Effects of Global Climate Oscillations on Intermonthly to Interannual Variability of Sea levels along the English Channel Coasts (NW France)

(1)

HAL Id: hal-02886424

https://hal.archives-ouvertes.fr/hal-02886424

Submitted on 16 Apr 2021

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Distributed under a Creative Commons Attribution - NoDerivatives| 4.0 International

License

Effects of Global Climate Oscillations on Intermonthly

to Interannual Variability of Sea levels along the English

Channel Coasts (NW France)

Imen Turki, Nicolas Massei, Benoît Laignel, Hassan Shafiei

To cite this version:

Imen Turki, Nicolas Massei, Benoît Laignel, Hassan Shafiei. Effects of Global Climate Oscillations on

Intermonthly to Interannual Variability of Sea levels along the English Channel Coasts (NW France).

Oceanologia, Polish Academy of Sciences, 2020, 62 (2), pp.226-242. �10.1016/j.oceano.2020.01.001�.

�hal-02886424�

(2)

Availableonlineatwww.sciencedirect.com

ScienceDirect

journal homepage:www.journals.elsevier.com/oceanologia

ORIGINAL

RESEARCH

ARTICLE

Effects

of

Global

Climate

Oscillations

on

Intermonthly

to

Interannual

Variability

of

Sea

levels

along

the

English

Channel

Coasts

(NW

France)

Imen

Turki

∗

,

Nicolas

Massei

,

Benoit

Laignel

,

Hassan

Shaﬁei

UMRCNRS6143— ContinentalandCoastalMorphodynamics‘M2C’,UniversityofRouen,Mont-Saint-AignanCedex,France

Received 25May2019;accepted20January2020 Availableonline5February2020

KEYWORDS Sealevel; Extremesurges; Multiscalevariability; Storms; Climateoscillations; Multiresolution analysis

Summary ThisworkexaminesthemultiscalevariabilityinsealevelalongtheEnglish Chan-nelcoasts(NWFrance)usingawaveletmultiresolutiondecompositionofwaterlevelvaluesand climateoscillationsinordertogaininsightsintheconnectionbetweentheglobalatmospheric circulationandthelocal-scalevariability ofthemonthlyextremesurges. Changesinsurges haveexhibiteddifferentoscillatorycomponentsfromtheintermonthly(_{∼3—6-months)}tothe interannual scales(∼1.5-years,∼2—4-years,∼5—8-years)with meanexplained variancesof ∼40%and_∼25%ofthetotal variabilityrespectively.Thecorrelationbetweenthe multireso-lutioncomponentsofsurges and28exceptionalstormyeventswithdifferentintensitieshas revealed thatenergetic eventsaremanifestedatall timescaleswhilemoderate eventsare limitedtoshortscales.

By consideringthetwohypothesesof(1)thephysicalmechanismsoftheatmospheric circu-lationchangeaccordingtothetimescalesand(2)their connectionwiththelocalvariability improvesthepredictionoftheextremes,themultiscalecomponentsofthemonthlyextreme surgeshavebeeninvestigatedusingfourdifferentclimateoscillations(SeaSurface Tempera-ture(SST),Sea-LevelPressure(SLP),ZonalWind(ZW),andNorthAtlanticOscillation(NAO)); resultsshowstatisticallysigniﬁcantcorrelationswith_{∼3—6-months,}_{∼1.5-years,}_{∼2—4-years,} and ∼5—8-years, respectively. Such physical links, from global to local scales, have been

∗ _{Corresponding} _author_at:_Imen_Turki,_UMR_CNRS ₆₁₄₃_{— Continental}_and_Coastal _{Morphodynamics}_‘M2C’,_University_of_Rouen,₇₆₈₂₁

Mont-Saint-AignanCedex,France.

E-mailaddresses:imen.turki@univ-rouen.fr(I.Turki),nicolas.massei@univ-rouen.fr(N.Massei),benoit.laignel@univ-rouen.fr

(B.Laignel),hassan.shaﬁei@univ-rouen.fr(H.Shaﬁei).

PeerreviewundertheresponsibilityofInstituteofOceanologyofthePolishAcademyofSciences.

https://doi.org/10.1016/j.oceano.2020.01.001

0078-3234/© 2020InstituteofOceanologyofthePolishAcademyofSciences.ProductionandhostingbyElsevierB.V.Thisisanopenaccess articleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

(3)

I.Turkietal./Oceanologia62(2020)226—242 227

consideredtomodelthemultiscalemonthlyextremesurgesusing atime-dependent Gener-alizedExtremeValue(GEV)distribution.Theincorporationoftheclimateinformationinthe GEVparametershasconsiderablyimprovedtheﬁttingofthedifferenttimescalesofsurgeswith anexplainedvariancehigherthan30%.Thisimprovementexhibitstheirnonlinearrelationship withthelarge-scaleatmosphericcirculation.

© 2020 Institute of Oceanology of the Polish Academy of Sciences. Production and host-ing by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Withthepresentcontextofglobalclimatechange,mostof researchersclaimthatthesea-levelvariabilityandthe in-creaseofstorminessareconsideredsignificanthazardsfor several low-lyingcoastalcommunities(e.g.Hanson et al., 2011 ;Nicholls et al., 2011 ).Hence,manyeffortshavebeen devoted to better understand the natural processes driv-ing the multiscale variability of the extreme sea levels to produce a more accurate estimation of their fluctua-tions and ensuring reliable coastal risk assessments. This challenge serves asthe basis for implementing an appro-priate adaptation strategyreducing thedisastrousrisksof the coastalflooding.Thesea-levelchanges,encompassing cyclical seasonal components superimposed on long-term trendsandstochasticvariability,areoneofthemost impor-tantphysicalprocessesaffectingthecoastalsystems.These changes areduetothefrequent temporalshifts linkedto the nonlinear,stochastic, or transient effects of external factorssuchastheglobalclimatepatterns(Pasquini et al., 2008 ).

Changesinsealevelsarelargelydrivenbythelarge-scale atmospheric andoceaniccirculation patternsoperatingon intermonthly andinterannualtimescales.Moreover, study-ing the links of the sea-level variabilitywith the climate teleconnections,related toprominentatmospheric modes asproxies,isnecessarytofullyunderstandtheinterplay be-tween theclimateoscillationsandtheoceanographic pro-cesses(e.g.Dangendorf et al., 2012 ;Levin, 1992 ;Yan, 2004 ;

Zampieri et al., 2017 ).Someworkshavebeenimplemented to understand and extract the signature of internal cli-matevariabilityfromtheobservedsea-levelpatterns(e.g.

Dangendorf et al., 2012 ;Levin, 1992 ;Yan, 2004 ;Zampieri et al., 2017 )withtheaimofcontributinginthemultiscale sea-levelpredictionsthatemergeasurgentprioritiesinthe state-of-the-artclimateresearch.

Large timescale studies around the English Channel (South coast of UK) have been presented in Haigh et al. (2009) work, where changes in the mean sealevels have beeninvestigatedbytheuseofhourlysea-levelrecords ex-tended by data archaeology. They have found that mean sea-leveltrendsvarybetween0.8and2.3mm/yeararound theChannel.Anewapproachofspectralanalysishasbeen usedby Turki et al. (2015a) toinvestigate theannual and theintermonthlychangesofthesealevelalongtheEnglish Channelcoasts(NWFrance).Suchmultiscalechangeshave been linkedtoa combinationofmeteorological andNorth AtlanticOscillation(NAO)climatedrivers,involvingseveral processesoverawiderangeofspatialandtemporalscales.

Beinganalarmingproblemforthecoastalvulnerability, extremeeventshavegainedtheattentionofthescientists whohave reportedthedynamics(e.g.,Haigh et al., 2010 ;

Idier et al., 2012 ;Masina and Lamberti, 2013 ;Tomasin and Pirazzoli, 2008 ;Turki et al., 2019 )andtheprojections(e.g.,

Vousdoukas et al., 2017 ) of extreme sea levels consider-ingthestationaryandthenonstationarycontributionsfrom tides,waves,sea-level-risecomponents(e.g.,Brown et al., 2010 ;Idier et al., 2017 ),andlarge-scaleclimateoscillations (e.g.,Colberg et al., 2019 ;Turki et al., 2019 ).

Anumberofstudieshaveexaminedtherelationship be-tween the global atmospheric patterns and the local sea levelstoestimate sea-levelextreme values by theuse of nonstationarystatisticalapproaches.Regionalstudiesof ex-tremesealevelshavebeenconductedinvariousworks us-ingdifferentmethodologies;themost commonofthem is relatedtotheGeneralizedExtremeValue(GEV)models.A reviewofextremeanalyseshasbeenperformedbyMendez and Woodworth (2010) ; they applied a nonstationary ex-tremevaluemodeltothemonthlymaximawithaspecial fo-cusonNAOtostudythevariabilityoftheextremesealevels alongtheEuropeancoasts.Masina and Lamberti (2013) have usedanonstationaryGEVmodeltodemonstrateacoherent behaviorbetweentheregionalandtheglobalscalesfroma detailedanalysisoftheannualmeansea-levelevolutionin theAdriaticseawiththeNAOandAO(ArcticOscillation) in-dices;theyhavesuggestedthattheincreaseintheextreme water levels sincethe 1990s is related to the changes in thewindregimeandtheintensiﬁcationofBoraandSirocco windsafterthesecondhalfofthe20thcentury.

IntheEnglishChannel,theextremesealevelshavebeen addressed by severalworks(e.g. Haigh et al., 2010 ;Idier et al., 2012 ;Tomasin and Pirazzoli, 2008 ;Turki et al., 2019 ) with the aim of investigating their dynamicsat different timescalesandtheirconnectionstotheatmospheric circu-lationpatterns.

