The-dependence-of-UK-extreme-sea-levels-and-storm-surges-on-the-North-Atlantic-Oscillation_2007_Continental-Shelf-Research_1.pdf

ARTICLE IN PRESS

0278-4343/$ - see

doi:10.1016/j.csr

�Correspondifax: +44151 79

E-mail addre

Continental Shelf Research 27 (2007) 935–946

www.elsevier.com/locate/csr

The dependence of UK extreme sea levels and storm surges onthe North Atlantic Oscillation

P.L. Woodworth�, R.A. Flather, J.A. Williams, S.L. Wakelin, S. Jevrejeva

Proudman Oceanographic Laboratory, 6 Brownlow Street, Liverpool L3 5DA, UK

Received 15 August 2006; received in revised form 1 December 2006; accepted 11 December 2006

Available online 3 January 2007

Abstract

The role of the North Atlantic Oscillation (NAO) in effecting changes in winter extreme high and low waters and storm

surges in UK waters has been investigated with the use of a depth-averaged tide+surge numerical model. Spatial patterns

of correlation of extreme high and low waters (extreme still water sea levels) with the NAO index are similar to those of

median or mean sea level studied previously. Explanations for the similarities, and for differences where they occur, are

proposed. Spatial patterns of correlations of extreme high and low and median surge with the NAO index are similar to the

corresponding extreme sea-level patterns. Suggestions are made as to which properties of surges (frequency, duration,

magnitude) are linked most closely to NAO variability. Several climate models suggest higher (more positive) average

values of NAO index during the next 100 years. However, the impact on the UK coastline in terms of increased flood risk

should be low (aside from other consequences of climate change such as a global sea-level rise) if the existing relationships

between extreme high waters and NAO index are maintained.

Crown Copyright r 2007 Published by Elsevier Ltd. All rights reserved.

Keywords: Sea-level changes; Extreme values; Storm surges; Flood forecasting; Climate changes

1. Introduction

This paper investigates the dependence of UKextreme sea levels and storm surges upon the NorthAtlantic Oscillation (NAO). The NAO is the majormode of North Atlantic atmospheric variability,with an NAO index defined by the differencebetween normalised sea-level pressures representa-tive of the Azores High and Icelandic Low (Hurrell,1995; Jones et al., 1997). Periods with large positiveindex correspond to strong westerly winds over the

front matter Crown Copyright r 2007 Published by

.2006.12.007

ng author. Tel.: +44151 795 4800;

5 4801.

ss: [email protected] (P.L. Woodworth).

UK and northern Europe. As the magnitude ofstorm surges depends primarily upon the wind stressover the continental shelf (Pugh, 2004), some kindof relationship between extreme sea levels and stormsurges and the NAO can be anticipated. However,so far as we are aware, the topic has not beenstudied before in detail, and has been noted asrequiring investigation (NERC, 2005).

There are two particular motivations for such aninvestigation. The first is a practical one. Rodwellet al. (1999) proposed that ocean-atmosphereinteractions provide a link between summertimecentral Atlantic sea surface temperatures andconditions in the atmosphere during the followingwinter, which determine storminess over Europe.

Elsevier Ltd. All rights reserved.

ARTICLE IN PRESSP.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946936

As a consequence, the NAO index in winter(DJFM) is to some extent predictable from knowl-edge of central Atlantic sea surface temperatures inthe preceding summer. Therefore, if a relationshipbetween extreme sea levels and NAO exists, there isthe possibility to provide advance warning ofanomalous winter flood risk on the basis ofsummertime sea surface temperature observations.

The second motivation stems from an interest inknowing whether major changes are likely in theclimatology of UK extreme sea levels and surges,for input to studies such as the UK Climate ImpactsProgramme (Hulme et al., 2002). One approach tothis question is to employ an Atmosphere OceanGeneral Circulation Model (AOGCM) to simulatetime series of future regional wind and air pressurefields, and to use those forcing fields in a stormsurge model to predict extreme sea levels. Severalgroups have demonstrated that such studies arefeasible (e.g. Von Storch and Reichardt, 1997;Flather and Smith, 1998; Lowe et al., 2001; Wothet al., 2006). However, an alternative approach is tomake use of the predictions of future winter NAOindex by the same AOGCM. If the dependenceof extreme sea levels on the NAO is known, oneshould then be able to infer changes in theclimatology of extremes. In practice, both ap-proaches are desirable.

