Deaths during the 1953 North Sea storm surge
S.N. Jonkman1 and I. Kelman2
From 31 January to 1 February 1953, a North Sea storm surge devastated coastal
areas of the United Kingdom, Belgium, and the Netherlands. Apart from enormous
economic damage and severe societal disruption, over 2,000 people died across the
three countries. This paper discusses the available data on loss of life in these three
countries and examines the application of these data for loss of life estimations and
general flood management practices.
1 Introduction and general description of the 1953 storm surge
Every year, floods cause enormous damage around the world. In the last decade of
the 20th century, floods killed about 100,000 people and affected over 1.4 billion
people. Although loss of life is often seen as the most significant consequence of
disasters, relatively few studies have focussed on the evaluation of historical
disasters and the lessons learnt. More than 50 years after the disastrous events which
formed the 1953 storm surge on the North Sea this paper discusses the available data
on loss of life, and discusses the application of these data for loss of life estimations
and general flood management practices.
Aim of this study
After giving a general description of the events and their damages (remainder of
section 1), the available data on loss of life in this event is discussed (section 2).
Section 3 deals with the application of this data in loss of life estimation for floods.
Lessons learnt and issues for future flood management are outlined in section 4.
From 31 January to 1 February 1953, a North Sea storm surge devastated coastal
areas of the UK, Belgium, and the Netherlands. Apart from enormous economic
damage and severe societal disruptions, over 2,000 people died across the three
countries. The floods occurred during a spring tide combined with low atmospheric
pressure causing elevated sea levels and strong winds piling up the water on shore.
Furthermore, the sea defences were reported to be in bad condition in all three
General description of 1953 storm surge
1 Bas Jonkman, Road and Hydraulic Institute, Ministry of Transport, Public Works and Water
PO Box 5044, 2600 GA, Delft, the Netherlands
Delft University of Technology, Faculty of Civil Engineering
telephone: +31-15-2518443; fax: +31-15-2518555; email: firstname.lastname@example.org
2 Ilan Kelman, Deputy Director, Cambridge University Centre for Risk in the Built Environment
The Martin Centre, 6 Chaucer Road, Cambridge, England, CB2 2EB, U.K.
telephone: +44-1223-331715; fax: +44-1223-331701; email: email@example.com
countries, as limited priority was given to maintenance and strengthening of flood
defences in the post-war years.
In the UK, vast areas in Lincolnshire, Norfolk, Suffolk, Essex and Kent were
inundated. The magnitude and extent of the effects of the storm surge which hit are
illustrated by Summers (1978):
On 31 January and 1 February 1953 a great storm surge, accompanied by
gale force winds, swept out of the north, causing widespread flooding of
coastal areas, and involving grievous loss of life and extensive damage to
property. The piled-up waters of the North Sea, whipped by the northerly
gales to huge tidal levels, smashed through the sea-wall defences in
hundreds of places from Spurn Head to Kent, scattering like proverbial
chaff before the wind thousands of tons of stone and concrete. The
damage extended over 1,000 miles of coastline and involved breaches in
the defences at some 1,200 sites. In some places not a mile of sea-wall
Parts of Belgium were also inundated. Of the 66 kilometres of coastline, 4.6 km were
severely damaged and at least 8 fatalities occurred (Martens, 2003).
In Belgium and the UK, only coastal areas were inundated, but the flooded area in
southwest Netherlands was much larger as the hinterland was below sea level. The
disaster caused enormous economic damage and at least 1,835 fatalities. About
250,000 people were affected and more than 47,000 cattle and 140,000 poultry were
killed. 3,000 Residences and 300 farms were destroyed, with more than 40,000
houses and 3,000 farms being damaged. Approximately 200,000 hectares were
inundated and the total material damage was estimated at 1.5 billion guilders
(http://www.delta2003.nl). Table 1 summarises the consequences.
Table 1: Overview of damage of 1953 storm surge.
