small differences in the beginning in combination with the nonlinear erosion processes are sufficient to result in a totally different end situation. It is because of this unpredictability of the breach growth Fig. 5 Simulated inundation pattern at different times after the blast. The filling took around 48 hours due to the large size of the polder

35 Hoes, et al.

0.00 - v/ater level in -0.50 - wienr gerwen Willemse 1955). It took until early in the morning on (met rMSL) -1.00 - Wednesday before the water entered the streets of the village of Middenmeer. At 5:30 AM, the water entered the Flevo-road (Keppel et al. 1995). Around 11:00 AM the water level in Middenmeer was around half a meter

-3.00 - - simulation results (Bijdorp 2007).

A measured by thie sacristan In our case, this information could be combined with

-4.00 - 3 C 3 C> c the final dimensions of the breach (which is known); the 13 C3 C ) c ö c ) l only effective calibration parameter left was the time I- ^ t \J c )4/4 5 12:0 0 - )4/4518:0 0 - )4/4 5 12:0 0 - /0<1/4 5 0:0 0 - /04/4 5 6:0 0 - 34/4518:0 0 - needed for the breach to reach hs final depth. The transition fi-om fig 4.a to fig 4.e must have taken place Fig. 6 Simulated water level rise in Wieringerwerf and after 12 hours in order to have a water level rise that fit historical measurements by the sacristan in the vicarage the data fi-om the sacristan in Wieringerwerf (Figure 6). villages towards the higher gi'ounds in the southeast This duration of 12 hours corresponds with the extensive (Bijdorp 2007). report fi'om the Ministery of Public Works after the The resulting inundation pattem fi-om Figure 5 was February 1953 flood (Rijkswaterstaat 1961) that suggests simulated with Sobek 1D-2D software (Delft Hydraulics, that the width of a dike, and the presence of a boulder 2000). The basis for our model was a 25m resolution clay dam in the dike with a stone revetment can seriously digital elevation map (DEM). This DEM was collected slow down the breach gi-owth process. For modern flood in 1997 with a laser altimeter mounted on a helicopter. simulations water authorities in the Netherlands assume Because of the course pixel resolution not all canals and smaller default periods for breach gi-owth. In 2010 the ditches are featured in the DEM. They were added to the provinces in the Netherlands combined all their 3000 schematisation as ID line elements with their flood simulations in one database. 88% of those were run appropriate profile and elevation and then were with a 'time to reach maximum depth' of less than 1 connected to the DEM. hour and none of the simulations used more than 7 hours. The model was calibrated with historical data of times and locations of water arrival and water level rise in the polder. Among the data used are the recorded data RECONSTRUCTION WORKS of the sacristan fi-om the village of Wieringerwerf, who refused to evacuate. The sacristan saw his yard getting Immediately after the liberation on the 5th of May inundated around 00:30 AM on April 18th, 1945. This 1945, the Public Works department worked on a plan to was 12 hours and 15 minutes after the blast. At 3:00 AM close the dike. They decided to build a 940 m long new dike in Lake IJssel around the deep scour holes. This solution was preferred over filling the scour holes, as this would have required much more sand and also the subsidence would be uncertain. The location and shape of the scour holes made that a new dike in the polder itself would have required more sand and boulder clay.

The first sand was dumped on the 21th of June 1945; on the 5th of August the dike was closed. On the 9th of August pumping could start to discharge around 616 million m3 of water. This is somewhat less than the before mentioned 692 million m3 that inundated the polder, as the water level of Lake Fig. 7 Reconstruction works around the breach in IJssel had already been lowered to -0.4 m MSL (Fig. 8) August 1945.The polder is on the left with some roofs to minimize the water vohmie in the Wieringermeer. In of farm houses visible in this picture and Lake IJssel is the subsequent months the 2 pumping stations Leemans on the right and Lely discharged 469 million m3, but not without problems. The suction side of the southern pumping the water entered his house, and at 4:00 AM, he station Lely was fi-equently blocked by fioating debris. measured 21 cm. At 5:00 AM there was 40 cm and at 8:00 AM aheady 57 cm. At noon it was 85 cm and the next evening at midnight he measured 177 cm (Cock and

36 Act of despair or full fledged experiineiil : Retrospective research on Ihe 1945 Wieringermeer Flood

buildings could not resist the beating of the waves during

Lake IJssel ( cm the storms in the spring and sunmter of 1945. Without MSL) these storms the damage would have been much less. For today's simulations this is an interesting aspect, as wave impact is never taken into account in a priori damage ZZ calculations. Factors mainly included are inundation depth, and sometimes flow velocity or duration (Merz 2010). 1 L T LJ1 L r r t ^ ? r r ^ 11 C H ° The reconstruction of the polder costed $ 10 million u 2 3. C i. I 1 r < c I :I I • : (see table 1) and took 5 years. Nevertheless, the farmers