2. Data sets

Our analysis makes use of 49 year (1955–2003)runs of a two-dimensional tide+surge model for theNW European continental shelf (Flather et al.,1998). The model has a relatively coarse spatialresolution (1/31 latitude by 1/21 longitude). How-ever, it has been demonstrated to provide a goodrepresentation of tides and surges on the shelf andwas formerly the model used for operational floodforecasting and warning in the UK (Flather, 2000).Eight tidal constituents (Q1, O1, P1, K1, M2, S2,K2 and N2) are included with tidal forcing at theopen boundary taken from a model of the NEAtlantic (Flather, 1981). Meteorological forcing isprovided by 6-hourly winds and atmosphericpressures from the Norwegian MeteorologicalInstitute (Det Norske Meteorologiske Institutt,DNMI). An inverse barometer approximation isassumed at the open boundary which, althoughnecessarily imperfect in its omission of dynamic air-pressure-induced signals, should be a good approx-imation of sea-level change in deep water (cf. Ponte,

1994). As a consequence, one expects the limitationsof such an approximation to have little impact onthe effective computation of surge heights over themodel domain. Reservations have been expressedconcerning the temporal–spatial homogeneity of theDNMI data set in data-sparse ocean areas, espe-cially around Greenland (WASA Group, 1998).However, such inhomogeneities are considered to besmall over the NW European continental shelf(Gunther et al., 1998; Langenberg et al., 1999), andcomparisons of the observed and modelled heightsof individual major surges have been found to be ofacceptable (sub-decimetric) accuracy at UK loca-tions (Flather et al., 1998).

Two model runs were made, first with tidalforcing only (‘t’), and then with both tidal andmeteorological forcing (‘tm’). Hourly fields of sea-level elevations were output in each run. Thedifference between the ‘tm’ and ‘t’ elevation fieldsprovides storm surge fields (‘s’), where ‘surge’defined in this way includes tide–surge interaction.In this study, we shall use primarily the ‘tm’ hourlyvalues, which in principle will correspond to the sealevels which would be recorded by tide gauges andare of most interest with regard to flood risk, andthe ‘s’ values, which one might anticipate to be mostsensitive to variability in the NAO.

The NAO index values, defined as the differencebetween the normalised sea-level pressures atGibraltar and southwest Iceland (Jones et al.,1997), were taken from the Climatic Research Unit,University of East Anglia web site (www.cru.uea.a-c.uk).

3. Sea-level relationships to the NAO

The 49 years of model run have 48 winter(DJFM) periods. For each winter, the ‘tm’ valueswere used to calculate extreme high and low waters,median sea level (MeSL) and mean sea level (MSL).Unless stated otherwise, extreme high (and low)waters were defined in terms of the 99 (and 1)percentiles of the winter hourly sea-level values, i.e.as the levels, which the sea is above (or below) 1%of the time (29 h each winter). The choice of suchpercentiles, instead of the true extremes (the 100 and0 percentiles), guards against the presence ofanomalous sea-level values, and is often essentialin analysis of tide gauge data (e.g. Woodworth andBlackman, 2004). Differences in findings due tosuch choices are mentioned below. MSL time series

ARTICLE IN PRESSP.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946 937

were defined as the arithmetic average of the ‘tm’hourly values in each winter period.

Fig. 1(b) shows the correlation coefficients be-tween winter MeSL and NAO index, while Fig. 1(e)presents the sensitivity of MeSL to changes in theNAO index obtained by linear regression. Correla-tion coefficients larger than 0.28 can be consideredsignificantly different from zero at 95% confidencelevel given 48 independent samples. The largestpositive and negative correlations are found in theNE and SW of the shelf, respectively, with the largestsensitivities in the shallow waters of the easternNorth Sea and German Bight, where large surges aregenerated by the westerly winds. Almost identicaldistributions are obtained with the use of MSLinstead of MeSL. Figs. 1(b,e) are very similar tothose of Wakelin et al. (2003), who investigated

-15° -10° -5° 0° 5° 10°

50°

55°

60°

-0.4

0.6

0.6

0.6

0.6

-15° -10° -5° 0°

-0.4

0.5

0.6

0.7

0.7

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-15° -10° -5° 0° 5° 10°

50°

55°

60°

20

30

30

30

40

40

50

70

-15° -10° -5° 0°

30

30

30

-100 -80 -60 -40 -20 0

a b

d e

Fig. 1. (a) Correlation coefficient between NAO index and winter extr

coefficients for MeSL and (c) extreme low water (1 percentile); (d) sens

index (mm/unit index); (e) corresponding sensitivity for MeSL and (f)

winter MSL from the same model for a slightlyshorter period (1955–2000), and of Woolf et al.(2003), who studied North Atlantic altimeter and tidegauge data with particular attention to the 1990s.

Fig. 1(a,d) and (c,f) show the correspondingdistributions for extreme high and low sea levels(99 and 1 percentiles), respectively. Figs. 1(a,c)indicate that extreme sea levels have weakercorrelations with the NAO than does MeSL,reflecting the fact that the index and MeSL arewinter-average quantities. Nevertheless, the spatialpatterns of correlation coefficients and of sensitivityto the NAO are similar. This indicates that changesin NAO index affect sea levels in a similar waythroughout the tidal range.