Inundated area (hectares)
Economic damage (1953 values)50 million pounds
1.5 billion guilders
Although the scale of disaster differs amongst the three countries, the event
mortality—defined as the number of fatalities divided by the number of affected
people—is the same order of magnitude for the UK and the Netherlands. Data on
other storm surges shows that a 1% average mortality seems to be a good first
approximation (Figure 1).
Figure 1: Mortality for some historical coastal storm surges with the 1% mortality
line drawn. Data are from EMDAT (2004).
2 Data on fatalities during the 1953 storm surge
Fatality data have been collected from different sources and aggregated in fact sheets
by country which are
published on the internet
These death figures account for only the people who died immediately, not including
those who later suffered premature deaths due to the psychological and/or physical
toll exacted on them from having experienced the storm surge. Despite extensive
research and material published on this event, reported death tolls vary substantially.
Only statements with explicit numbers, including zero, are recorded in these fact
sheets. One important source which these fact sheets do not yet fully cover is local
For the Netherlands, two main sources state the number of fatalities by village
(Waarts, 1992 and Delta, 2003). The official death toll has long been 1,835, but
Slager (1992) found that the actual death toll was 1,836: a newborn boy who
drowned in Cappele had never been included in the official counts. The website
2,278 victims as dying immediately, but 483 married women were counted twice.
Each of these women is listed once under her birth surname and once as a married
woman under her husband’s surname. Subtracting the 483 double counts from the
2,278 total yields the 1,795 figure reported by Delta (2003) as being the total number
who died immediately. Holland (2003) then confirms that “The difference between
1,835 and 1,795 is accounted for [by] a number of people that died in an uncertain
period after the disaster because of the illness and suffering they were undergoing
during the first hours or days.”
For the UK, more sources are available, the most prominent being Grieve (1959),
Pollard (1978), and Summers (1978). Compiling many sources yields a total death
toll between 304 and 313 with 307 being the most common number quoted. In
reality, that figure refers to deaths on land in eastern England. In addition,
approximately 132 people died when the Princess Victoria ferry sank in the Irish
Sea. Nineteen others died in Scottish waters (Hickey, 2001) and several dozen were
killed when ships sank in the North Sea. One part of the UK Fact Sheet (Kelman,
2003) is displayed in Figure 2.
Figure 2: Example from CURBE Fact Sheet 3: UK Deaths from the 1953 Storm
Surge (Kelman, 2003).
For Belgium, the different sources report total fatality numbers between 8 and 22
(see Gerritsen et al., 2003). The highest estimates factor in fatalities outside the
coastal area, due to river flooding near Antwerpen. Most reports note that 8 deaths
occurred along the Belgian coast.
Flood fatalities are rarely examined across several flood disasters to identify trends in
medical causes of deaths along with the vulnerabilities which led to those deaths.
Investigation of relatively recent inland floods in the USA and Europe revealed some
patterns with respect to death causes and vulnerabilities (Jonkman and Kelman,
2005). However, higher-fatality events seem to exhibit different mortality patterns
than the smaller-scale events. Insights into causes and circumstances are important
knowledge and analysis needed for the mitigation of the potential consequences of
floods. Based on event descriptions, further discussion here examines the most
relevant causes and circumstances of the fatalities in the 1953 flood.
Causes and circumstances of 1953 flood disaster deaths
Data on fatalities in the Netherlands due to this disaster have been collected by
Waarts (1992) and Duiser (1989) from memorial volumes and official reports. Both
reports give loss of life and hydraulic circumstances by municipality. Overall, the
data covers 1,726 fatalities which occurred in 45 locations. Based on the descriptions
from memorial volumes, three categories of fatalities are distinguished: fatalities due
to high flow velocities, due to rapidly rising waters, and due to other causes. Table 2
shows the distribution of reported fatalities over the three categories.
Table 2: Categorised data on fatalities caused by the 1953 disaster in the
Netherlands, based on Waarts (1992)
Rapidly rising waters
High flow velocities
Most fatalities occurred at locations where there was insufficient warning and where
the water rose rapidly to form a deep flood. In addition, qualitative descriptions of
the 1953 floods in the Netherlands by Slager (1992) show that in the high-mortality
locations, large numbers of buildings collapsed due to the severe conditions and poor
building quality. As a consequence the poorest communities suffered most fatalities.