. Apri l 194 5 - C < S O started to cultivate their land direefly in the spring of Fig. 8 Water level at lake IJssel April - June 1945. On 1946 with less intensive crops. Doing this, they were April the 17th the water level was + 12 cm MSL. The able to harvest a nearly complete yield in 1946 (Dienst water level dropped 15cm between the 17th and the Wederopbouw Wieringermeer 1945). 20th This was solved by deploying German torpedo-nets in fl-ont of the suction side of the pumping station. DISCUSION AND CONCLUSION Next, 55 million m3 water was discharged under free flow through a sluice towards a nearby polder. 101 Interest in preventing floods has increased in recent million m3 was discharged with 11 emergency pumps. years due to extraordinary floods, e.g. the 2004 Tsunami, At the 11th of December - 125 days after the first pump Hurricane Katrina in 2005, the floods in Pakistan in 2010 started - the water in the polder was back at -5 m MSL. and 2011, Japan's Tsunami in 2011, and the necessity to With the flood water gone and the surface of the prepare for the impacts of climate change. Research polder fmally dry, the damage became visible. The institutes, water authorities, and consultancy firms in Wieringermeer showed an unpleasant view of ruins of many countries conduct flood studies, improve what had been churches, farms, and houses. Without emergency plans and develop measures to arrange a exception all buildings were at least damaged: in more robust environment. Within this process the focus Wieringerwerf 155 houses out of 165 were destroyed, is on possible fitture scenarios, but every now and then it and in Slootdorp 120 houses out of 150 and in is good to look back to past events to leai-n lessons. Middeimieer 176 houses out of 221 were destroyed. Several aspects fi-om the inundation of the Furthermore, fi-om the 513 farmhouses 90% was Wieringermeer are now - 65 years later - still relevant. destroyed. This devastation was mainly because the A polder inundates by a combination of A) water that is rapidly redistributed through the existing drainage Table 1 Reconstruction cost (Terpstra 1980) network of canals and ditches and B) water that flows Item Damage radially, but slower over the fields starting fi-om the Rebuilding farm houses $ 4 775 000.- breach location. The water redistributed by the network Communal cost, labour quarters $ 2 590 000.- might iiumdate the lowest fields at quite some distance etc from the breach before overland flow can reach the same Restoration of canals and drains $ 675 000.- places. Furthermore, this water might unexpectedly Removing debris $ 555 000.- block the evacuation routes. Rebuilding dikes $ 415 000.- Only near the breach the flow velocities are high Recovery of electricity, water $ 400 000.- enough to initiate structural damage to buildings. The services actual damage to building structures is likely to occur Rebuilding schools $ 230 000.- because of the wave impact of storms during the period Roads, bridges and office $ 185 000.- that a polder is inundated. This damage is never buildings explicitly considered in the flood damage estimation Planting trees $ 165 000.- models used at present. Emptying the polder $ 100 000.- Two aspects determine the velocity with which a TOTAL $ 10 000 000.- polder floods. First of all the upstream water level and *The actual damage was higher as lost household water volume. In case of Lake IJssel the total surface of contents, farm equipment, cattle, the harvest of 1945 2770 km2 in 1945 is nowadays reduced by the and cost to temporarily accommodate the residents is construction of the Flevopolder and the Houtribdijk in not part of the reconstruction cost. the sixties and seventies to 1100 km2. The same disaster

