Mass evacuation because of possible flooding is a worldwide phenomenon. Several deltas in the world face the same problems with evacuation management as a layer of flood risk management. Worldwide events illustrate the possible results as well as the problems of evacuation. For example, during hurricanes and flooding, people and movable goods might be saved through evacuation. Evacuation of buildings in the event (or based on the threat) of fire, terrorist attacks, nuclear or chemical accidents, volcanic eruptions, bush fires or earthquakes can reduce the loss of life by removing people from danger. A few examples with regard to evacuation and flood risk management are given for the following events (for all of events outside Dutch Deltas the frequency of the events is far more than the frequency for evacuation and flooding in the Netherlands):
The Netherlands, evacuation of the area of Rivierenland (1995) and Zeeland (1953); United States, Louisiana Delta (2005);
France; Xynthia (2010); Japan; Tsunami (2012);
United States, New York and hurricane Irene (2011); Australia, Brisbane flooding (2010 and 2011).
The examples in this chapter shows that evacuation in the Dutch delta is (far) more complex because of uncertainties, the size of the event, the impact of a flood and the number of people. Therefore the Dutch delta can be seen as a reference delta because complexity is less in other deltas. The Dutch Delta has a low frequency for flooding because of dikes, dunes and structures. This is to protect large areas that represent a high economic value. Two-thirds of the total number of people lives in a flood-prone area, and 70% of the Gross Domestic Product is earned in flood-prone areas. Because of the prevention levels that are already in place and calls to increase them even more (Deltacommissie 2008), it can be questioned whether investment in emergency management with regard to flood risk should be made. Risk assessments show that flooding is considered to be the disaster with the highest consequences in the Netherlands (BZK 2009). Earlier research showed that the risk of flooding is relatively high compared to other possible incidents (ten Brinke et al. 2008a; RIVM 2004). The low frequency of flooding also means that there is almost no real (or out-dated) experience with evacuation, but the risk persists. Therefore, evacuation is a low-frequency event for the Netherlands. Experiences with mass evacuation and flooding are limited. After the last mass evacuation (in 1995 see (van Duin et al. 1995; Meurs 1996)) and mayor flood (1953, see for example (Slager 2003; Lumbroso and Vinet 2011; Vinet et al. 2012), the approach to emergency management and prevention changed. Attention for emergency planning rised and a delta program to improve levees were executed. However the public perception of the flood risk is very low (Terpstra 2009; NIPO 2006), as is the level of attention to this issue from decision makers and policy makers which also tend to decline over time (ten Brinke et al. 2008a). An evacuation in the Dutch Delta is a complex operation because of the size of the treat, the available lead time, uncertainties, and the number of people involved related to the road capacity (Barendregt et al. 2005; Jonkman 2007; Kolen and Helsloot 2012b). It is therefore
known that for some areas a preventive evacuation cannot always by completely executed, although the focus remains on preventive evacuation. Large-scale flooding can occur from rivers as a result of extreme discharges as well as from the sea due to storm surges. Forecasts of hydraulic loads for extreme water levels in rivers (Rhine and Meuse) have a relatively long lead time and a small bandwidth of uncertainty. In addition, signals will be available for floods in smaller upstream catchments in Germany and Belgium. However, in the case of a storm surge, the lead time is relatively short and uncertainties are high. The areas of South and North Holland are hardly able to evacuate in time. Other coastal areas, such as the province of Zeeland, can evacuate in less time because the load on the infrastructure is less.
The Dutch delta however can also be seen as a non-representative delta because of the preventive measures taken. These (manmade) measures reduced the probability of flooding and therefore the risk by far. As a consequence the economy and culture (as the governance style “polderen” which is based on cooperation and negotiation) could develop. In the Dutch Delta the flood risk is reduced that far so the risk is not perceived any more by citizens, in other words other risk are perceived to be more relevant. The limited risk perception however is also a complexity for emergency planning and one of the elements of adaptive evacuation management as one of the research objectives of this thesis.
The Netherlands and the evacuation of the area of Rivierenland in 1995 and Zeeland in 1953.