The numbers reported by municipality by Waarts (1992) and Duiser (1989)
correspond relatively well with the numbers given on http://www.delta2003.nl. See
Jonkman et al. (2003) for a further comparison of the numbers reported in different
Martens (2003) mentions that 8 coastal fatalities occurred: 7 in Oostende and 1 on
sea. Of the 7 fatalities in Oostende, 3 died in the city centre: 2 due to drowning and 1
due to a heart attack. The other 4 drowned when a house in Sas Slijkens flooded.
Other sources report higher death tolls for Belgium: between 10 and 22. See
Gerritsen et al. (2003) for an overview.
The high death toll is mainly due to the unexpected occurrence of the flood after
sunset and without warning. Death tolls were highest in seaside towns with low-
quality buildings, often consisting of wooden prefabricated houses. For example, 39
fatalities occurred at the Felixstowe prefab estate. At Canvey Island, the death toll
amounted to 58, with many people surprised in their houses, several of which
collapsed, after a sea defence failed. At Jaywick, many of the 37 fatalities were
elderly who were surprised or trapped in their homes.
An indication of the vulnerability of the elderly and the importance of building
collapse is in the following quotations:
And consistently, all round the east coast, the eventual death tolls showed
that it was the elderly, who went to bed early and had meagre reserves of
energy even if they had time to realise what was happening when the
water hit them, who were most vulnerable… It is significant that the
stretches along the stretches of coast where the casualties were heaviest –
the Mablethorpe Sutton area of Lincolnshire, the Hunstanton to Lynn
area, Jaywick, Canvey Island – were largely seaside shanty-towns
consisting mainly of timber bungalows never intended for winter
occupation.” (Pollard, 1978).
At Canvey Island, 42 out of 58 fatalities were aged over 60. At Jaywick, 28 of the
reported 34 fatalities were over 60. In south Lynn, all 14 fatalities were over 60.
The difference in death toll between Harwich (8 fatalities) and Jaywick (37 fatalities)
can also be attributed to different building quality. Shortly after the war, people were
living in all kinds of temporary housing, prefabricated cottages, and wooden chalets.
Deaths were not necessarily drowning. Of the 41 post mortems carried out at Canvey
Island, 14 people died of other causes than drowning, such as shock and heart attack.
Although loss of life is seen as one of the major consequences of floods, limited
methods exist to estimate the number of flood fatalities under different
circumstances. Investigations are reported here into how the available data from the
1953 storm surge can be used to develop flood mortality functions. These functions
relate the local flood mortality (i.e. fraction of the population killed) to the local
flood characteristics (e.g. flood depth, flow velocity, and rate of water rising).
Loss of life estimation based on historical data
A concept is proposed in which different hazard zones are distinguished—based on
the assumption of a breach in a sea defence or flood defence—which represent
different flood conditions and different mortality patterns. The following three
hazard zones are used here:
• • • •Breach zone: This zone is dominated by the physical forces of water near the
breach. High flow velocities can cause people to lose their stability and buildings to
• • • •Zone with rapidly rising waters: Due to the rapid rise of water depth, people will
not be able to reach higher ground or higher floors of buildings. This rapid rise is
especially hazardous in combination with large water depths.
• • • •Remaining zone: In other areas, fatalities may be due to other causes. For example,
hypothermia, exhaustion, or building collapse after a long period of hydraulic load
being imposed. People’s behaviour will be an important factor in mortality.
Within each zone, flood characteristics are generally relatively homogeneous. These
zones are representative for loss of life estimation for large-scale coastal floods, even
if all three zones do not exist such as if no structural sea defence is used.