This conclusion is confirmed if one considerscorrelations with, and sensitivities to, the NAO for

5° 10° -15° -10° -5° 0° 5° 10°

50°

55°

60°

0.3

0.40.4

0.4

0.2 0.4 0.6 0.8 1.0

5° 10°

30

40

-15° -10° -5° 0° 5° 10°

50°

55°

60°

10

20

20

30

40

20 40 60 80 100

mm/unit

c

f

eme high water (99 percentile) for 1956–2003; (b) corresponding

itivity of change in winter extreme high water to change in NAO

extreme low water.

ARTICLE IN PRESSP.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946938

extreme high and low waters, where the extremesare expressed relative to MeSL in the same winter.Fig. 2(a) demonstrates that almost no significantcorrelation remains between high water extremesand NAO, except for part of the NE North Sea andSkaggerak, where sensitivities are also relativelylarge (Fig. 2(c)). This is consistent with findingsfrom tide gauge data by Woodworth and Blackman(2004) and Tsimplis et al. (2005) who observed agreater response of extreme high waters to NAOchange than of MSL at some Scandinavian stations.Correlations between extreme low waters, aftermedian subtraction, and NAO are negativelycorrelated in the western North Sea (Fig. 2(b)).However, the corresponding sensitivities are small(Fig. 2(d)).

-15° -10° -5° 0° 5° 10°

50°

55°

60°

0

0

0.1

0.2 0.3

-15°

-1.0 -0.8 -0.6 -0.4 -0.2 0.0

-15° -10° -5° 0° 5° 10°

50°

55°

60°

0

0

0

0

10

2030

-15°

-100 -80 -60 -40 -20 0

a b

dc

Fig. 2. (a) As Fig. 1(a) but with extreme high water values expressed r

water values expressed relative to the winter median; (c) sensitivity of c

index); (d) corresponding sensitivity for extreme low water.

It is well known that the winter NAO index in thesecond half of the 20th century contained a largesecular trend (0.23 units/decade over 1956–2003).Therefore, a question arises as to whether thecorrelations in Fig. 1(a–c) are due to the long termtrend rather than interannual or decadal variabilty.This has been addressed by detrending the NAOindex and the extreme high and low water andMeSL time series yielding distributions similar tothose of Fig. 1(a–c). This indicates a similar sea-level response to the NAO at shorter and longertimescales.

Tests were made to see if the choice of 99 (and 1)percentiles as extreme high (and low) water levelsyielded similar conclusions to the choice of percen-tiles nearer to the true extremes (99.9 and 0.1) or to

-10° -5° 0° 5° 10°

50°

55°

60°

-0.3

0.1

0.2 0.4 0.6 0.8 1.0

-10° -5° 0° 5° 10°

50°

55°

60°

-10

-10

0

20 40 60 80 100

mm/unit

elative to the winter median; (b) as Fig. 1(c) but for extreme low

hange in winter extreme high water relative to median (mm/unit

ARTICLE IN PRESSP.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946 939

the use of the true extremes themselves. Distribu-tions of correlation coefficients and sensitivities forthe 99.9 and 0.1 percentiles and for the trueextremes were found to be very similar at mostlocations to those of Figs. 1 and 2, with slightlyweaker correlations and larger sensitivity values.The original choice was maintained for this paper soas to treat the model information in the same way towhich tide gauge data are likely to be analysed.

The present study does not require a removal ofperigean (approximately 4.5 year) tidal effects frompercentiles in addition to medians, as in the Wood-worth and Blackman (2004) analysis of a quasi-global tide gauge data set. That is because thenumerical model omits those smaller tidal constitu-ents which result in perigean variations (e.g. L2)(Pugh, 1987).

4. Properties of Storm Surges

It is of interest to ask why the extreme high andlow waters and MeSL have broadly similar correla-tions with the NAO. Fig. 3 shows an estimate of theduration of large winter storm surge events obtainedby considering periods of data for which ‘s’ valuesexceed the time-averaged (over 48 winters) 95percentile surge at each grid box in the model. Thischoice of threshold selects 145 h of large surge in a

-15°

-15°

-10°

-10°

-5°

-5°

0°

0°

5°

5°

10°

10°

50° 50°

55° 55°

60° 60°

10

15

15

20

20

20

25

25

0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0

hours

Fig. 3. Duration of a large winter surge event obtained with the

use of the time-averaged 95 percentile as described in the text

(hours).

typical winter. The duration of a large storm surgeevent is determined by the difference between theup- and down-crossing times of ‘s’ values across thethreshold level, in a similar manner to which waveperiods are calculated (e.g. Vassie et al., 2004). Asimilar method was employed previously by Wothet al. (2006). Values of large surge duration varyspatially, with larger values in the deeper Atlanticwaters where air pressure changes are relativelymore important than winds in generating surgeevents. Smaller values are found in the shallowwaters of the North Sea. The averaged durationacross the model domain is 14.4 h. However oneselects reasonable values for the threshold, it is clearthat the typical large surge duration will be at leastcomparable to the tidal period and, therefore, willcontribute to the still water level throughout thetidal range.