Because of the high water levels of the rivers, the area of Rivierenland was evacuated in 1995 (250,000 people); however, the area did not
flood (van Duin et al. 1995). The event had an estimated return period of 50 years (WL 1995; Parmet 1996). Afterwards, discussions arose about the need for evacuation. The call for evacuation was placed in the perspective of water boards who claimed a larger budget to invest in flood prevention (Meurs 1996). During the event, the “dijkgraven” as chairman of the Waterboards indicated that they could not guarantee absolute safety. This information was used by mayors and decision makers on the provincial and national level to call for evacuation.
The coastal flood of 1953 (southwest of the Netherlands, 1836 fatalities) showed limitations of a better warning system as well and the perception of the risk. During 1953, the warning was delayed due to procedures, and limitations of media as the broadcasts on radio stopped after midnight. In additions the decision makers (as mayors) did not understand the warning and set other priorities (Slager 2003). More lead time and more early response to warning by authorities could have led to more available time for citizens to save themselves. Because the lack of a warning and ignorance of the warning by decision makers many people were surprised by the event. The flood of 1953 (with an estimated return period of 300 years) resulted in a delta plan for flood protection and a greater focus on flood prevention based on the philosophy that this might never occur again.
Figure 7: Flooded highway during a period of extreme water levels in 1995 and the evacuation of the area of Rivierenland
United States, Louisiana Delta (2005 and 2008).
Evacuations elsewhere in the world demonstrate the impact of strategic choices and decision ma– king on evacuation. After the flood in New Orleans caused by hurricane Katrina in 2005, which caused more than 1100 fatalities in Louisiana (Jonkman et al. 2009). The levees were designed for an event with a return period of 30 years (Sills et al. 2008). Some literature suggested that earlier and more robust involvement at the national level (such as FEMA and the Red Cross) might have reduced the consequences (Parker et al. 2009). Despite more people evacuated preventive than was expected, based on a survey among citizens it was expected that about 70% of the people would evacuate (Van Heerden and Streva 2004).
In the same area that was evacuated during Hurricane Gustav, the lessons learned were put into practice during the evacuation in 2008 (Cole 2008). However, the area did not flood but Hurricane Gustav itself caused damage inside the flood-prone area as well as outside of this area, where people were sheltered. Services in the shelters in the Baton Rouge area were down for a longer period than for those who did not evacuate and remained in the city of New Orleans (Boin 2009). Before the storm made landfall, it was described as the ‘mother of all storms’ by leaders. This phrasing, or symbol, was used to make people aware and willing to respond, but it can be questioned what kind of frame is needed the next time to make people evacuate.
France, Xynthia (2010)
REFERENCE
This section is based on ‘The impacts of Storm Xynthia Feb. 27-28, 2010 in France: lessons for flood risk management’ as published the journal of flood risk management (Kolen et al. 2012c)
On Sunday, 28 February 2010, around 2.00 am, the storm Xynthia (official French name ‘Tempête Xynthia, Feb. 27-28, 2010’) reached the west coast of France and caused large-scaled flooding of coastal areas. The storm claimed dozens of casualties and caused major damage along the Iberian Peninsula in France, Germany and the Benelux countries. A total of 65 people were killed. In most of the in total, 47 of the casualties were caused by the flooding along the coasts of the Vendée and Charente-Maritime in the western region of France. The return period was about 100 years (Kolen et al. 2012c).
Loss of life in the flooded area was caused by multiple reasons. Some people were trapped in their houses because electric lockers on doors and windows could not be opened any more when the electricity went down and flood water entered the houses. The remaining casualties
occurred as a result of the storm winds, which caused hazards such as falling trees (the wind also caused problems for the emergency services as well). A few days before the storm, the meteorological institute issued a warning for strong winds but also predicted high waves at the coast. Combinations of various factors (depression, wind, tide) were predicted to lead to a temporary rise of the sea level, causing parts of the coast to flood. Although Météo-France had reported the risk of rising water levels, they could not forecast exactly how high the water would rise. In France, flood warnings are given based on forecasts made by regional hydrological/meteorological centres (Ministere de L'Ecologie 2010). Currently, the subsequent conversion of weather forecasts by Meteo France into estimates of local water levels and the subsequent (in the case of exceeding alarm criteria) warnings from the mayors of the municipalities is explicitly a responsibility of the prefectures. Therefore, each prefecture has its own hydrologist or system. This requires that these hydrologists should identify the possible flood risk (using the information of forecasters of Météo-France) based on the forecasts of the storm and tide and inform others.