Haigh et al. (2010) investigatedtheinterannualandthe interdecadal extreme surges in the English Channel and theirstrongrelationship withtheNAOindex.Theirresults showed weak negative correlations throughout the Chan-nelandstrongpositive correlationsatthe boundaryalong theSouthernNorthSea.Usinganumericalapproach,Idier et al. (2012) studiedthespatialevolutionofsomehistorical stormsintheAtlanticSeaandtheirdependenceontides.

Recently, Turki et al. (2019) have examined the multi-scale variabilityof thesea-level changesin theSeine Bay (NWFrance)inrelationwiththeglobalclimateoscillations from the SLP composites; they have demonstrated dipo-lar patterns of high-low pressures suggesting positive and

(4)

negativeanomaliesattheinterdecadalandtheinterannual scales,respectively.

Despitetheseimportantadvances,noparticularstudies existon sea-leveldynamics andextreme events linked to thelarge-scaleclimateoscillationsalongtheEnglish Chan-nel coastlines. The aforementioned works of Turki et al. (2015a ,2019 )havefocusedonthemultiscalesea-level vari-abilityalongtheFrenchcoastsrelatedtotheNAOandthe Sea-LevelPressure(SLP)patterns; however,theyhavenot addressed theregionalbehavior oftheextreme sealevels inrelationwiththeglobalclimateoscillations.

The present contribution aims to investigate the re-gionaldynamicsofthesealevelandtheextremeeventsin someselectedEnglishChannelcoasts(NWFrance)andseeks to understand their multiscale connection to the large-scaleatmosphericcirculation.Aspectralapproachbasedon thecontinuousandmulti-resolutionwavelettechniqueshas beenusedhereininorderto(i)visualizeandinvestigatethe spectralcontent of the sealevels onintermonthly tothe interannualscales,(ii)identifythelinksbetweenthe sea-level ﬂuctuationsand theincrease of thehistorical storm events,(iii)quantifythemultiscalechangesinthemonthly maximaofextremesurgesandtheirconnectionwiththe at-mosphericcirculationbyapplyinganonstationaryapproach usingtheclimateoscillations.

The rest of the paper is organized into four sections:

Section 2 gives a description of thedata sourcesand the methodologicalapproachusedtoinvestigatethemultiscale variabilityofthesealevelwiththeatmosphericcirculation. Local-scale changes in sea levels along the English Chan-nelandtheirconnectionwiththeclimateoscillationshave been addressedanddiscussed insections 3 and4 , respec-tively.Finally,allﬁndingsaresummarizedandconcludedin

Section 5 .

2. Data

and

methods

2.1. Sea-levelandclimatedatasets

The present research is focused on the Normandy coasts along the Southern side of the English Channel (North of France); it has been conducted in the framework of the French researchprogram ANR RICOCHETand the interna-tionalproject COTEST fundedby CNES-TOSCAandrelated tothefuturemissionSurfaceWaterandOceanTopography (SWOT).

Sea-leveldata,obtainedfromthreetidegaugesites,are usedinthepresentstudy:(1) Dunkirkstationwhichisafew kilometersawayfromBelgianborders,(2 )LeHavrestations situatedontherightbankoftheestuaryoftheSeineRiver, and(3 )CherbourgstationlocatedontheCotentinPeninsula andattheopeningoftheAtlanticSea.Thesestations pro-videtime-series ofhourlyobservationsmeasured between 1964 and 2010. The tide gauges are operated and main-tainedbytheNationalFrenchCenterofOceanographicData (SHOM).The observations whichcorrespond tothe hydro-graphiczerolevelarereferencedtozerotidegauge(Fig. 1 ). The storm events produced inthe EnglishChannel largely occurred along Cherbourg, Le Havre, and Dunkirk coasts. The surgesin these regions are equal or higher than the

valuesof2yearreturnperiod(Re)whichrepresents1.2m, 1.3m,and1.6m,respectively(SHOM).

Thelarge-scaleatmosphericcirculationsarerepresented in the present analysis by four different climate indices which are considered as fundamental drivers in the At-lanticregions:theSeaSurfaceTemperature(SST),theNorth Atlantic Oscillation (NAO), the Zonal Wind (ZW) compo-nentextractedat850hPa,andtheSea-LevelPressure(SLP). Monthlytime-seriesofclimateindexhavebeenprovidedby theNCEP-NCARReanalysisﬁelds1_with_the_same_period

dur-ingwhich thesea-levelobservationswereconducted (i.e. 1964—2010).

2.2. Methods

Thetotalsea-levelheight,resultingfromtheastronomical andthemeteorologicalprocesses,exhibitsatemporal non-stationaritywhichisexplainedbyacombinationofthe ef-fects of the long-term trends in the mean sea level, the modulationby the deterministic tidalcomponent and the stochastic signal of surges, and the interactions between tides and surges. The occurrence of extreme sea levels is controlled by periods of astronomically-generated high tides,inparticular,atinter-annualscalewhentwo phenom-enaofprecessioncausesystematicvariationofhightides. The modulation of the tides contributesto the enhanced riskofcoastalﬂooding.Therefore,theseparationbetween tidalandnon-tidalsignalisanimportanttaskinanyanalysis ofsea-leveltime-series.

Bythe hypothesis of independence between the astro-nomicaltidesandthestochasticresidualofsurges,the non-linearrelationshipbetweenthetidalmodulationandsurges isnotconsideredinthepresentanalysis.Usingtheclassical harmonicanalysis,thetidalcomponenthasbeen modeled asthe sumof afinitesetof sinusoids atspecific frequen-ciestodeterminethedeterministphase/amplitudeofeach sinusoidandpredict theastronomicalcomponentoftides. In order to obtain a quantitative assessment of the non-tidalcontributioninstorminesschanges,technicalmethods based onMATLAB t-tide package have been used to esti-mateyear-by-year tidalconstituents. Ayear-by-year tidal simulation(Shaw and Tsimplis, 2010 ) hasbeen appliedto thesea-level time-series todetermine the amplitude and thephaseoftidalmodulationsusingharmonicanalysis fit-tedto18.61-,9.305-,8.85-,and4.425-yearsinusoidal sig-nals(Pugh, 1987 ). The radiational componentshave been alsoconsideredfortheextractionofthestochastic compo-nentofsurges(Williams et al., 2018 ).Thehourlysea-level measurementspresentsomeshortgaps(ofsomedays) dis-tributedduringthefirst30years(1964—1995)ofthetotal time-series. These gaps, withrespect tothe total series, represent2%inCherbourgandLeHavreand5%inDunkirk; theyhave been processed by the hybrid model for filling gapsdeveloped byTurki et al. (2015b) .The modelapplies apurely statistical approachtothe stochastic component ofthesealevel(residualcomponent)basedon Autoregres-siveMovingAverage(ARMA)techniqueswithintroducingSLP in ARMA as a main physical process driving the residual

1_{http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.} reanalysis.derived.html

(5)

Figure1 GeographicallocationofthestudyareaandthedifferenttidegaugesalongtheSouthernEnglishChannelcoasts(NW France):Dunkirk,LeHavre,andCherbourg.

Figure2 Thehourlysurgesduringtheperiod1964—2010in(a)Dunkirk,(b)LeHavre,and(c)Cherbourg.

sea level. They have used this model toprocess the sea-levelrecordsprovidedbythetidegaugesalongtheAtlantic Frenchcoasts.ThemodelhasbeenusedalsobyTurki et al. (2019) fordataprocessing.Thetotalsealevelandsurgesat Dunkirk,LeHavre,andCherbourgareillustratedinFig. 2 .

The general approachdeveloped herein intendsto: (1) identifythemain timescalesofthevariabilityof thetotal sealevelsandsurges,(2)characterize themultiscale rela-tionships betweenthelocal-scalehydro-climatological sig-nal of thesurgesand thehistorical stormevents onlarge scales,(3)examinetheabilityofthepotentialmonthly cli-matepredictors(NAO, ZW,SLP,SST)todescribethe time-scaledecompositionofthemonthlyextremesurgesby iden-tifyingthephysicallinkbetweenbothvariablesateachtime

scale.Suchphysicallinkhasbeenusedforamultiscale pa-rameterization of the non-stationary GEV models. Firstly, thehourlyhydro-climatologicalsignals(i.e.totalsealevel and surges) have been analyzed using continuouswavelet transform(CWT)toexplorethespectralcontentof oceano-graphicsignals.Thetypicalscalesofthesea-level variabil-ity(Turki et al., 2015a ,2019 )havebeendetectedfrom dif-ferentrecords.

The CWT is a well-known method that has been used over the past decade for data analysis in hydrology, geo-physics, and environmental sciences (Labat, 2005 ; Sang, 2013 ;Torrence and Compo, 1998 ).Thecontinuouswavelet transformproduceseitheratime-scaleortime-periodwith the means of the Fourier transform contour diagram on

(6)

which timeis indicated onthex-axis, period,or scale on they-axis,andamplitude(or variance,or power,etc.) on thez-axis.

Secondly, the hourly surges have been decomposed by waveletmultiresolutionanalysis intodifferent internal componentscorrespondingtodifferenttime-scales.Briefly, themultiresolutionanalysisconsistsof theiterative filter-ingof thetime-series usingaseriesof low-passand high-pass filters which eventually produce one high-frequency “rough” componentcalled“waveletdetail” andonelower frequency“coarse” componentcalled“smooth” or “approx-imation”.

The smooth, produced at each scale, is then subse-quentlydecomposedintoasecondwaveletdetailanda sec-ondsmooth,thelatterisdecomposedagaininthesameway untilonelastsmoothremainsandwillnolongerbe decom-posed.Summingupallwaveletdetailsandthelastsmooth (i.e.thelowest-frequencycomponent)givesbackthe orig-inalsignal.Insummary,thetotalsignalhasbeenseparated intoarelativelysmallnumberofwaveletcomponentsfrom high tolow frequencies that altogetherexplains the vari-abilityofthesignal;thiswillbeillustratedlaterusingthe hourlymeasurementsandthemonthlymaximaofsurges.