Fig. 4(a) demonstrates an alternative way ofmaking the same point with the use of all ‘s’ values,not just the largest ones. It shows power spectra of‘s’ at the 7 representative locations indicated in Fig.4(b). Small amounts of tidal energy are apparentwhich originate from the modelled tide in the ‘t’ and‘tm’ runs being slightly different, owing to themodifications in water depth. It can be seen thatthere is little energy in ‘s’ at frequencies higher thanthat of the semidiurnal tide (0.081 cycles/h).

It is also of interest to ask why there aredifferences between the relationships of extremehigh and low waters and MeSL to the NAO. Thesecan be inferred from the statistical properties ofwinter ‘s’ values displayed in Fig. 5. It is well knownthat the prevailing westerlies result in decimetricvalues of wind-setup in the German Bight, and thatthe standard deviation of ‘s’ is also largest there.The distribution of ‘s’ values departs from a normalone in several ways. For one thing, it is clear thattime series will be highly serially correlated. Inaddition, the distribution on the shelf is skewedtowards large positive values, especially in theshallow waters of the southern North Sea, EnglishChannel and eastern Irish Sea, and has largekurtosis (peakiness). The skewness arises partlyfrom tide–surge interaction, but primarily from thefact that the meteorological forcing is itself skewedwith long tails in low air pressures and high windspeeds (Wilks, 1995; Barry et al., 2003). Therefore, itis inevitable that there will be some differencesbetween extreme high and low waters and MeSL,with the former usually presenting the largerdifferences, as indicated by comparing Fig. 2(c–d).

ARTICLE IN PRESS

a

b

Fig. 4. (a) Power spectra of ‘s’ for the 7 locations shown in (b).

Arbitrary units. The spectra were computed from continuous

hourly values of ‘s’ for 1955-2003.

P.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946940

5. Surge relationships to the NAO

Fig. 6 presents distributions of correlation coeffi-cients and sensitivities for the 99, 50 and 1percentiles of winter ‘s’ values, instead of the ‘tm’values of Fig. 1. The 50 percentile distributions(Figs. 1(b,e) and 6(b,e)) are almost identical as onewould expect, while those for low waters are alsoalmost the same. The 99 percentile correlationcoefficients are also similar (Figs. 1(a) and 6(a)),with larger sensitivity values for ‘s’ in the easternNorth Sea (Figs. 1(d) and 6(d)).

It is important to realise that the hours of data,which contain the extreme values for ‘s’ are notnecessarily the same as those, which form theextremes for ‘tm’. Surges can occur throughout thetidal cycle, and, for example, a large positive surgecould occur at low tide resulting in a middling ‘tm’value. Nevertheless, it is of interest to consider inmore detail the relationship of large positive surgesto the NAO as they would be of potentialimportance to flood risk if they did occur at hightide. If there is a clear relationship, one can askwhether it is due to there being more surges, or oneswith larger amplitude or duration.

As before, use is made of the time-averaged (over48 winters) 95 percentile surge at each model gridbox to define the threshold for a large surge event.Fig. 7(a) shows the correlation between the numberof large surge events in a winter and the NAO index.The pattern is similar to those seen before. Thesensitivity between the number of events and NAO,obtained by linear regression, is shown in Fig. 7(d).An increase of one NAO unit results in typically 3or 4 extra events in the North Sea compared to anaverage of 10 events in a typical winter (with thischoice of threshold). Fig. 7(b) and (c) show thecorresponding correlation coefficients for the totalnumber of hours each winter with a surge overthreshold, and the average height of the surgesabove the threshold, respectively, while Fig. 7(e)and (f) give the corresponding sensitivities. Com-parison of Fig. 7(d) and (e) confirms that surgeevents correspond to approximately 14 h overthreshold. One concludes that, in the North Seawhere correlations are relatively large, any increasein the NAO index results in (1) an increase in thenumber of hours over threshold, (2) for a bandstretching from northern Scotland to northernDenmark and the Skaggerak, an allocation of thosehours preferentially into more events, and (3) onlyminor change in the surge amplitude. Point (2) willbe returned to below.