37 Hoes, et al. nowadays will result in a smaller inundation depth. If Costa, I.E. (1985). Floods from dam failures, US one considers to compartmentalise a polder, one should Geological. Survey rep. m- 85-560, Denver. also consider to compartmentalise the upstream water Cock, E. de, Willemse W.J.Z. (1955). De inundade van body. The second aspect is breach growth, which de Wie-ringermeer in 1945, een rapport over de determines the water level rise. In our simulation of the evacuatie en de terugkeer van de Wieringermeer flood the final breach size was used as Wieringermeerbevolking, Alphen a/d Rijn: N. input. Next, the breach growth process is selected such Samson. that the water level rise corresponds with historical Dienst Wederopbouw Wieringermeer. (1946). Verslag measurements. However, for many simulations the final van de werkzaamheden over de periode 1940-1945, measurements are not known in advance, and are the Den Haag. choice of the modeller. The predictive value fi'om these Directie van de Wieringermeer. (1955). Wording en simulations must be cautiously interpreted. Smaller or opbouw van de Wieringermeer, geschiedenis van de larger breaches are just as realistic, because of the non- ontgiiming en kolonisatie van de eerste linearities of the erosion processes. IJsselmeerpolder. Wageningen: Veenman The present state-of-the-art in flood risk management Fujisawa, K., Kobayashi A., Aoyama S. (2009). is applying models which need to be validated with real Theoretical de-scription of embankment erosion data to avoid erroneous conclusions. This case study owing to overflow, Geo-technique 59(8): 661-671. shows that historical floods provide valuable data for the Gottschalk, M.K.E. (1977). Stormvloeden en validation of models and reveals lessons that are rivieroverstromin-gen in Nederland. Assen: van applicable in current day flood risk management. Gorcum. Jelgersma, S., Sfive M., van der Valk L. (1995). Holocene storm surge signatures in the coastal dunes REFERENCES of the Westem Nether-lands, Marine Geology: 125 (1-2): 95-110. Apel, H., Thicken A.H., Merz B., Blöschl G. (2004). Jong, L. de. (1982). Koninlaijk der Nederlanden in de Flood risk assessment and associated uncertainty. Tweede Wereldoorlog, deel Xb: 1441. Leiden: Natural Hazards and Earth System Sciences 4: 295¬ Martinus Nijhoff. 308. Jonkman S.N., Bockarjova M., Kok M., Bernardine P. Apel, H., Thicken A.H., Merz B., Blöschl G. (2006). A (2008). Integrated hydrodynamic and economic probabil-istic modelling system for assessing flood modelling of flood damage in the Netherlands, risks. Natural Hazards 38: 79-100. Ecological Economics 66: 77-90. Apel, H., Aronica G.T., Kreibich H., Thicken A.H. Kamp, A.F. (1937). -land, verleden en (2009). Flood risk analyses - How detailed do we toekomst va de Zuiderzee. Amsterdam: Querido. need to be? Natural Hazards 49: 79-98. Keppel C, Brugman N., Maris-Eriks W. (1995). 17 april Baart, F., Bakker M.A.J., van Dongeren A., den Heijer 1945: Het koolzaad bloeide zo mooi goudgeel. C, van Heteren S., Smit M.W.J., van Koningsveld Middenmeer: His-torisch Genootschap M., Pool A. (2011). Using 18th century storm-surge Wieringermeer. data fi-om the Dutch coast to improve the confidence Ki-eibich, H., Piroth K, Seifert L, Maiwald H., Kunert U., in flood-risk estimates. Natural Hazards and Earth Scwarz, J., Merz B., Thicken A.H. (2009). Is flow System Sciences 11:2791-2801. velocity a significant parameter in flood damage Bijdorp, B.A. (2007). 1940-1945, Wieringermeer in de modelling? Nat. Ha-zards Earth Syst. Sci. 9: 1679¬ bezet-tingsjaren, Middenmeer: Historisch 1692. Genootschap Wierin-germeer. Lumbroso D.M., Vinet F. (2011). A comparison of the Bouwer, L.M., Bubeck P., Wagtendonk A.J., Aei1s causes, effects and aftermaths of the coastal fiooding J.C.J.H.. (2009). huindation scenarios for flood of England in 1953 and France in 2010, Natural damage evaluation in polder areas. Natural Hazards Hazards and Earth System Sciences II: 2321-2333. and Earth System Sciences 9: 1995-2007. Manen S.E. van, Brinldiuis M. (2005). Quandtative flood Chang, D.S., Zhang L.M. (2010). Simulation of the risk assessment for polders. Reliability Engineering erosion process of landslide dams due to overtopping and System Safety 90: 229-237. considering variations in soil erodibility along depth. Merz B., Kreibich H., Schwarze R., Thicken A. (2010). Natural Hazards and Earth System Sciences 10: 933¬ Review article "Assessment of economic flood 946. damage", Nat. Ha-zards Earth Syst. Sci. 10: 1697¬ 1724.

38 Act of despair or fiiU fledged experiment : Retrospective research on the 1945 Wieringermeer Flood

Pladdet-Dees, P. (1946). De Terp als reddingsboei. Visser P.J. (1998). Breach gi'owth in sand dikes Ph.D. Rijkswaterstaat. (1961). Verslag over de stomivloed thesis. Delft: Delft University of Technology. 1953. 's-Gravenhage: Staatsdrukkerij. Vorogushyn S., Apel H., Merz B. (2011). The impact of Roniburgh, C.P.P. (1987). De inundatie van de the un-certainty of dike breach development time on Wieringermeer-polder in april 1945. Amsterdam: flood hazard. Physics and Chemistry of the Earth 36: Stichting voor het bevol-kingsonderzoek in de 319-323. drooggelegde Zuiderzeepolders. Waalewijn, A. (1985). Waterstaat in Londen 1940-1945. TAW. (1994). Het Zwin, successen en lessen Den Haag: Ministerie van Verkeer en Waterstaat. bresgroeiexperimenten. Delft: Technische WL-Delft Hydraulics. 2000. Delft 1D-2D, Manual Adviescommissie voor de Wa-terkeringen. Sobek over-land flow module. Delft: WL-Delft Terpstra P., Vogt P. (1980). 50 jaar Wiermgermeer. Hydraulics. Leeuwar-den: van Seijen. Zhu, Y. (2006). Breach gi'owth in clay dikes. Ph.D. thesis. Delft: Delft University of Technology.

39