Figure 3 illustrates the different zones in relation to breach location for a
characteristic large-scale coastal flood in the Netherlands. The boundaries of the
rapidly rising waters might be caused by the presence of elements creating barriers,
such as building rows, lowered streets, dikes, or steep contours. For other types of
floods, the situation and proportional area of the hazard zones might be different.
Rapidly rising water
Figure 3: Hazard zones distinguished for loss of life estimation. Numbers indicate
subzones, delineated by features such as buildings, streets, dikes, and contours.
A relationship between causes of death and flood zones exists, as certain causes of
death dominate in each zone. For example, the causes of death associated with high
velocities, such as human instability and building collapse, will be dominant in the
breach zone. Nonetheless, multiple causes of death will usually occur within one
flood zone because people’s characteristics also influence vulnerability, including
age, gender, physical ability, mental ability, and warning type received.
For the three hazard zones in Figure 3, basic mortality functions have been derived
from the 1953 flood data for Netherlands and UK. Some available data from storm
surges in Japan have been added to the dataset. For all these cases, multiple data
points (locations) were available, allowing statistical analysis. Each location is
considered to be a separate event with different flood conditions.
Mortality functions for the two non-breach zones are shown in Figures 4 and 5. For
the breach zone, insufficient data were available, hence a damage criterion is
proposed limiting the combination of water depth and flow velocity.
water depth (m)
Figure 4: Mortality as a function of water depth for locations in the two zones with
rapidly-rising water. The best-fit function is FD(h)=exp((h-4.58)/0.69), where FD(h)-
mortality [-]; h – water depth [m]
water depth (m)
NL 1953 (33)
UK 1953 (11)
Jap 1959 (28)
Jap 1934 (8)
Jap 1954 (10)
Figure 5: Mortality as function of water depth for the remaining zone. Different
flood cases are included and numbers between brackets show the number of points
included. The best-fit function is FD(h)=exp((h-10,7)/1,59)
These functions relate flood mortality to flood characteristics. In ongoing research,
so-called correction factors are developed to correct the mortality numbers for factors
such as level of warning given (although “level of warning received” and “level of
warning obeyed” could be two other parameters), the strength of buildings, and
rescue actions attempted.
The available case studies are all coastal floods which occurred unexpectedly and for
which no organised evacuation took place beforehand. For other case studies, the
effects of evacuation would have to be incorporated. The population present
originally (usually referred to as “population at risk”) has to be adjusted for the
number of evacuees. By combining an evacuation model with loss of life models,
these functions could be applied for estimating loss of life for different flood cases
thereby indicating the potential importance of proper warning, obedience to
warnings, and timely and well-organised evacuation.
The data, descriptions, and sources from this tragic event provide important lessons
for contemporary flood and emergency management practices as well as increasing
awareness that storm surge disasters still threaten these coastlines. The issues to
•Many people were not aware of their vulnerability to storm surge flooding and
received no warning meaning that they were surprised by the floodwater.
•Darkness made rescue, including self-rescue, difficult.
•Lack of communications inhibited warning other areas that flooding was imminent
and prevented requesting external assistance.
•The elderly proved to be exceptionally vulnerable as they were trapped in their
homes with few capabilities or support to escape.
Lessons learnt and concluding remarks
•In areas with poor buildings which collapsed, many families succumbed.
•Overall, most fatalities occurred in areas with vulnerable and low quality buildings.
•Many lives were saved by individual bravery, whether or not they had any training
or equipment. Post-war troops stationed around eastern England contributed
remarkably to rescue efforts there.
•Many sea defences collapsed, indicating that the appropriateness of relying on
structural sea defences should be examined (see also Fordham, 1999 and Kelman,
2001), particularly where other priorities will preclude needed maintenance, and
that potential failure modes of structural defences should be investigated more
Based on these issues, both recommendations and questions arise related to
preventing loss of life in North Sea storm surge, although some points could readily
apply to flood management internationally:
•Improved systematic recording of flood fatalities, and the wider health effects of
floods, would be essential for providing a more solid basis for recommendations. In
particular, vulnerable locations and people could be clearly identified helping to
reduce vulnerability against coastal flood disasters.