Finally andwith the aim of addressing the nonstation-arybehaviorofextremesurges,themonthlymaximaofthe surgeshavebeencalculatedanddecomposedwiththe mul-tiresolution analysis.Then,a nonstationaryextreme value analysisbasedontheGEVdistributionwithtime-dependent parameters(Coles, 2001 )has beenimplementedtomodel theseriesofthemonthlymaximasurges.Thereareseveral GEV families which depend onthe shape parameter, e.g. Weibull(ɛ< 0),Gumbel (ɛ= 0),andFréchet(ɛ>0).The threeparametersoftheGEV(i.e.locationμ,scaleψ,shape ɛ)areestimatedbythemaximumlikelihoodfunction.

The nonstationaryeffectwasconsideredby incorporat-ingtheselectedclimateindices(NAO,SST,ZW,andSLP)into theparametrizationoftheGEVmodels.AkaikeInformation Criterion (AIC)hasbeen usedtoselectthemost appropri-ateprobabilityfunctionmodels.Themethodsofmaximum likelihoodwereusedfortheestimationofthedistributions parameters.Theapproachusedconsidersthelocation(μ),

thescale (ψ),andtheshape(ɛ)parameterswithrelevant covariates, whichare described by aselected climate in-dex:

μ(t)=β0,μ+β1,μY1+...+βn,μYn, (1)

ψ(t)=β0,ψ+β1,ψY1+...+βn,ψYn, (2)

ε(t)=β0,ε+β1,εY1+...+βn,εYn, (3)

where β0,β1,…,βn arethe coefﬁcients,andYi is the

co-variaterepresentedbytheclimateindex.Foreachspectral component,onlyoneclimateindexcanbeusedamongthe parameters μ, ψ,and ɛof the nonstationary GEV model. Withtheaimofoptimizingthebestuseoftheclimate in-dexintothedifferentGEVparameters,aseriesof sensitiv-ityanalyseswereimplementedforeachtimescale.TheAIC measures the goodness of ﬁtof the model (Akaike, 1974 ) totherelationAIC=—2l+2K;wherelisthelog-likelihood valueestimatedfortheﬁttedmodel,andKisthenumberof themodelparameters.Higherrankedmodelsshouldresult fromlowerAICscores.

3. Results

3.1. Themultiscalesea-levelvariability

Thevariabilityofthetotalsealevel(SL)andthesurges(S) hasbeen investigatedusingthecontinuouswavelet trans-form(CWT).InthespectrumofFig. 3 ,thecolorscale repre-sentsanincreasingpower(variance)frombluetored.The CWTdiagramshighlight theexistence ofseveral modesof variability;eachcolor representsoneof theenergybands with certain ranges of frequencies. The annual mode is clearlyillustrated forSLwhileitis largelydissipatedforS spectrumwithadiminutionhigherthan90%inDunkirkand Cherbourg.Thedissipationoftheannualspectrumbecomes lessimportant(withaspectrumdiminutionof72%)forSin LeHavreasitgetsclosertotheSeineBay.wherethe hydro-logicalsignature,induced bytheSeine River,stillremains tobeobservedwithinthestochasticcomponentofthesea level.

At the interannual scales, the CWT diagrams highlight twomodesof∼2—4-yrand∼5—8-yratallthestations, par-ticularlyCherbourgandLeHavre.Suchlowfrequenciesare well-structuredwith45%and65%oftheexplainedvariance higherthanthosecalculatedfromtheCWTofSL.TheCWT diagramsofSdisplayanewmodeof∼1.5-yrwhilethe an-nualmode, observed in the SL diagrams, hasbeen disap-peared.

The multiresolution analysis has been applied to local surgeswiththeaimofachievingthefulltimescale decom-position of the signal. The process results in separation of different components or wavelet details for each sig-nal(Figs. 4—6 ).Wefocusedonlyonfrequenciesranging be-tween∼3-monthsand∼5—8-years,whoseﬂuctuations cor-respondtotheoscillationperiodslessthanhalfthelength oftherecordandexhibitahigh-energycontributiononthe varianceofthetotalsignal.

Resultshavebeenexploredtoinvestigatethedynamics ofsurgesatdifferenttimescales. Thevariabilityis clearly dominatedby thehigh-frequenciesof∼3-months and ∼6-months,explainingavariancebetween35%and45%ofthe totalenergyrespectively(Table 1 ).Then,thelowfrequency components of ∼1.5-yr, ∼2—4-yr, and ∼5—8-yr explain a meanvarianceof25%ofthetotalenergy(Table 1 ).

Figs. 4—6 show a series of oscillatory components of surgesfromintermonthlytointerannualscales,notclearly identiﬁedbyasimplevisualinspectionofthesignal. Simi-laritiesofsurgecomponentshavebeenhighlyobservedfor theinterannualmodesof∼5—8-yrand∼2—4-yrwhilethey seemtobesigniﬁcantlylessimportantforhighfrequencies of∼3-monthsand∼6-months.Thisresultsuggeststhat dif-ferentphysicalphenomenashouldbeconsideredtoexplain thelocalvariabilityofsurges(∼3-monthsand∼6-months); theyareinduced bycombiningtheeffects of meteorolog-icaland oceanographic forces including changes in atmo-sphericpressures andwindvelocitiesin shallowwater ar-eas.Beyondthe∼1.5-yrtimescale,thelarge-scalepatterns ofsurges,describedbythelowfrequencycomponents,tend tobequitesimilarbetweenthedifferentsitesintermsof intensity andamplitude. Suchvariability exhibitsa global contributionofsomephysicalprocessesrelatedtothe cli-mate oscillations. The extent of these oscillations is not

(7)

Figure3 Continuouswavelettransform(CWT)ofthetotalandthenon—tidal(surges)sealevelat(a)Dunkirk,(b)LeHavreand (c)Cherbourgduringtheperiod1964—2010.

Table1 The explainedvariance expressedaspercentageoftotal varianceofsurges for allsites:Dunkirk,Le Havreand Cherbourg.

∼ 3months _{∼ 6}months _{∼ 1.5-}y _{∼ 2—4}-y _{∼ 5—8-y}

Dunkirk 20% 15.6% 12.8% 8.6% 5.3%

LeHavre 27.7% 18.2% 14% 8.8% 5.7%

(8)

Figure4 Waveletdetails(components)resultingfrommultiresolutionanalysisofsurgesfrom1964to2012attheintermonthly (_∼3months,_{∼ 6}months)andinterannual(_∼1.5-yr,_∼2—4-yr,_∼5—8-yr)timescalesinDunkirk.Theblack,greyandwhiteboxesare usedforstormsmanifestedsimilarlyinLeHavreandCherbourg(samereturnperiodobserved),similarlyinCherbourgandDunkirk andsimilarlyforallsites,respectively.

Figure5 Waveletdetails(components)resultingfrommultiresolutionanalysisofsurgesfrom1964to2012attheintermonthly (∼3months,∼6months)andinterannual(∼1.5-yr,∼2—4-yr,∼5—8-yr)timescalesinLeHavre.Theblack,greyandwhiteboxesare usedforstormsmanifestedsimilarlyinLeHavreandCherbourg(samereturnperiodobserved),similarlyinCherbourgandDunkirk andsimilarlyforallsites,respectively.

(9)

Figure6 Waveletdetails(components)resultingfrommultiresolutionanalysisofsurgesfrom1964to2012attheintermonthly (_∼3months,_∼6months)andinterannual(_∼1.5-yr,_∼2—4-yr,_∼5—8-yr)timescalesinCherbourg.Theblack,greyandwhiteboxes areusedfor stormsmanifestedsimilarlyinLe HavreandCherbourg(same returnperiod observed), similarlyinCherbourgand Dunkirkandsimilarlyforallsites,respectively.

strictlysimilarandchangesaccordingtothetimescale vari-ability since the dynamicsof surgesis not necessarily re-latedtothesametypeofatmosphericcirculationprocess. Thisrelationshipwillbeaddressedlaterinthesecondpart ofthissection.

A totalof 28historical stormyevents (fromT1 toT28),

produced in theEnglish Channel during the period1964— 2010 and withsurgeshigher than2-yr returnperiodlevel (Re), havebeenextractedfromSHOMdatabase(Table 2 ). The historical events havebeen attachedtothe different spectralcomponentsofsurges(Figs. 4—6 )byverticalbars withdifferentcolors(black,greyandwhite).

Intheseﬁgures,wedistinguishthreetypesofevents:(1 ) the ﬁrst one, depicted by the black bars, representsthe stormsoccurringinLeHavreandCherbourgwiththesame Re;(2 )thesecond one,depictedbythewhitebars,shows thestormssimilarlymanifestedinCherbourg,LeHavreand Dunkirk;(3 )thethirdone,depictedbythegreybars, illus-tratesthestormsoccurringinLeHavreandCherbourgwith thesameRe.Thepercentagesofeachofthebarsare21.4, 57.2,and21.4%,respectively.

ThestormsinFigs. 4—6 showthatthesignatureofeach eventisfullyidentiﬁedattheintermonthlyscaleswhileit ismanifesteddifferentlyattheinterannualscales.

TheeventsofNovember1965(T1)andJanuary1968(T2),

producedatneaptides,arehighlightedbythepeakofthe surgesattheinterannualscales;theyarelimitedto∼1.5-yr and∼2—4-yrinLeHavreandCherbourgwhereReisof2—5 years.ThestormsT3(July1969)andT4(February1974)

ex-hibitahighenergyformostofthecomponentsinLeHavre (Reof 5—10years),asshown inFig. 5 ;furthermore,their effects are limited to ∼1.5-yr component in Dunkirk and

Cherbourg. As documented by Bessemoulin (2002) , these storms (in particular T3) were resulted from the big

at-mospheric depression Ex-Bertha withhigh windvelocities (morethan150km/h)andbarometricgradients(maximum of14hPa).T5ismanifestedindifferentlocationswithReof

2-years;itsimpactisrestrictedtothe∼1.5-yrcomponentof surges.Similarbehaviorhasbeen identiﬁedfor theevents T9—12,T15—16,andT18producedatneaptides.TheeventsT6

(January1978),T13(February1989),andT14(January1990,

‘DARIA’)coincide withspring tidesand highreturnperiod (Re>5years);theyhavebeenrecordedattheinterannual scalesin LeHavreandCherbourg.According tothe previ-ousworksofBessoumoulin (2002) andPirazzoli et al. (2005) , theseeventswereinducedbycoldfrontsknownasKatasplit frontresponsiblefordecreasingofthepressureandashift ofthewindfromSSWtoWSW.

