How much do tides affect the simulated climate of the ... · How much do tides affect the simulated climate of the Mediterranean Sea? The Nemerte Project ... PROTHEUS regional climate

PRACE Day16 Scientific & Industrial Conference 11 May 2016

How much do tides affect

the simulated climate of

the Mediterranean Sea?

The Nemerte Project

Gianmaria SanninoENEA C.R. Casaccia (Roma)Climate Modelling and Impacts Laboratory

NEMERTE – Numerical Experiment on the Mediterranean model response to Enhanced Resolution and TidE

Project leader: Gianmaria Sannino, ENEA, ITALY

Collaborators:

Patrick Heimbach, Massachusetts Institute of Technology (MIT), UNITED STATESGabriel Jorda Sanchez, Mediterranean Institute for Advanced Studies (IMEDEA), SPAIN Vincenzo Artale, ENEA, ItalyAdriana Carillo, ENEA, ItalyEmanuele Lombardi, ENEA, ItalyGiovanna Pisacane, ENEA, Italy MariaVittoria Struglia, ENEA, Italy

FERMI@CINECAArchitecture: 10 BGQ FramesModel: IBM-BG/QProcessor type: IBM PowerA2 @1.6 GHzComputing Cores: 163840 Computing Nodes: 10240 RAM: 1GByte / core (163 PByte total)Internal Network: 5D TorusDisk Space: 2PByte of scratch space Peak Performance: 2PFlop/s

N. 7 in Top 500 rank (June 2012)

National and PRACE Tier-0 calls

NEMERTE Resource Awarded:18 Million core hours on FERMI

The mission of PRACE (Partnership for Advanced Computing in Europe) is to enable high impact scientific discovery and engineering research and development across all disciplines to enhance European competitiveness for the benefit of society. PRACE seeks to realize this mission by offering world class computing and data management resources and services through a peer review process.

OUTLINE

HOW

DID

NEMERTE

BEGIN?A fascinating story involving ocean models and HPC

Workshop– Instanbul – July2012

The Mediterranean Sea is a semi-enclosed basin displaying an active thermohaline circulation that is sustained by the atmospheric forcing and controlled by the narrow and shallow Strait of Gibraltar

Mediterranean Thermohaline Circulation (MTHC)

The atmospheric forcing drives the Mediterranean basin toward a negative budget of water and heat, and toward a positive budget of salt. Over the basin, evaporation exceeds the sum of precipitation and rivers discharge, while through the surface a net heat flux is transferred to the overlying atmosphere. MassconservationinthebasinrepresentsthelastingredientnecessarytoactivatetheMTHC

ChartoftheStraitofGibraltar,adaptedfromArmi &Farmer(1988),showingtheprincipalgeographicfeaturesreferredtointhetext.

Areasdeeperthan400mareshaded

Espartel SillCamarinal Sill

Strait of Gibraltar Background

Tarifa Narrow


Strait of Gibraltar Background: 3D Bathymetry


Image of acoustic backscatter during ebb tide over Camarinal Sill inthe Strait of Gibraltar (Wesson and Gregg, 1994)

Strait of Gibraltar Background: Physics

Strongmixingandentrainmentmainlydrivenbytheveryintensetides.

semidiurnal constituents and particularly for M2. The opposite behavioris found in the Atlantic layer, where amplitude increases monotonicallyfrom east to west. Phases of semidiurnal constituents also show suchmonotonic characteristic. This behavior was reported by García-Lafuente et al. (2000) between the eastern exit of the strait and themain sill of Camarinal and the analysis here not only confirms this resultbut also extends it to the western exit of the strait. Between GC and CS,the amplitude of M2 transport in theMediterranean layer is reduced bymore than 50% while it is nearly 6 times greater in the Atlantic layer atCS (Table 2). A very clear transfer of the tidal signal from theMediterra-nean to the Atlantic layer takes place between GC and CS during theflood tide and vice-versa during the ebb tide. This fact led Bray et al.(1990) to propose the conceptual model of a membrane-like interfacebetween CS and GC sections whose role would be the transfer of thetidal signal from one layer to the other. Phase difference betweenthese locations in the Atlantic layer indicates that tidal transport propa-gates fromCS to GC, a fact that is ascribed to the strong baroclinic natureof the internal tide. An inspection of values in Table 2 indicates that thesame behavior applies between CS and ES sections.