•The capability to respond to a large-scale flood emergency with appropriate
technical rescue personnel, equipment, and training should be reviewed. Relying on
individual heroism and ad hoc methods is an inappropriate response strategy.
•Population demographics should be examined for sources of vulnerability. Between
1951 and 1991, the UK’s population increased 12% while the population of many
coastal areas in eastern England increased between 17% and 92%. Are elderly
people still residing in places likely to be inundated?
•What awareness level exists in terms of acting on warnings received? How could
warnings be improved (e.g. Handmer, 2000) in terms of accuracy, timeliness, and
•Heavy reliance is currently placed on new technology, such as satellite monitoring,
instant communication systems, and evacuation modelling. To what extent is such
reliance justified? What contingency planning exists in case of technology failure,
such as communication systems breakdowns which have occurred in recent floods?
•Even with current technology, no more than 48 hours warning of a major storm
surge event is likely to be available. Is that time sufficient for warning, preparation,
and appropriate action, including evacuation of large populations?
•How can individual actions leading to potentially fatal vulnerabilities be tackled?
Examples are flood tourism, opposing evacuation orders, and gratuitous risk-taking
behaviour during floods.
We have the ability and resources to avoid loss of life in floods. We have the
responsibility to do so by tackling the above questions thoroughly.
Duiser J.A. 1989, Een verkennend onderzoek naar methoden ter bepaling van
inundatieschade bij dijkdoorbraak (in dutch – research on the methods for
estimation of the consequences of flood damage), TNO report
EM - DAT. (2004).
http://www.cred.be/emdat, Université Catholique de Louvain, Brussels.
Fordham, M. (1999) Participatory Planning for Flood Mitigation: Models and
Approaches. The Australian Journal of Emergency Management, 13(4), 27-34.
Gerritsen, H., Jonkman, B., Kelman, I., Martens, C., and Waarts, P. (2003). CURBE
Fact Sheet 7: Belgium Deaths from the 1953 Storm Surge, Version 3, 27 July
2003 (Version 1 was
Grieve, H. (1959). The Great Tide, The County Council of Essex, Essex, UK
Handmer, J. (2000) “Are Flood Warnings Futile? Risk Communication in
Emergencies”, The Australasian Journal of Disaster and Trauma Studies,
Hickey, K. R. 2001. “The Storm of 31 January to 1 February 1953 and its Impact on
Scotland”, Scottish Geographical Journal, 117(4), pp. 283-295.
Holland, A. (2003). Personal communication in March 2003 from the Project
Coordinator of Delta 2003, http://www.delta2003.nl
Jonkman S.N. and Kelman I., (2005), “An analysis of causes and circumstances of
flood disaster deaths”, Disasters, in press.
Jonkman, B., I. Kelman, and P. Waarts. (2003). CURBE Fact Sheet 6: Netherlands
Deaths from the 1953 Storm Surge, Version 1, 31 March 2003, from
Kelman, I. (2001). “The Autumn 2000 Floods in England and Flood Management”.
Weather, 56(10), 346-348,353-360.
Kelman, I. (2003). CURBE Fact Sheet 3: UK Deaths from the 1953 Storm Surge,
Version 4, 10 November 2003 (Version 1 was 30 November 2002), from
Martens C. (2003). De storm van 1953: literatuurstudie naar de gevolgen van de
overstromingen voor de Belgische Kust, Nota I/NO/11226/03.085/CMA, 15
Pollard, M. (1978). North Sea Surge: The Story of the East Coast Floods of 1953,
Terence Dalton Limited, Lavenham and Suffolk, UK
Slager K.: 1992, De ramp - een reconstructie (in Dutch: The disaster – a reconstruction), de
Koperen Tuin Publishers
Summers, D. (1978). The East Coast Floods, David & Charles (Publishers) Limited,
Waarts P. (1992). Methode voor de bepaling van het aantal doden als gevolg van
inundatie, TNO B-91-1099.
International Disaster Database from
31 March 2003), from