6. Secular trends in extreme high and low waters,

MeSL and storm surges

The NAO trend during the past few decades hasresulted in long term trends in sea level on the NWEuropean continental shelf. Fig. 8(a)–(c) demon-strates that secular trends in winter extreme highwater, MeSL and extreme low water during1956–2003 were similar in having largest values inthe eastern North Sea and Skaggerak. Trends

ARTICLE IN PRESS

-15° -10° -5° 0° 5° 10°

50°

55°

60°

50

50

50

50

-100 -75 -50 -25 0 25 50 75 100 125 150

-15° -10° -5° 0° 5° 10°

50°

55°

60°

150

150

150

200

250

0 50 100 150 200 250 300 350 400 450 500

mm

-15° -10° -5° 0° 5° 10°

50°

55°

60°

0.4

0.4

0.4

0.4

0.4

0.4

0.4

0.80.8

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

-15° -10° -5° 0° 5° 10°

50°

55°

60°

0

0

0

0

0

0.8

1.6

2.4

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

a b

dc

Fig. 5. (a) Average value, (b) standard deviation, (c) skewness and (d) reduced kurtosis of ‘s’ using values from 48 winter periods.

P.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946 941

around the UK were considerably smaller than thosein the German Bight, as noted previously byLangenberg et al. (1999). (Trends for the April–No-vember part of the year were negligible throughoutthe entire region, reflecting partially the lower energyin air pressure and zonal wind variability in that partof the year.) In addition, there were large differencesbetween extreme high and low waters and MeSL inUK waters. Fig. 8(a) contains an additional band oflarge trend across the northern part of the North Sea,located approximately at the areas of maximumsensitivity of number of large surge events and hoursover threshold to NAO change in Fig. 7(d,e) (i.e.point 2 above).

Large differences in trends in extremes areobtained when selecting model data for slightly

different epochs, as applies to the study of meansea-level trends from tide gauge data (e.g. Pugh,1987; Tsimplis et al., 2005). For example, it isknown that the winter NAO index was particularlylarge and positive during the mid-1990s (Hurrell,1995). Consequently, the use of data sets, which endin the mid-1990s (e.g. choice of epoch 1956–1994instead of 1956–2003) results in much larger trendsin NAO index (0.38 units/decade) and in the sealevels of the eastern North Sea (Fig. 8d–f).

Secular trends in 99, 50 and 1 percentile surge (‘s’)during 1956–2003 (not shown) were broadly con-sistent with those in still water level (‘tm’). The 50percentile trends were identical to those of Fig. 8(b),as one would expect. The spatial patterns for the 99and 1 percentiles were in close agreement with their

ARTICLE IN PRESS

-15° -10° -5° 0° 5° 10°

50°

55°

60°

-0.3

0.5

0.6

-15° -10° -5° 0° 5° 10°

-0.4

0.6

0.6

0.7

0.7

0.7

-15° -10° -5° 0° 5° 10°

50°

55°

60°0.3 0.4

0.4

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

-15° -10° -5° 0° 5° 10°

50°

55°

60°

-20

30 40

50

60

60

70

-15° -10° -5° 0° 5° 10°

20

3030

30

40

-15° -10° -5° 0° 5° 10°

50°

55°

60°

10

20

20

3040

50

-100 -80 -60 -40 -20 0 20 40 60 80 100

mm/unit

a b c

fed

Fig. 6. (a) Correlation coefficient between NAO index and winter 99 percentile surge for 1956–2003; (b) corresponding coefficients for 50

percentile surge and (c) 1 percentile surge; (d) sensitivity of change in winter 99 percentile surge to change in NAO index (mm/unit index);

(e) corresponding sensitivity for 50 and (f) 1 percentile.

P.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946942

‘tm’ counterparts but with values larger by approxi-mately 25% in the areas of large trend in the NorthSea. Such differences can be anticipated as, it will berecalled, surges occur throughout the tidal cycle andthe 29 h each winter which define the 99 (or 1)percentile need not be the same for both ‘s’ and ‘tm’.

7. Discussion and conclusions

There will be differences between sea-levelchanges in the real ocean and those in the presentmodel study. In particular, a depth-averaged modelcannot simulate sea-level variations arising fromsteric (density) changes in the ocean. For example,the trends discussed above (Fig. 8) do not includesteric and other (glacial, hydrological, etc.) con-tributions which are a consequence of climatechange (Church et al., 2001). Steric and other

climate-related changes can be expected to occuron timescales longer than a typical storm surge, andtherefore to contribute approximately equally to allpercentiles (high and low water extremes andMeSL). Such a link between sea level and theNAO, via thermosteric changes, has been exploredrecently by Tsimplis et al. (2006). They concludedthat sensitivities of the order 10mm per unit indexcould arise from thermosteric fluctuations linked tothe NAO, additional to those due to winds and airpressures investigated in the present study.