The events T7, T8, andT17 were produced during

win-ter periods at the equinoxes of spring tides with Re of 5—10 years; they are manifested at the full interannual timescales(from∼1.5-yrto∼5—8-yr)inLeHavreand Cher-bourg. Such events were generated by two huge anticy-cloneswithlargepressuregradientsfromtheNorthtothe West: theﬁrst onewasresulting fromthe NWwinds; the secondonewascomingfromtheAtlantictoreachtheNorth Sea andthe Scandinavian countries, responsible for more than70% of thesubmersionphenomenain theNWFrance (Costa et al., 2004 ).

ThetwostormsofT19(December1999)andT20

(Decem-ber 2001)highlighted aclear signaturelimited to∼1.5-yr inDunkirkandCherbourgwhiletheyaremanifestedat dif-ferentinterannualscalesthanLeHavre,closetotheSeine Bay,where the inﬂuence of the hydrological variability is

(10)

Table2 List ofstorminesswith asurge returnperiod (Re)higherthan2years andproducedbetween 1964and2010in Dunkirk,LeHavre,andCherbourg.Theblack,grey,andwhiteboxesareusedforstormsmanifestedsimilarlyinLeHavreand Cherbourg(samereturnperiodobserved),similarlyinCherbourgandDunkirkandsimilarlyforallsites,respectively.

Code Date Surge return period Tide cycle

T1 November, 1965 H5–H10 (Dunkirk)

H2–H5 (Le Havre, Cherbourg)

Neap de

T2 January, 1968 H5–H10 (Dunkirk)

Neap de

T3 July, 1969 H2 (Dunkirk, Cherbourg)

H5–H10 (Le Havre)

Neap de

T4 February, 1974 H2 (Dunkirk, Cherbourg)

H2–H5 (Le Havre)

Spring de

T5 December, 1976 H2 (Dunkirk, Le Havre, Cherbourg) Neap de

T6 January, 1978 H5–H10 (Dunkirk)

H5 (Le Havre, Cherbourg)

Spring de (equinoxes) T7 December, 1979 H5–H10 (Le Havre, Cherbourg)

H5 (Dunkirk)

Spring de (equinoxes) T8 December, 1981 H5–H10 (Le Havre, Cherbourg)

H5 (Dunkirk)

Spring de

T9 December, 1982 H2 (Dunkirk, Le Havre, Cherbourg) Neap de

T10 January, 1984 H2 (Dunkirk, Le Havre, Cherbourg) Neap de

T11 March, 1986 H2 (Dunkirk, Le Havre, Cherbourg) Neap de

T12 October, 1987 H2 (Dunkirk, Le Havre, Cherbourg) Neap de

T13 February, 1989 H5–H10 (Cherbourg, Dunkirk) H2–H5 (Le Havre)

Spring de

T14 January, 1990

(DARIA)

H5–H10 (Dunkirk, Le Havre, Cherbourg) Neap de

T15 May, 1992 H2 (Dunkirk, Le Havre, Cherbourg) Neap de

T16 September, 1993 H2 (Dunkirk, Le Havre, Cherbourg)

T17 December, 1995 H5–H10 (Dunkirk, Le Havre, Cherbourg) Spring de (equinoxes)

T18 October, 1998 H2 (Dunkirk, Le Havre, Cherbourg) Neap de

T19 December, 1999 H2 (Dunkirk, Cherbourg)

H5–H10 (Le Havre)

Spring de

T20 December, 2001 H2 (Dunkirk, Cherbourg)

H5–H10 (Le Havre)

Spring de

T21 February, 2004 H5–H10 (Dunkirk, Cherbourg) H5 (Le Havre)

Neap de T22 April, 2005 H2–H5 (Dunkirk, Le Havre, Cherbourg) Spring de T23 March, 2006 H2–H5 (Dunkirk, Le Havre, Cherbourg) Spring de (equinoxes)

T24 November, 2007 H2 (Dunkirk, Le Havre, Cherbourg) Neap de

T25 March, 2008 H5–H10 (Dunkirk, Le Havre, Cherbourg) Spring de (equinoxes) T26 January, 2009 H2–H5 (Dunkirk, Le Havre, Cherbourg) Neap de T27 February, 2010 H5–H10 (Dunkirk, Le Havre, Cherbourg) Spring de

T28 December, 2010 H2–H5 (Dunkirk)

(11)

I.Turkietal./Oceanologia62(2020)226—242 235 importantduringtheﬂoodingeventsbetween20thcentury

andthebeginningof21st_century₍_{Massei et al., 2017}_).

T21 (February 2004) was produced at neap tides with

strongwindsof140km/h(SHOM, 2014 );thiseventismainly carriedoutbytheinterannualcomponents.T22 andT23(Re

between2and5years),wererecordedforthesitesatthe scales of∼1.5-yrand∼2—4-yr whileT24 (Reof2 years)is

limitedto∼1.5-yr.The lasteventsT25,T27,andT28 (Re>

5 years)occurred duringthewinterperiodstheequinoxes ofspringtides,aremanifestedatalltimescalesinLeHavre andCherbourgandseemtobelimitedto∼4—8-yr compo-nentinDunkirk.

Accordingly,themultiscalevariabilityofthesurges,asa response toextreme events,highlights that14%of storms aremanifestedatalltheinterannualtimescaleswhile32% of themarelimitedto∼2—4-yr. Thiscorrespondence sug-gests the relationship between the return period of the stormsandthemultiscalevariabilityofthesurges; storms withRe<5yearsaremanifestedattimescalessmallerthan 5yearswhilestormswithhighReof5—10yearsshouldbe observedfromtheintermonthlytotheinterannualscalesof ∼5—8-yr.

Thisdiagnosisofthehistoricaleventsandtheir manifes-tationalongtheNorthFrenchcoasts(FromDunkirkto Cher-bourg)allowsthefollowingclassiﬁcation:(i)theevents pro-ducedatneaptidesandsimilarlyaffectingthestudiedsites of theNormandycoastsat theinterannualscalesof ∼1.5-yr(T5,T9-12,T15—16,andT18)and∼2—4-yr(T22andT23);(ii)

theeventsinducedbyKatasplitfontunderspringtidesand evolvingsimilarlyinLeHavreandCherbourgandwithhigher impactin Dunkirk(T6 andT13—14);(iii) theevents induced

byacombinationoftwoanticycloneseffects,fromthe At-lantic to the North Sea, and affecting similarly Le Havre andCherbourgandwithalessremarkableimpactinDunkirk (T17 andT7-8);(vi)theevents producedatspringtidesand

highlyaffectingLeHavreatdifferenttimescalescompared toDunkirkandCherbourgwheretheirmanifestationis lim-itedto∼1.5-yr(T3—4andT19—21).

Thedistributionofthestormsisnothomogenousintime; moreover, their manifestation,according todifferent cat-egories of returnperiod,takes anonlinearbehavior since the number of theevents in a windowof one year is not thesamethroughoutthetime.Indeed,allsignificantevents (Re > 2 years) have been identified during only 20 years of measurements from 46 years, the total period of the presentstudy(1964—2010).Thesedynamicsexplainthe al-teringphasesofhigh(Re>2years)andmoderate(Re< 2 years)stormsinaccordancewiththeirreturnperiod. Mod-erate phases arelonger with3—4 successive yearsduring thefirst35years(1964—2000)anddecreaseto2yearsafter 2000.Thephasesofhighstormsexhibitdifferentcategories of events: 75% of them are produced between November andFebruaryand25%ofthemareproducedbetweenMarch andOctober.

Hence, the distribution of storminess is controlled by a series of seasonal patternsrelated tochanges in ocean waveclimateandenergyconditions.Theseseasonal oscilla-tionsareclearlyidentiﬁedattheintermonthlysurges(∼3— 6-months)forallsites.Atthisscale,similaroscillationshave beendemonstratedbyTsimplis and Woodworth (1994) from the analysis ofthe mean sealevels.According tothe ob-tained results,thestormyevents,produced underspeciﬁc

meteorologicalandoceanographicconditionsontides,have exhibitedastrongseasonaldependenceonthetideswhich underline acomplexnetwork induced bya nonlinear con-tributionof manyphysical processes.The hydrological ef-fectshouldbealsoconsideredinthepresentcontextofthe SeineBaywheretheriverdischargecontributesinthetotal variabilityofthestochasticcomponent,inparticularduring theextremeeventsoftheﬂoodingproducedbetween1999 and2001.

Haigh el al. (2010) usedEnglishChannelsea-leveldataset from18tidegaugestoevaluatethechangesintheextremes throughout the20 thcentury.Theyhave identiﬁed differ-ent intra-and inter-decadal variabilityin surges withthe strongest intensity on the late 1950s. Their studies have shown strong relationship between the storm surges and theNAOindexwithweaknegativecorrelationsthroughout theChannelandstrongpositivecorrelationsatthe bound-aryalongtheSouthernNorthSea.Theinteractionsoftides withsurgeshavealsobeeninvestigatedbythesameauthors showingtheirincreaseeastwardalongtheEnglishChannel. The tide-surge interaction has been also studied by Idier et al. (2012) usinganumericalapproachtocomputesome historicalstormsintheAtlanticandtheNorthSea. Accord-ingtotheiranalyses,thetide-surge interactioninthe En-glish Channelissigniﬁcantly important;itincreasesinthe easternhalfoftheEnglishChannelanddecreasesin west-erlydirection.Inthepresentresearch,thestochasticsignal of surgeshasbeen individually studiedanditsinteraction withtideshasnottakenintoaccount.Theneap-springtidal cycle has been considered to investigate thedynamics of thestormyeventsalongtheFrenchcoastsandtheir signa-turesindifferentintermonthlyandinterannualcomponents oftheextremesurges.