During the eve before the dike breaches, the expected wind speed was less than the wind speed during the 1999 storm (when no flooding occurred). Therefore, the threat of coastal flooding was not widely recognised. The combination of the wind speed and the high tide was not taken into account. Prefectures and local authorities claim that they were not focused on the rising water levels and the need to warn of flood risks because this information was "melted" into the usual list of hazards and subsequent storm recommendations. The use of technical language in the flood warnings, e.g., the word ‘setup’, or in French, ‘sur-cote’, as the additional rise of the water level caused by the influence of the wind on the tide, was not understood (Anziani 2010). As a result, no large-scale evacuations were recommended or carried out (Figaro 2010).
After the disaster, an emotionally and politically charged debate emerged on the causes of this disaster. This debate focused on a controversial subject: over the last twenty years, holiday homes were built at many locations along the French coast. It was well-known that these
Figure 9: The areas affected by the floods in Charente-Maritime (Charente-Maritime 2010)
coastal areas were (and are) prone to flooding. Most dikes are (implicitly) designed for a hydraulic load level that occurs once every 100 years. This debate also focused on difficulties in providing an adequate flood warning to the population at risk and related to when and how to decide to evacuate. Commentaries from the involved parties reveal powerlessness in decision making. L'Express, for example, asked why there was no evacuation after the warning by Météo-France. The prefects say that it was not that simple. ‘I signed a code red on Saturday, the 27th of February, at 16.00 pm, which was sent to the officials, along with a press release to inform the general public’, says Beatrice Lagarde, the sub prefect of the Vendée, she also said ‘There were no warnings about floods or failing flood defences. We cannot fantasise about risks and dangers ourselves. And what were we to do at the time that the risk spread over the entire territory of the Vendée – 600,000 persons? Where could we have gone at 22.00 pm to evacuate the 400,000 occupants who were threatened? To the Sahel?’ (L’Express 2010). Japan, the Tsunami after The Great Eastern Japan Earthquake (2011)
The Great Eastern Japan Earthquake and Tsunami of 11 March 2011, inundated over 560 square kilometres of land, devastating a large number of coastal communities, causing over 19,000 casualties (including missing people) (Gokon and Koshimura 2012; Mori et al. 2011). The total economic loss estimates were 210 to 330 billion US dollars (Economist 2011). Another estimation of the damage was made by CATDAT. They estimated the damage between 100 to 500 billion dollars. Approximately 60% of the damage was caused by the tsunami (CATDAT 2011). Many houses were washed away or were completely devastated (Gokon and Koshimura 2012). Six hours after the earthquake on March 11, the International Atomic Energy Agency reported a nuclear emergency at the Fukushima Daiichi nuclear power plant. Due to the strong earthquake, the process of shutting down the three operating reactors was automatically initiated. This process requires diesel generators to power the water pumps that are supposed to supply water to the fuel rods in the reactors in order to cool them down. The operation of the diesel generators failed on the 11th of March, which should have prompted a system of back-up generators to activate. Due to the tsunami inundation, the back-up generators were damaged, and they did not work. After the first explosion, the area within a radius of 20 kilometres from the nuclear plant was evacuated. After the second and third explosions, an exclusion zone was established within a radius of 30 kilometres around the Fukushima Daiichi nuclear power station, and the Japanese authorities took immediate action to cool down the overheated reactors and to prevent con–tamination of the surrounding region.
As many catastro– phic tsunamis have been recorded in the history of Tohoku and seismo- logists have re– marked on the high probability of a ma- jor earthquake that could generate a large tsunami in Japan, the region was considered highly prepared for a tsunami. However,
the event of March Figure 10: Overview of the tsunami survey results and impact of the tsunami in
11 exceeded expectations and overwhelmed the flood risk measures. Dams, barriers and structures flooded or collapsed, evacuation planning was overwhelmed and multiple buildings were washed away (Mori et al. 2011). The tsunami of March 11, 2011, whose return period has been suggested to be 1000-1200 years (Fujita 2011), exceeded assumptions made in advance of the disaster as used for preparation (CDMC 2011). The return period however remains topic of discussion because during the last century an event of the size has occurred in Japan. Due to the frequent occurrence of tsunamis in the Sanriku region, the local society was well prepared and willing to evacuate. Moreover, the so-called “tendenco” local culture of mutual trust may have prevented many casualties. The literal meaning of tendeco is that people trust that their families will also be properly sheltered, and as a consequence, during a tsunami alarm, they shelter themselves immediately without looking for their family members first, which could take precious time (SA-OIC-KU 2011). In other regions in Japan, this tendeco did not take place. The early warning system worked effectively, as the tsunami alarm was issued only three minutes after the earthquake (SA-OIC-KU 2011). Approximately 57% of the people began evacuation directly after the earthquake without additional warning because the felt the earthquake, 31% of the people evacuated after receiving the warning and 12% did not evacuate for a number of reasons (Hayashi 2012).