The present study has shown that extreme sealevels and storm surges around the UK do exhibit adependence on the NAO, with sensitivities of theorder of several 10 s mm per unit index (Fig. 1). Thesensitivities for extreme high waters tend to belarger than those for MeSL and extreme low watersin the eastern North Sea. However, for most of the

ARTICLE IN PRESS

-15° -10° -5° 0° 5° 10°

50°

55°

60°

-0.3

0.60.6 0.7

0.7

-15° -10° -5° 0° 5° 10°

-0.4

0.6

0.6

0.7

-15° -10° -5° 0° 5° 10°

50°

55°

60°

-0.2

0.4

0.50.5

0.50.6

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

-15° -10° -5° 0° 5° 10°

50°

55°

60°

-2

-1

1

2

4

4

4

4

4

4

-5 -4 -3 -2 -1 0 1 2 3 4 5

-15° -10° -5° 0° 5° 10°

5060

60

70

70

-75 -60 -45 -30 -15 0 15 30 45 60 75

-15° -10° -5° 0° 5° 10°

50°

55°

60°

-5

10

10

15

1515

15

-50 -40 -30 -20 -10 0 10 20 30 40 50

a b c

fed

Fig. 7. (a) Correlation coefficient between the number of large surge events each winter and the NAO; (b) corresponding coefficients for

total time for surges over threshold each winter and (c) average height of surges over threshold; (d) sensitivity of the number of large surge

events to change in NAO index (number per unit index); (e) corresponding sensitivities for total time over threshold (hours per unit index)

and (f) average height over threshold (mm per unit index).

P.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946 943

model domain, and in particular for waters aroundthe UK, there are no major differences between therelationships of each parameter to the NAOaveraged over the epoch of the data set (cf.Fig. 2). Any time-dependence of such relationships,as have been studied for MSL (Yan et al., 2004;Tsimplis et al., 2005; Jevrejeva et al., 2005), willrequire the analysis of longer model data sets.

A conservative assessment would suggest sensi-tivities of extreme high water (Fig. 1(d)) or largesurges (Fig. 6(d)) of several 10 smm per unit NAOindex, with largest values on the east coast.However, this hardly amounts to major flood risk,even if the NAO index becomes more positive in the21st century as suggested by several AOGCMs(Osborn, 2004; Tsimplis et al., 2005). For compar-

ison, one may note that the standard deviation ofthe interannual variability of MSL around the UKis approximately 50mm (Woodworth et al., 1999).Consequently, we do not believe that the UK will besubject in future to significantly enhanced flood riskdue to NAO-related changes in air pressures andwinds. This finding is consistent with that for theNorth Sea by Butler et al. (2006), who employed thesame model data sets as in the present study, andwith those of other recent modelling investigations(e.g. EU PRUDENCE study, Woth et al., 2006). Ofcourse, other increased risks associated with climatechange could result from an overall MSL rise(Flather et al., 2001).

Several directions for further work can beidentified. While we do not consider the coarse

ARTICLE IN PRESS

-15° -10° -5° 0° 5° 10°

50°

55°

60°

-15° -10° -5° 0° 5° 10° -15° -10° -5° 0° 5° 10°

50°

55°

60°

-15° -10° -5° 0° 5° 10°

50°

55°

60°

-15° -10° -5° 0° 5° 10° -15° -10° -5° 0° 5° 10°

50°

55°

60°

-2.0 -1.6 -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2 1.6 2.0

mm/year

a b c

fed

Fig. 8. (a) Secular trend in winter extreme high water (99 percentile) for 1956–2003 (mm/year); (b) corresponding trend for MeSL and (c)

extreme low water (1 percentile); (d) secular trend for extreme high water for 1956–1994; (e) corresponding trend for MeSL and (f) extreme

low water.

P.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946944

resolution of the present model to impact seriouslyon our general conclusions for the NW Europeanshelf, the existence of any short spatial scale NAOdependence will be studied more effectively with anew model with an order of magnitude improve-ment in spatial resolution. This will shortly beimplemented by the Proudman OceanographicLaboratory at the Storm Tide Forecasting Serviceof the Met Office and will eventually be employed tohindcast extreme sea levels around UK coastsutilising European Centre for Medium-RangeWeather Forecasts (ECMWF) reanalysis fields.

For estimation of coastal flood risk, the compu-tation of joint probabilities of a number ofparameters is required including tide, surge, waves,river flow and precipitation. Pugh and Vassie(1980), Dixon and Tawn (1992) and Svensson andJones (2002, 2004) provide examples of jointprobability methods and applications. The specialrelationship between the NAO and extremes has so

far been focused on individual regional parameters:for example, extreme still water levels and surgelevels in the present paper, North Sea surges also byButler et al. (2006), waves by Tsimplis et al. (2005)and Wolf and Woolf (2006), severe storms byAlexander et al. (2005) and rainfall by Fowler andKilsby (2002). Tsimplis et al. (2005) went some wayin assessing the overall importance of the NAO tocoastal sea levels and by implication to a range ofUK and northern European coastal processes.However, a considerable amount of further workis required using statistically rigorous methodsbefore a full appreciation of the impact of theNAO on UK coastal flood risk can be obtained.