3.2. Howdoesthevariabilityinextremesurges dependontheglobalatmosphericcirculation?

Thispartfocusesontheconnectionbetweenthemultiscale variabilityof thelocalsurgesalongthe Frenchcoastsand theglobalclimateoscillations.

The monthly extrema have been simulated from the hourly records of surges in the considered sites. The monthlydistributionoftheextremevalueshasbeen calcu-latedforeach scalefromtheintermonthlytothe interan-nualmodesandillustratedbythemeansofboxplot presen-tations(exampleforDunkirkinFig. 7 ).Thesedistributions exhibitthat themaximum surgeschange accordingtothe timescale;themultiscaleevolutionshowsanoscillatory be-haviorfortheshort-scales(intermonthlyandtheﬁrst inter-annualof ∼1.5-yr)shifting betweenhighand lowchanges inAprilandSeptember.Thisshiftismorevisibleatthe in-termonthlyscalesof∼3-monthsand∼6-months.Forlarger scales,theextremesurgesdisplayauniformevolution dur-ingthetime.

Withtheaimofaddressingthenonstationarybehaviorof theextremesurges,wehaveappliedthewavelet multireso-lutiondecompositionofmonthlyextremaforeachsite.The processhasresultedintheseparationof different compo-nents,i.e.5waveletdetailstraducedbytwointermonthly scales (∼3-months and ∼6-months) andthree interannual scales (∼1.5-yr, ∼2—4-yr, and ∼5—8-yr). The explained

(12)

Figure7 Theboxplotofthemonthlymaximaofsurgesontheintermontly(_∼3monthsand_∼6months)totheinterannual( ∼1.5-yr;∼2—4-yrand∼5—8-yr)scales;caseofDunkirk.

variances of these scales are higher than those resulting fromthemultiresolutionanalysisofthehourlysurges:the varianceoftheintermonthlyscaleshasameanpercentage of45%ofthetotalsignalwhiletheinterannualscalesvary between28%and36%.

Then, the nonstationary behavior of the monthly ex-treme surges has been addressed by investigating their physicalconnectionswiththedifferentclimateindices(i.e. SST,SLP,ZW,andNAO).Thetotalsignal(notdecomposed)of theclimateindiceshasbeenlinkedtotheextreme compo-nentscalculatedpreviously.Foreachspectralcomponent, a seriesofMonteCarlosimulationshave been carriedout toidentifythemoststatisticallysignificantcorrelationwith theclimateindex.Thebestcorrelationofeachsurge com-ponent(i.e.∼3-months and∼6-months,∼1.5-yr, ∼2—4-yr and5—8-yr)withthesuitableclimateindex(i.e.SST, SLP, ZW,andNAO)isillustratedinFig. 8 .Thecorrelation coeffi-cientsvarybetween0.5and0.6attheintermonthlyscales (i.e.∼3-monthsand∼6-months)showinggoodagreements of SST index with the extreme surges. The ∼1.5-yr com-ponentis linkedtoSLPindexwithcorrelation coefficients varyingbetween0.65and0.7.Significantcoefficientshigher than0.7 havebeen observedforthe interannualscalesof ∼2—4-yr and∼5—8-yr exhibitingstronglinks withZWand NAOindicesrespectively.

Inthesecondpartofthissection,astationaryanda non-stationaryextremevalueanalysesbasedontheGEV distri-butionwithtime-dependentparameters(Coles, 2001 )have

beenimplementedtomodelthespectralcomponentofthe monthlymaximaofthesurges.

Here,weareinterestedtoseparatelymodelthe differ-entcomponentsoftheextremesurgesbyusingtheGEV dis-tribution.TheGEVdistributionusesthemaximumlikelihood methodwithaparametrizationofthedifferentparameters oflocation,scale,andshape.

Five stationary GEV models (GEV01, GEV02, GEV3, GEV04,andGEV05)havebeenusedtoﬁtthedifferent spec-tralcomponents of the monthly extreme surges. The AIC criterionhasbeencalculatedforeachspectralcomponent (Table 3 )tocomparetheirdistributions.Resultsshowhigher AIC scores for the intermonthly scales of ∼3-months and ∼6-months (GEV01 and GEV02); they decrease at larger-scalesof∼5.8-yr(GEV05).Thedifferencesbetweenthe ob-servedextreme valuesand theempiricalones, calculated from the stationary GEV models (from GEV01 toGEV05), shouldbeexplainedbysigniﬁcantdiscrepancieswhichvary according to the timescale of the extreme surges. These discrepanciesdecrease in thesites where theinterannual scales(i.e.GEV04andGEV05)experienceauniformtrendof extremes.

With the aim of improving the modelling of extreme surges,ﬁvenonstationaryGEVmodels(GEV1,GEV2,GEV3, GEV4andGEV5)have been appliedtoallspectral compo-nents.TheparametersofthenonstationaryGEVdistribution havebeenestimatedusingthemaximumlikelihoodmethod (Coles, 2001 ). Maximizing the log-likelihood function has

(13)

Figure8 Multiresolutiondecompositionofthemonthlymaximaofsurges(blackline).Thesignalofclimateoscillationsassociated withthedifferentindicesSST,SLP,ZW,NAO(greyline)hasbeencorrelatedtothespectralcomponentsofsurges.Onlythe con-nectionmaximizingthecorrelationcoefﬁcientbetweenaselectedclimateindexandthecomponentofsurges(fromintermonthly tointerannualtimescales)ispresented(thenormalizedvalueshavebeencalculatedtosuperposebothsignals).Thecorrelation coefﬁcientsarealsodisplayedateachtimescale.

Table3 AkaikeInformationCriterion(AIC)testresultsfor distributionmodelsofextremesurgesusingthestationaryGEV model.

∼ 3months _{∼ 6}months _{∼ 1.5-}y _{∼ 2—4}—y _{∼ 5—8-y}

Dunkirk -450 -352 -324 -352 -345

LeHavre -410 -322 -320 -350 -352

Cherbourg -395 -301 -314 -345 -350

Table4 AkaikeInformationCriterion(AIC)testresultsforthedistributionmodelsoftheextremesurgesusingthe nonsta-tionaryGEVmodel.

∼ 3months/SST ∼ 6months/SST ∼ 1.5-y/SLP ∼ 2—4—y/ZW ∼ 5—8-y/NAO

Dunkirk -310 -282 -266 -248 -232

LeHavre -308 -279 -260 -252 -243

Cherbourg -315 -285 -264 -245 -245

beenperformedbymeansofthe‘trustregionreﬂective al-gorithm’(Coleman and Li, 1996 ).

Foreachmodel,thebestclimateindexdescribingthe in-ternaloscillationsofthemonthlymaximahasbeenusedas acovariableinGEVdistributiontoaddressthe nonstation-arybehavioroftheextremevalues.Foreachspectral com-ponent, thestructureof themostappropriate nonstation-aryGEVdistributionhasbeenselectedbychoosingthemost adequateclimateindexthatminimizestheAkaike informa-tioncriterion (Akaike, 1974 ).Here,the totalsignalof the climate indexhasbeenusedasshowninFig. 8 .The good-nessoftheﬁtofeachmodelhasbeencheckedthroughthe visualinspectionofthequantile-quantile(Q-Q)plots;these plotscomparetheempiricalquantilesagainstthequantiles oftheﬁttedmodel.Thesubstantialdeparturefromthe di-agonalindicatesaninadequacyoftheGEVmodel.

ResultsprovidedbyGEV1—5revealabetterperformance (thelowestvaluesofAIC)ofextremeestimationcompared to the previous models of GEV01—05 and give the most appropriate distributions by the use of the climate oscil-lations. Indeed, the SLP, ZW,and NAO indices have been used for the interannual scales of ∼1.5-yr, ∼2—4-yr, and ∼5—8-yrrespectivelywhiletheSSTindexhasbeenselected as the most adequate for the intermonthly scales of ∼3-months and ∼6-months (Table 4 ). The Q-Q plots for the alltimescalesofthemonthly maximainDunkirkare illus-tratedinFig. 9 ;theyconﬁrmthesuitabilityoftheselected models.

Accordingly,the nonstationary GEVmodelshave exhib-ited high improvements at all timescales, in particular at theintermonthlyscales(GEV01/GEV1;GEV02/GEV2)where theAICscoreshavesigniﬁcantlydecreased.

(14)

Figure 9 The quantileplotbetween observed andmodelledextremesurges obtained using thenonstationaryGEV modelat differenttimescales.

4. Discussion

Results providedin thepresent researchhighlight the ori-ginofthenonlinearrelationshipbetweenlarge-and local-scales. Indeed, the intermonthly and the interannual ex-treme surgesmayberelatedtodifferent combinationsof several,notnecessarilysimilar,oscillatingatmospheric pat-terns.Theuseofamultiresolutionapproachtoinvestigate the dynamicsof the extreme surges intothe downscaling studiesprovestobeusefulforassessingthenonlinear inter-actions between large-scale climate variability and local-scale hydrodynamicchangesof thesea-level hydrodynam-ics.

Investigatingthephysicalrelationshipsoftheclimate os-cillations withthe multiscale surges have shown that the multimodelclimateensembleshouldbeusedtobetter un-derstandthiscomplexity,inparticularforextremeevents.