The first order of the Government was the deployment of 100,000 Japanese troops, 190 aircrafts and 45 boats to immediately start the search and rescue operations (University 2011). Within one week, more than 70,000 Japanese and 2,000 international search and rescue personnel were deployed in Tohoku, consisting mainly of soldiers, fire fighters, doctors and engineers (MOD 2012).
Hurricane Irene, United States (2011) (after (Kolen et al. 2012a))
Hurricane Irene was a powerful Atlantic hurricane that caused extensive damage throughout the Caribbean and along the eastern coast of the United States. The first major hurricane of the 2011 annual hurricane season, it began to show signs of organising east of the Lesser Antilles and public advisories were sent out late on August 21st. On August 22nd, Hurricane Irene made landfall (see Figure 11) as a Category 1 hurricane in Puerto Rico, resulting in severe flooding and property damage.
As Irene intensified, it travelled north of Hispaniola, killing seven people. It transitioned into a Category 3 major hurricane while passing through the Bahamas, resulting in widespread structural damage. Irene’s first landfall in the United States occurred on August 27th as a Category 1 hurricane in the outer banks of eastern North Carolina and moved through south-eastern Virginia. The second landfall in the United States was on August 28th in the Coney Island area of Brooklyn, New York. This caused extensive damage to eastern upstate New York and Vermont, initiating their worst flooding in centuries. In the U.S., Hurricane Irene generated mandatory evacuation orders
for 2.3 million people, and it resulted in at least 40 deaths and long-term power outages for approximately 9 million people. Although not yet finalised, the combined monetary losses in the Caribbean and in the United States are estimated to be 10.1 billion U.S. dollars.
Hurricanes however, also for New York, are a frequent event for the US (OEM 2013). As a consequence citizens as well as emergency services have experience with these events. Australia, Queensland floods (see also (Coates et al. 2012))
During the Queensland floods from December 2010 to January 2011, 35 people drowned, 9 people are missing and 386,000
square miles flooded (Coates et al. 2012). Three-quarters of the state of Queensland was declared as a disaster zone (Bureau-of-Meteorology 2012). The area has a history on flooding with 2 other large scale flooding events in the last 150 years.
The flash flood (10 January) in the city of Toowoomba and the Lockyer Valley and flooding from the Brisbane river on the 11 and 12 of January that flooded parts
of Brisbane had the highest impact. During these two events, 24 people died. A warning for the flash flood was issued by the authorities (Commonwealth Bureau of Meteorology) 1.5 hours before the flash flood occurred in Toowoomba. Citizens also warned each other when they recognised the threat, approximately three-quarters of an hour later than the authorities. Many people evacuated after receiving both warnings. One important lesson is that a preventive evacuation was not possible, although it was the cornerstone of the policy in Australia in response to flooding. Sheltering in place or vertical evacuation might be an alternative and should be further developed.
Reflection on experiences
The cases discussed in this section illustrate that evacuation is a possible measure to reduce the risk in delta areas. However, it is also shown that the local circumstances vary widely and the impact of ontological and epistemic uncertainty and human failure. The complexity of evacuation rises when the consequences increase and when the available time for evacuation decreases. Also unforeseen (or unplanned) events can occur as falling trees, shutters on houses which cannot be opened etc. The general lessons of these events are as follows:
In addition to the probability of flooding also the area a risk and prone to flooding from rivers, storm surges or hurricanes is uncertain at the moment when decisions are made about evacuation;
Each delta has flood protection (as structures, dunes or levees) or planned mitigating measures, such as shelter areas. However these all have (implicit) design standards and