Acknowledgements

This paper is a contribution to the NERC FloodRisk from Extreme Events (FREE) programme.Jason Lowe (Hadley Centre) is thanked for valuable

ARTICLE IN PRESSP.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946 945

advice. Some of the figures in this paper weregenerated using the Generic Mapping Tools (Wesseland Smith 1998).

References

Alexander, L.V., Tett, S.F.B., Jonsson, T., 2005. Recent observed

changes in severe storms over the United Kingdom and

Iceland. Geophysical Research Letters 32, L13704.

Barry, R.G., Chorley, R.J., Chase, T., 2003. Atmosphere,

Weather and Climate, eighth ed. Routledge, London, 536pp.

Butler, A., Hefferman, J.E., Tawn, J.A., Flather, R.A., Hors-

burgh, K.J., 2006. Extreme value analysis of decadal

variations in storm surge elevations. Journal of Marine

Systems (in press), doi:10.1016/j.jmarsys.2006.10.006.

Church, J.A., Gregory, J.M., Huybrechts, P., Kuhn, M.,

Lambeck, K., Nhuan, M.T., Qin, D., Woodworth, P.L.,

2001. Changes in sea level. In: Houghton, J.T., Ding, Y.,

Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X.,

Maskell, K., Johnson, C.A. (Eds.), Climate Change 2001: The

Scientific Basis. Contribution of Working Group I to the

Third Assessment Report of the Intergovernmental Panel on

Climate Change. Cambridge University Press, Cambridge, p.

881pp.

Dixon, M.J., Tawn, J.A., 1992. Trends in UK extreme sea levels:

a spatial approach. Geophysical Journal International 111,

607–616.

Flather, R.A., 1981. Results from a model of the northeast

Atlantic relating to the Norwegian Coastal Current. In:

Saetre, R., Mork, M. (Eds.), The Norwegian Coastal Current,

vol. 2. University of Bergen, Bergen, pp. 427–458.

Flather, R.A., Smith, J.A., Richards, J.D., Bell, C., Blackman,

D.L., 1998. Direct estimates of extreme storm surge elevations

from a 40-year numerical model simulation and from

observations. The Global Atmosphere and Ocean System 6,

165–176.

Flather, R.A., Smith, J.A., 1998. First estimates of changes in

extreme storm surge elevations due to the doubling of CO2.

The Global Atmosphere and Ocean System 6, 193–208.

Flather, R.A., 2000. Existing operational oceanography. Coastal

Engineering 41 (1–3), 13–40.

Flather, R.A, Baker, T.F., Woodworth, P.L., Vassie, I.M.,

Blackman, D.L., 2001. Integrated effects of climate change on

coastal extreme sea levels. In: Proceedings of the 36th

DEFRA Conference of River and Coastal Engineers, Keele

University, 20–22 June 2001, pp. O3.4.1–O3.4.12.

Fowler, H.J., Kilsby, C.G., 2002. Precipitation and the North

Atlantic Oscillation: a study of climatic variability in northern

England. International Journal of Climatology 22, 843–866.

Gunther, H., Rosenthal, W., Stawarz, M., Carretero, J.C.,

Gomez, M., Lozano, I., Serrano, O., Reistad, M., 1998. The

wave climate of the northeast Atlantic over the period

1955–1994: the WASA wave hindcast. The Global Atmo-

sphere and Ocean System 6, 121–163.

Hulme, M., Jenkins, G., Lu, X., Turnpenny, J.R., Mitchell, T.D.,

Jones, R.G., Lowe, J., Murphy, J.M., Hassell, D., Boorman,

P., McDonald, R., Hill, S., 2002. Climate Change Scenarios

for the United Kingdom: The UKCIP02 Scientific Report.

Tyndall Centre for Climate Change Research, 120pp.

Hurrell, J.W., 1995. Decadal trends in the North Atlantic

Oscillation: regional temperatures and precipitation. Science

269, 676–679.

Jevrejeva, S., Moore, J.C., Woodworth, P.L., Grinsted, A., 2005.

Influence of large scale atmospheric circulation on European

sea level: results based on the wavelet transform method.

Tellus 57A, 183–193.

Jones, P.D., Jonsson, T., Wheeler, D., 1997. Extension to the

North Atlantic Oscillation using early instrumental pressure

observations from Gibraltar and south-west Iceland. Inter-

national Journal of Climatology 17, 1433–1450.

Langenberg, H., Pfizenmayer, A., von Storch, H., Sundermann,

J., 1999. Storm related sea level variations along the North

Sea coast: natural variability and anthropogenic change.

Continental Shelf Research 99, 821–842.

Lowe, J.A., Gregory, J.M., Flather, R.A., 2001. Changes in the

occurrence of storm surges around the United Kingdom

under a future climate scenario using a dynamic storm surge

model driven by the Hadley Centre climate models. Climate

Dynamics 18, 179–188.