The seasonalandtheinterannual physicalrelationships betweenthelocalhydrodynamicsandtheclimate variabil-ity have been investigated in numerous previous works. Theyfocusedontheatmosphericcirculationwithdifferent relatedmechanisms(e.g.,Feliks et al., 2011 ;Lopez-Parages et al. 2012 ;Zampieri et al., 2017 ).

The use ofthe SSTindextoestimate theintermonthly extremesurgesshowsthattheirseasonalvariabilityshould berelatedtothevariationoftemperatureintheseasurface

controlledbyanimportantatmosphere-oceaninteractionin theAtlanticSea.Forshort-scales,Bell and Goring (1998) in-vestigatedtherelationshipbetweentheseasonalvariations ofthesealevelandtheSSTalongthenorth-eastcoastsof theNewZeland.Theyhavedemonstratedthatthedominant influenceofSSTontheannualcycleofsealevelisrelated tothethermo-stericsea-leveladjustmentswith50—80%of thevarianceintheannualfrequency.Theirworkshave sug-gestedthatthesealevelgenerallypeaksattheendofApril (australautumn)laggingtheSSTcyclebyaround2months, anditsseasonalmodulationisexplainedbychangesinthe oceaniccurrent patternsandtheseasonal couplingof cli-mateoscillationeffects.Previousstudieshaveindicatedthe impactofthemidlatitudeSSTgradientsontheatmospheric circulationinfluencingtheamplitudeandthelocationofthe stormtracks(e.g.,Nakamura and Yamane 2009 ;O’Reilly and Czaja 2015 ; Small et al., 2014 ).Forlarge-scales, the key roleof SSTinthe atmosphere-ocean interaction hasbeen alsoaddressedbyseveralworks.Accordingtosome obser-vationalstudies,themeridionalshiftsinthelocationofthe GulfStreamextensionfrontsaresignificantlylinkedtothe large-scalevariationsintheatmosphericcirculationatthe intraseasonaltimescales(e.g.,Frankignoul et al., 2011 ). Re-cently,Wills et al. (2016) haveexamined the atmosphere-oceaninteractionovertheGulfStreamanditsrelationwith twopotentialdifferentpatternsofatmosphericcirculation

(15)

I.Turkietal./Oceanologia62(2020)226—242 239 anomalies:onepatternleadstoSSTﬁeldandanother

resul-tantpatternlagstheSSTﬁeldandisconsideredasan atmo-sphericresponsetomidlatitudeoceanforcing(Wills et al., 2016 ).

The spatialdistributionoftheSSTanomaliesis strongly relatedtothestudy area,theinfluenceof theseabed to-pography andtherunoff conditions.Consideringthecases of the BristolChannel(Mayes and Wheeler, 2013 )and the SeineBayintheEnglishChannel,negativeanomaliesofSST aregenerallylocatedneartocoastalzonesconnectedto hy-drological systemsandinfluencedbyriversandtheannual regime of precipitations. The Northern Atlantic SSTfields havebeenhypothesizedtobelinkedtotheinterannualand thedecadalpersistenceofatmosphericcirculation anoma-lies(Namias, 1969 ;1972).Theseanomalies havebeen de-finedtobeconnectedtotheAtlanticMultidecadal Oscilla-tion(AMO)asanaturalmodeofvariability(e.g.,Schlesinger and Ramankutty, 1994 ).

Lopez-Parages et al. (2012) havestudied thestatistical predictability of the rainfall variability and statedthat it can beimproved byselectingthemostsuitablepredictors depending on the period on which the prediction is car-ried out.Theyhave alsosuggestedthat thenonstationary link of the rainfall with SST takes place when the dipo-lar patternsof the rainfall are reinforcedwith the nega-tive phases of theAMO(Atlantic Multidecadal Oscillation) along the20th _century. _In_this_study,_the_seasonal

modula-tions ofsurgeshave displayedstrong linkswithchangesin SST phases. The SST patterns, due tothe intrinsic modes of the atmospheric circulation variability, play a key role in regulatingtheglobal climatechangeandthus providing asourceofpotentialpredictabilityfortheclimate ﬂuctua-tionsonseasonaltimescales. Forlargerscales,theSST in-dexcanberelatedtotheAtlanticMultidecadalOscillation. The correlation between the ∼1.5-yr interannual vari-abilityofextremesurgesandtheSLPindexhasbeen demon-strated for the study sites in this work. In this context,

Turki et al. (2019) investigatedtheconnectionbetweenthe localdynamicsofthesurgesandtheglobalatmospheric cir-culation from SLPcomposites. They have suggested dipo-lar patternsofhigh-low pressureswithaseriesof anoma-lies at the interannual and the interdecadal scales which shouldberelatedtothephysicalmechanismslinkedtothe North-Atlantic andocean/atmosphericcirculation oscillat-ingatthesametimescales.AsdocumentedbyFrankignoul et al. (2011) , the SLP fields and the baroclinic instabil-ity of wind stress arerelated tothe Gulf Streampath as given by NCEP reanalysis; the dominant signal is a north-ward(southward)displacementoftheGulfStreamwhenthe NAO reaches positive (negative) extrema. Zampieri et al. (2017) used the daily mean SLP fields to analyze the in-fluence of theAtlantic seatemperature variabilityonthe day-by-daysequenceoflarge-scaleatmosphericcirculation patterns over the Euro-Atlantic region. They have found significant changes in the frequencies of certain weather regimes associated withthe phase shifts of the AMO. For hydrological applications, severalworks have investigated themultiscalerelationshipsbetweenthelocalhydrological changesandtheclimatevariability.Lavers et al. (2010) as-sociatedthe7.2-yrtimescalestoSLPpatternswhicharenot exactlyreminiscentoftheNAOanddefinecentersofaction whichareshiftedtotheNorth.

TherelationshipbetweentheZWindexandthe∼2-4-yr interannual variabilityof extreme surgeshas been signifi-cantlyvalidatedforthedifferentsitesoftheEnglish Chan-nelcoasts(asshownbyGEV4).TheZWisgenerally associ-atedwiththeatmosphericjetsinducedbythegeostrophic equilibriumwiththezonalmeanheightgradients.The wind-stressanomaliesinduceazonalwind-stressincreasingover thesubtropicalandthesubpolargyreboundaries.Thefast barotropicresponsetothewindstressisexplainedby east-erly winds in the Tropics and westerly winds in the mid-latitude (Wang, 2001 ). According tosome previous works (e.g.,L’Heureux and Thompson 2006 ;Seager et al. 2003 ), the inter-hemispheric symmetry of both temperature and zonalwindbearsgreatresemblancetoregressionpatterns associatedwiththeobservedElNiño.Lu et al. (2008) inves-tigatedEl NiñominusLa Niñacompositesfor theair tem-perature, zonal wind, and the tropopause pressure level. Theyderivedtheseparametersfromthedifferenceof pat-ternsbetween14warmyearsand12coldyears.Theyhave demonstratedthatthetroposphericzonalwindisintensified neartheequatorwardflankofthejet,resultingina equa-torwardstrengtheningofthejet.Andrade et al. (2012) in-vestigated the extreme temperature in Europe and its occurrenceinrelationtothelarge-scaleatmospheric circu-lation.Theyusedthezonalwindcomponentat 850hPato identifythepositiveandnegativephasesoftheextreme pe-riods.Theysuggestedthatbothphasesofextreme tempera-turechangesarecommonlyconnectedtostronglarge-scale changesinzonalandmeridionaltransportsofheatand mois-ture,resultinginchangesinthetemperaturepatternsover westernandcentralEurope(Corte-Real et al., 1995 ;Trigo et al., 2002 ). The studies from Mizuta (2012) and Zappa et al (2013) havedemonstratedthephysicallinksbetween the ZW and the extreme events from 11 Global Climate Model runs, suggesting the complexrelationship between the climate oscillation and the jet stream activity. They havefoundaslightincreaseinthefrequencyandstrength ofthestormsoverthecentralEuropeanddecreasesinthe thenumberofthestormsovertheNorwegianand Mediter-raneanseas.

IntheEnglishChannel andalongtheUKandthe North-ern coasts, changes in trends of the extreme waters and stormsurgeshavebeen explainedbythevariationsofthe energypressureandZW variabilityadditionaltothe ther-mosteric ﬂuctuations. These studies are in closer agree-mentswiththepresentresearchwherestrongphysicallinks betweentheZWandtheinterannualcomponent∼2-4-yrof theextremesurgeshavebeendemonstratedalongthe En-glishChannelcoasts(NorthFrance)

Inthepresentstudy,thelargerscalesofthe∼5-8-yr ex-tremesurgeshavebeenlinkedtotheNAOindex.In agree-mentwithother previous works(e.g. Marcos et al., 2012 ;

Philips et al., 2013 ),theNAOisconsideredasaninﬂuencing climate driverfor thelarge-scaleatmospheric circulation. Thephysicalmechanismsexplainingtheeffectsofthe con-tinuouschangesin NAOpatternsonthesea-level variabil-ityhavebeenaddressedinseveralstudies(e.g.,Trisimplis and Josey, 2001 ).Investigatingthelow frequencies ofthe sealevelshasshowntheexistenceofthelong-term oscilla-tionsthatoriginatefromlarge-scaleclimatevariabilityand thus controltheinterannualextreme surgesalongthe En-glishChannel. ThekeyroleoftheNAOontheinterannual

(16)

sea-level variabilityhas been explained by some previous works: Philips et al. (2013) investigated the influence of theNAOonthemeanandthemaximumextremesealevels in the BristolChannel/Severn Estuary. They have demon-strated that when high NAO wintersincrease in the posi-tivephase,windspeedsalsoescalatewhileincreasingthe negative NAO warmers results in low wind speeds. Then, thecorrelation between thelow/highextremesurgesand the NAOin theAtlantic has demonstrateda proportional-itybetweenNAOvaluesandtheaugmentationinthe win-ter storms.Feliks et al. (2011) defined significant oscilla-torymodesperiodsof∼2.8-yr,∼4.2-yr,and∼5.8-yrinboth observedNAOindexandNAOatmosphericmarineboundary layersimulationsforcedwithSST;theyhavesuggestedthat theatmosphericoscillatorymodesshouldbeinducedbythe GulfStreamoceanicfront.