NERC, 2005. Flood risk from extreme events (FREE), the

science of flooding. Science Plan of a Research Programme

of the UK Natural Environment Research Council. 16pp.

/http://www.nerc.ac.uk/funding/thematics/free/S.

Osborn, T.J., 2004. Simulating the winter North Atlantic

Oscillation: the roles of internal variability and greenhouse

gas forcing. Climate Dynamics 22, 605–623.

Ponte, R.M., 1994. Understanding the relation between wind-

and pressure-driven sea level variability. Journal of Geophy-

sical Research 99 (C4), 8033–8039.

Pugh, D.T., 1987. Tides, Surges and Mean Sea-level: A Hand-

book for Engineers and Scientists. Wiley, Chichester, 472pp.

Pugh, D.T., Vassie, J.M., 1980. Applications of the joint

probability method for extreme sea level computations.

Proceedings of the Institution of Civil Engineers 69 (Part 2),

959–975.

Pugh, D.T., 2004. Changing Sea Levels: Effects of Tides, Weather

and Climate. Cambridge University Press, Cambridge, 280pp.

Rodwell, M.J., Rowell, D.P., Folland, C.K., 1999. Oceanic

forcing of the wintertime North Atlantic Oscillation and

European climate. Nature 398, 320–323.

Svensson, C., Jones, D.A., 2002. Dependence between extreme

sea surge, river flow and precipitation in eastern Britain.

International Journal of Climatology 22, 1149–1168.

Svensson, C., Jones, D.A., 2004. Dependence between sea surge,

river flow and precipitation in south and west Britain.

Hydrology and Earth System Sciences 8 (5), 973–992.

Tsimplis, M.N., Woolf, D.K., Osborn, T.J., Wakelin, S., Wolf, J.,

Flather, R., Shaw, A.G.P., Woodworth, P., Challenor, P.,

Blackman, D., Pert, F., Yan, Z., Jevrejeva, S., 2005. Towards

a vulnerability assessment of the UK and northern European

coasts: the role of regional climate variability. Philosophical

Transactions of the Royal Society of London 363, 1329–1358.

Tsimplis, M.N., Shaw, A.G.P., Flather, R.A., Woolf, D.K., 2006.

The influence of the North Atlantic Oscillation on the sea-

level around the northern European coasts reconsidered: the

thermosteric effects. Philosophical Transactions of the Royal

Society of London A 364, 845–856.

Vassie, J.M., Woodworth, P.L., Holt, M.W., 2004. An example

of North Atlantic deep ocean swell impacting Ascension and

St Helena islands in the central South Atlantic. Journal of

Atmospheric and Oceanic Technology 21 (7), 1095–1103.

ARTICLE IN PRESSP.L. Woodworth et al. / Continental Shelf Research 27 (2007) 935–946946

Von Storch, H., Reichardt, H., 1997. A scenario of storm surge

statistics for the German Bight at the expected time of

doubled atmospheric carbon dioxide concentration. Journal

of Climate 10, 2653–2662.

Wakelin, S.L., Woodworth, P.L., Flather, R.A., Williams, J.A.,

2003. Sea-level dependence on the NAO over the NW

European Continental Shelf. Geophysical Research Letters

30 (7), 1403.

WASA Group, 1998. Changing waves and storms in the

Northeast Atlantic. Bulletin of the American Meteorological

Society 79, 741–760.

Wessel, P., Smith, W.H.F., 1998. New, improved version of

Generic Mapping Tools released EOS. Transactions of the

American Geophysical Union 79, 579.

Wilks, D.S., 1995. Statistical Methods in the Atmospheric

Sciences: An Introduction. San Diego, Academic Press.

Wolf, J., Woolf, D.K., 2006. Waves and climate change in the

north-east Atlantic. Geophysical Research Letters 33,

L06604.

Woodworth, P.L., Tsimplis, M.N., Flather, R.A., Shennan, I.,

1999. A review of the trends observed in British Isles mean sea

level data measured by tide gauges. Geophysical Journal

International 136, 651–670.

Woodworth, P.L., Blackman, D.L., 2004. Evidence for systema-

tic changes in extreme high waters since the mid-1970s.

Journal of Climate 17, 1190–1197.

Woolf, D.K., Shaw, A.G.P., Tsimplis, M.N., 2003. The influence

of the North Atlantic Oscillation on sea level variability in the

North Atlantic Region. The Global Atmosphere and Ocean

System 9 (4), 145–167.

Woth, K., Weisse, R., von Storch, H., 2006. Climate change and

North Sea storm surge extremes: an ensemble study of storm

surge extremes expected in a changed climate projected by

four different regional climate models. Ocean Dynamics 56

(1), 3–15.

Yan, Z., Tsimplis, M.N., Woolf, D., 2004. Analysis of the

relationship between the North Atlantic Oscillation and sea-

level changes in Northwest Europe. International Journal of

Climatology 24, 743–758.