Wecanconcludethatdifferentatmosphericoscillations could be related to the variations in the extreme surges according tothetimescale considered; for each scale,its bestphysicalconnectionwithaspeciﬁcclimateoscillation isproven.Thisresultsuggeststhattheatmospheric circula-tionactsasaregulator controllingthemultiscale variabil-ityofextremesurgeswithanonlinearconnectionbetween thelarge-scaleatmosphericcirculation andthelocalscale hydrodynamics. Such nonlinear characteristics depend on thedynamicsofthedifferentsequencesoftheatmospheric andwatervapourtransportpatternsduringthemonthprior tothesea-levelobservations(e.g.,Lavers et al., 2015 ).

The multiscale dependence ofthe local-scale hydrody-namic changes on the internal modesof the global-scale climateoscillationsisstillunderdebate.Theuseofthe di-verseindicesof SST, NAO,ZW,andSLPcouldimprovethe estimationoftheextremevaluesintheoceanographytasks anddeepenthescientiﬁcunderstandingofthephysical dy-namicsofsurgesincoastalenvironments.

In thiswork, the signal ofsurges hasbeen linearly ex-tracted fromthetotal sealevel time-seriesby theuse of the classicalharmonicanalysis andthanks tothe assump-tionthatthewaterlevelisthesumofthemeansealevel, tides,andsurges.Thisassumptionisnotcompletelyvalidin theEnglishChannelwhere thesigniﬁcanttide-surge inter-actions(Tomassin & Pirazzoli, 2008 )andtheeffectsofthe sea-levelriseontidesandsurgesareimportant(e.g.Idier et al., 2017 ).

Neglecting this nonlinear interaction between the surges,tides,andthesea-levelriseinthepresentwork sug-gests someuncertainties inthe estimationofthe spectral componentsfromthe instantaneous andmonthly extreme surges.Theseuncertainties,extractedfromthe multireso-lutionanalysisof,couldnotaffect themainresultsofthe presentworksincethenonstationarydownscalingfromthe globalatmosphericcirculationtothelocalsurgesshouldbe similartoupscalingtheshortscalesoflessthandaystothe scalesofmonths,i.e.theupscalinginwhichthesigniﬁcant interactionbetweenthesurgesandtidesismoreimportant

5. Conclusion

The results of the study explain key role of the cli-mate patterns in the nonstationy dynamics of the ex-treme surges; the ‘switch on’ and ‘off’ of each

cli-mate index is strongly related to the multiscale vari-ability. The use of the climate drivers could help improving the intermonthly and the interannual fore-casting. Furthemore, they can be considered for GCM simulationsthataredesignedtolookforthephysical mech-anisms explainingtheteleconnections betweenthe atmo-sphericcirculationandthesea-levelvariability.

Dependingontheatmosphericcirculation,describedby theclimate indices and weatherforecasts, thestochastic models necessary for estimating the extreme surges can bedevelopedandimprovedaccordingwiththeirmultiscale variabilityandtheirphysicalconnectionswiththeglobal cli-mateoscillations.Furthermore,similarassessmentsshould beundertakentoimproveunderstandingofthestorminess uncertainty.

Acknowledgments

The authors are grateful to the international project COTEST funded by CNES-TOSCAand related tothe future missionofSurfaceWaterandOceanTopography(SWOT). Au-thorsalsowouldlike tothankNational NavyHydrographic Service and National Center for Environmental Prediction for providingsea level andatmospheric data. Finally, the authorswouldliketoexpresstheirgratitudetotheeditor andthe unknown reviewersfor the excellentsuggestions, whichimprovedtheoriginalmanuscript.

References

Akaike,H.,1974.Informationtheoryasanextensionofthe maxi-mumlikelihoodprinciple.In:Petrov,B.N.,Csaki,F.(Eds.), Sec-ondInternationalSymposiumonInformationTheory.Akademiai Kiado,Budapest,267—281.

Andrade,C.,Leite,S.M.,Santos,J.A.,2012.Temperatureextremes inEurope:overviewoftheirdrivingatmosphericpatterns.Nat. Hazards Earth Syst. Sci. 12, 1671—1691, https://doi.org/10. 5194/nhess-12-1671-2012.

Bell,R.,Goring,G.,D.,G.,1998.SeasonalVariabilityofSeaLevel andSea-surfaceTemperature ontheNorth-eastCoast ofNew Zealand.Estuar.Coast.Shelf.Sci.46(2),307—319,https://doi. org/10.1006/ecss.1997.0286.

Bessoumoulin,P.,2002.LestempêtesenFrance.AnnalesdesMines 9—14.

Brown,J.,Souza,A.,Wolf,J.,2010.Surgemodellingintheeastern IrishSea:Presentandfuture stormimpact. OceanDynam.60 (2),227—236,https://doi.org/10.1007/s10236-009-0248-8.

Colberg,F.,McInnes,K.L.,Grady,J.,Hoeke,R.,2019.Atmospheric circulation changes and their impact on extreme sea levels aroundAustralia.Nat.HazardsEarthSyst.Sci.19,1067—1086,

https://doi.org/10.5194/nhess-19-1067-2019.

Coleman,T.F.,Li,Y.,1996.Aninteriortrust regionapproachfor nonlinearminimizationsubjecttobounds.SIAMJ.Optim.6(2), 418—445,https://doi.org/10.1137/0806023.

Coles,S.,2001.AnIntroductiontoStatisticalModelingofExtreme Values.Springer,London,209pp.

Corte-Real,J.,Zhang,Z.,Wang,X.,1995.Large-scalecirculation regimesandsurfaceclimaticanomaliesovertheMediterranean. Int.J.Climatol. 15,1135—1150, https://doi.org/10.1002/joc. 3370151006.

Costa,S.,Cantat,O.,Pirazzoli,P.A.,Lemaitre,M.,Delahaye,D., 2004. Vents forts et submersions de tempête en Manche Orientale:analysemétéo-marinesurlapériodehistorique

(17)

ré-I.Turkietal./Oceanologia62(2020)226—242 241

cente.ActesduColloquedel’AssociationInternationalede Cli-matologie.Climat,mémoiredutemps.Lesrelations climat-es-pace-société,277—280.

Dangendorf, S.,Wahl,T., Hein,H.,Jensen,J., Mai,S., Muders-bach,C.,2012.LevelVariabilityandInﬂuenceoftheNorth At-lanticOscillationonLong-TermTrendsintheGermanBight. Wa-ter4(1),170—195,https://doi.org/10.3390/w4010170.

Feliks,Y.,Ghil,M.,Robertson,A.W.,2011.Theatmospheric circula-tionovertheNorthAtlanticasinducedbytheSSTﬁeld.J.Clim. 24(2),522—542,https://doi.org/10.1175/2010JCLI3859.1.

Frankignoul,C.,Sennechael,N.,Kwon,Y-O.,Alexander,M.A.,2011. InﬂuenceoftheMeridionalShiftsoftheKuroshioand the Oy-ashioExtensionsontheAtmosphericCirculation.J.Climate24, 762—777,https://doi.org/10.1175/2010JCLI3731.1.

Haigh,I.,Nicolls,R.,Wells,N.,2009.Meansealeveltrendsaround the English Channel over the 20th _century _and _their _wider

context.Cont.Shelf.Res.29,2083—2098,https://doi.org/10. 1016/j.csr.2009.07.013.

Haigh,I.,Nicolls,R.,Wells,N.,2010.Assessingchangesinextreme sealevels:ApplicationtotheEnglishChannel1900—2006.Cont. Shelf.Res30,1042—1055,https://doi.org/10.1016/j.csr.2010. 02.002.

Hanson, S., Nicholls, R., Ranger, N., Hallegatte, S., Dorfee-Morlot, J., Herweijer, C., Chateau, J., 2011. A global rank-ing of port cities with high exposure to climate extremes. Clim. Change. 104 (1), 89—111, https://doi.org/10.1007/ s10584-010-9977-4.

Idier,D.,Dumas,F.,Muller,H.,2012.Tide-surgeinteractioninthe EnglishChannel.Nat.HazardsEarthSyst.Sci. 12,3709—3718,

https://doi.org/10.5194/nhess-12-3709-2012.

Idier,D.,Paris,F.,LeCozannet,G.,Boulahya,F.,Dumas,F.,2017. Sea-levelriseimpactsonthetidesoftheEuropeanShelf.Cont. Shelf.Res.137,56—71,https://doi.org/10.1016/j.csr.2017.01. 007.

Labat, D., 2005.Recentadvancesin waveletanalyses: Part1.A reviewofconcepts.J.Hydrol.314(1—4),275—288,https://doi. org/10.1016/j.jhydrol.2005.04.003.

Lavers,D.A.,Hannah,D.M.,Bradley,C.,2015.Connecting large-scaleatmosphericcirculation,riverﬂowandgroundwaterlevels inachalkcatchmentinsouthernEngland.J.Hydrol.523,179— 189,https://doi.org/10.1016/j.jhydrol.2015.01.060.

Lavers,D.A.,Prudhomme,C.,Hannah,D.M.,2010.Large-scale cli-mate, precipitationand British riverﬂows:Identifying hydro-climatologicalconnectionsanddynamics.J.Hydrol.395(3—4), 242—255,https://doi.org/10.1016/j.jhydrol.2010.10.036. Levin,S.A.,1992.Theproblemofpatternandscaleinecology:the

RobertH.MacArthurAwardLecture.Ecology73,1943—1967.

L’Heureux,M.L.,Thompson,D.W.J.,2006.Observedrelationships betweentheElNiño—SouthernOscillationandtheextratropical zonal-meancirculation.J.Climate.19,276—287,https://doi. org/10.1175/JCLI3617.1.

López-Parages,J.,Rodríguez-Fonseca,B.,2012.Multidecadal mod-ulationofElNiñoinﬂuenceontheEuro-Mediterraneanrainfall. Geophys.Res.Lett.39,art.1009no.L02704,https://doi.org/ 10.1029/2011GL050049.

Lu, J., Chen, G., Frierson, D., M., W., 2008. Response of the Zonal Mean AtmosphericCirculation to ElNiño versus Global Warming.J.Climate21,5835—5851,https://doi.org/10.1175/ 2008JCLI2200.1.

Marcos, M., Chust,G., Jordá,G., Caballero, A.,2012. Effect of sealevelextremesonthewesternBasquecoastduringthe21st century.Clim.Res.51(3),237—248,https://doi.org/10.3354/ cr01069.

Masina, M., Lamberti,A.,2013. Anonstationary analysisforthe Northern Adriatic extreme sea levels, J. Geophys. Res. 118, 3999—4016,https://doi.org/10.1002/jgrc.20313.

Massei,N., Dieppois,B.,Hannah, D.M.,Lavers,D.A.,Fossa, M., Laignel,B.,Debret,M.,2017.Multi-time-scalehydroclimate dy-namicsofaregionalwatershedandlinkstolarge-scale atmo-sphericcirculation:Applicationto theSeineriver catchment. France. J. Hydrol. 546, 262—275, https://doi.org/10.1016/j. jhydrol.2012.04.052.

Mayes,J.,Wheeler,D.,2013.Regionalweatherandclimatesofthe BritishIsles-Part1:Introduction.Weather68(1),3—8,https: //doi.org/10.1002/wea.2041.

Menendez,M., Woodworth, P.L.,2010. Changes inextreme high water levels basedon a quasi-global tide-gauge data set. J. Geophys.Res.115,art.no.C10011,https://doi.org/10.1029/ 2009JC005997.

Mizuta, R., 2012. Intensiﬁcation of extratropical cyclones asso-ciated with the polarjet changein the CMIP5 global warm-ingprojections.Geophys.Res.Lett.39,1—6.art.no.L19707,

https://doi.org/10.1029/2012GL053032.

Nakamura,M.,Yamane,S.,2009.Dominantanomalypatternsinthe near-surfacebaroclinicityand accompanyinganomaliesinthe atmosphereandoceans.PartI:NorthAtlanticbasin.J.Climate 22,880—904,https://doi.org/10.1175/2010JCLI3017.1. Namias,J.,1969. OnthecausesofthesmallnumberofAtlantic

hurricanesin1968.Mon.WeatherRev.97(4),346—348.

Nicholls,R.J.,Marinova,N.,Lowe,J.A.,Brown,S.,Vellinga,P.,De Gusmao,D., Hinkel,J.,Tol,R.S.,2011.Sea-levelrise andits possibleimpactsgivena“beyond4Cworld” inthetwenty-ﬁrst century.Philos.T.Roy.Soc. A369,161—181,https://doi.org/ 10.1098/rsta.2010.0291.

O’Reilly,C.H.,Czaja,A.,2015.TheresponseofthePaciﬁcstorm trackand atmosphericcirculation to Kuroshio Extension vari-ability,Q.J.Roy.Meteor.Soc.141(686),52—66,https://doi. org/10.1002/qj.2334.

Pasquini,A.I.,Lecomte,K.L.,Depetris,P.J.,2008.Climatechange andrecentwaterlevelvariabilityinPatagonianproglaciallakes, Argentina.GlobalPlanet.Change63,290—298.

Philips, M.R., Rees, E.F., Thomas, T., 2013. winds, sea level andNorthAtlanticOscilation(NAO)inﬂuences:Anevaluation; GlobalPlanet.Change100,145—152,https://doi.org/10.1016/ j.gloplacha.2012.10.01.

Pirazzoli,P.A.,2005.A reviewofpossibleeustatic, isostaticand tectoniccontributionsineightlate-Holocenerelativesea-level histories from the Mediterranean area. Quaternary Sci. Rev. 24 (18—19), 1989—2001, https://doi.org/10.1016/j.quascirev. 2004.06.026.

Pugh,D.J.,1987.Tides, SurgesandMeanSea-Level:AHandbook forEngineersandScientists.JohnWiley,Chichester472pp.

Sang,Y.F.,2013.Areviewontheapplicationsofwavelettransform inhydrologytimeseriesanalysis.Atmos.Res.122,8—15,https: //doi.org/10.1016/j.atmosres.2012.11.003.

Schlesinger, M.E., Ramankutty, N., 1994. An oscillation in the globalclimatesystemofperiod65—70years.Nature367,723— 726.

Seager,R.,Harnik,N.,Kushnir,Y.,Robinson,W.,Miller,J.,2003. Mechanismsofhemisphericallysymmetricclimatevariability.J. Climate16,2960—2978.

Shaw,A.G.P.,Tsimplis,M.N.,2010.The18.6yrnodalmodulationin thetidesofSouthernEuropeanCoasts.Cont.ShelfRes.30(2), 138—151,https://doi.org/10.1016/j.csr.2009.10.006.

SHOM, 2014. Caractérisation de 7 évènements de tempêtes de l’automne-hiver2013—2014N°001/2014.

Small, R.J., Tomas, R.A., Bryan, F.O., 2014. Storm track re-sponse to ocean fronts in a global high-resolution climate model. Clim. Dynam. 43, 805—828, https://doi.org/10.1007/ s00382-013-1980-9.

Tomasin,A.,Pirazzoli,P.A.,2008.ExtremeSeaLevelsintheEnglish Channel:CalibrationoftheJointProbabilityMethod.J.Coastal Res.24,1—13,https://doi.org/10.2112/07-0826.1.

(18)

Torrence,C.,Compo,G.P.,1998.Apracticalguidetowavelet anal-ysis.B.Am.Meteorol.Soc.79,61—78,https://doi.org/10.1175/ 1520-0477(1998)0790061:APGTWA.0.CO;2.

Trigo,R.M.,Osborn,T.J.,Corte-Real,J.,2002.TheNorthAtlantic OscillationinﬂuenceonEurope:climateimpactsandassociated physicalmechanisms.Clim.Res.20,9—17,https://doi.org/10. 3354/cr020009.

Trisimplis,M.N., Josey,S.A.,2001. ForcingoftheMediterranean Seabyatmosphericoscillationsover theNorthAtlantic. Geo-phys.Res.Lett.28,803—806.

Tsimplis,M.N.,Woodworth,P.L.,1994. Theglobaldistributionof theseasonalsealevelcyclecalculatedfromcoastaltidegauge data.J.Geoph.Res.99(C8),16031—16039.

Turki,I.,Laignel,B.,Chevalier,L.,Massei,N.,Costa,S,2015b.On theInvestigationoftheSeaLevelVariabilityinCoastalZones us-ingSWOTSatelliteMission:exampleoftheEasternEnglish Chan-nel(WesternFrance).IEEEJ.Sel.Top.Appl.8(4),1564—1569,

https://doi.org/10.1109/JSTARS.2015.2419693.

Turki, I.,Laignel,B.,Kakeh,N., Chevalier,L.,Costa, S.,2015a. MethodologyforFillinggapsand Forecastin sealevel: appli-cation tothe easternEnglish Channeland the NorthAtlantic Sea(western France). Ocean Dynam. 65(9—10), 1221—1234,

https://doi.org/10.1007/s10236-015-0824-z.

Turki,I.,Massei,N.,Laignel,B.,2019.LinkingSeaLevelDynamic andExceptionalEventstoLarge-scaleAtmosphericCirculation Variability:CaseofSeineBay,France.Oceanologia61(3),321— 330,https://doi.org/10.1016/j.oceano.2019.01.003.

Vousdoukas, M.I., Mentaschi, L., Voukouvalas, E., Verlaan, M., Feyen,L.,2017.ExtremesealevelsontherisealongEurope’s

coasts.Earth’sFuture5(3),304—323,https://doi.org/10.1002/ 2016EF000505.

Wang,C.,2001.AtlanticClimateVariabilityandItsAssociated At-mosphericCirculationCells.J.Climate15,1516—1536.

Williams,J.,IrazoquiApecechea,M.,Saulter,A.,Horsburgh,K.J., 2018. Radiational tides: their double-counting in storm surge forecasts and contribution to the Highest Astronomi-calTide.OceanSci.14,1057—1068,https://doi.org/10.5194/ os-14-1057-2018.

Wills,S.W.,Thompson,D.W.J.,Ciasto,L.M.,2016.OntheObserved Relationships between Variability in GulfStream Sea Surface TemperaturesandtheAtmosphericCirculationovertheNorth Atlantic.J. Climate29,3719—3730, https://doi.org/10.1175/ JCLI-D-15-0820.1.

Yan,Z.,Tsimplis,M.N.,Woolf,D.,2004.Analysisoftherelationship betweenthe NorthAtlantic Oscillationand sea-levelchanges innorthwesternEurope.Int.J.Climatol.24,743—758,https: //doi.org/10.1002/joc.1035.

Zampieri,M.,Toreti,A.,Schindler,A.,Escoccimarro,E.,Gualdi,S., 2017. Atlantic multi-decadaloscillation inﬂuence onweather regimesoverEuropeandtheMediterraneaninspringand sum-mer.GlobalPlanet.Change,151,92—100,https://doi.org/10. 1016/j.gloplacha.2016.08.014.

Zappa, G., Shaffrey, L.C., Hodges, K.I., Sansom, P., Stephen-son,D.B.,2013.Amultimodelassessmentoffutureprojections ofNorthAtlanticandEuropeancyclonesintheCMIP5climate models. J. Climate26, 5846—5862, https://doi.org/10.1175/ JCLI-D-12-00573.1.