The two-layer sketches presented in Fig. 4 illustrate key aspects ofthe tidal dynamics in the strait and explain the aforementioned diver-gences. Fig. 4.a represents the hypothetical mean exchange and meaninterface position in the absence of tidal forcing. Tidal transports as-sociated with M2 within each layer (small two-headed arrows withsize proportional to the transport, which is indicated by the numbersbeside) and the total barotropic transport (large arrows with numbersinside) have been included for ES, CS and GC sections (see Fig. 1 for lo-cations). Notice that M2 transport by itself (3.1 Sv) would reverse themean flow, which is around 0.8 to 1 Sv in each layer (Baschek et al.,2001; Garcia-Lafuente et al., 2002a; Sánchez-Román et al., 2009), if itwere equally distributed among both layers. This is not the case at thestrait's boundaries because most of the tidal flow moves through thepassive, slow-flowing layer (Mediterranean layer in the east, Atlanticlayer in the west), making them reverse periodically. At these bound-aries, however, the tidal transport in the active, fast-flowing layers is in-sufficient to reverse the total flow. In CS the amplitude of M2 in anylayer is greater than 1 Sv and, except for some neap tides, the totalflow in each layer reverses during part of the tidal cycle.

Fig. 4.b sketches the flow during flood tide (thick gray line in thesea level graph) that implies westward barotropic tidal transport(thick arrow beside the graph). At this time, the flow is hydraulicallycontrolled at CS and only a limited volume of the west-going tidal flowthat has flown through GC is able to surpass the sill, the rest remainingtrapped between both sections. Consequently, the interface is pushed up-ward and forces water in the Atlantic layer to move away from the area.Part of it moves eastward, thus reducing the amplitude of tidal currentsin the upper layer at GC, but most of it goes on to the west, transferringthe signal from the Mediterranean to the Atlantic layer. Some tidal flowcan be transferred cross-strait reaching the continental shelf of both theSpanish and Moroccan coasts and promoting some coastal recirculation.This topic will be addressed in the next section. At the ES section, westof CS, the Mediterranean flow is also hydraulically controlled. In fact, thehydraulic control in ES is more permanent than in CS (Sánchez-Román,2008b; Sannino et al., 2007, 2009) and the reduced fraction of the tidaltransport crossing CS is not even able to overflow ES. Consequently, thewater accumulates in the Tangier basin forcing the flow to change to sub-critical and originating the well-known internal hydraulic jump down-stream CS (Armi and Farmer, 1988; Sanchez-Garrido et al., 2008, 2011).Once more the interface is pushed up displacing the Atlantic waterabove toward the Atlantic Ocean and completing the transfer of thelarge tidal signal that is observed in the lower layer in the east to theupper layer in the west. A first guess of the interface vertical excursionin the Tangier basin is obtained by integrating the equation

S∂η∂t ¼ Q2;E−Q2;C ð6Þ

where η is the interface depth, S is its horizontal area assumed constantand equal to the area at its mean depth in the inter-sill region (160 m),and Q2,E and Q2,C are the tidal transports in the Mediterranean layerthrough the ES and CS sections, respectively. The equation,which is appli-cable to any constituent, gives a vertical excursion of around 75 m for M2

(35 m for S2), which is an important fraction of the water depth in theTangier basin (~500 m).

During ebb tide the situation reverses, but the interpretation israther similar. The main difference is the flooding of the hydrauliccontrol in CS by the end of the flood tide when the tidal currentweakens and the subsequent release of the internal jump that propa-gates as an internal bore with the associated short-wavelength inter-nal wave packet to the east (Fig. 4.c). The flooding of CS control allowsthe accumulated Mediterranean water in the Tangier basin to floweastward, thus reversing the flow in this layer at CS, although partof the water keeps on flowing westward, smoothing out the ampli-tude of tidal currents at ES in this layer. The Mediterranean water ac-cumulated between CS and GC in the previous flood evacuates towardthe Mediterranean through GC, a process favored by the hydrauliccontrol in Tarifa narrows that prevents the large volume of Atlanticwater moving to the east during the ebb tide to cross the control sec-tion. As this water accumulates between CS and GC, it pushes down theinterface and forces the Mediterranean water below to flow back

Fig. 4. Two-layer sketch to explain tidal transport divergence in each layer. (a) Plate of themean exchange with indication of M2 tidal transport. The mean flow of around 0.8 Sv ineach layer has not been indicated to prevent the figure from being overcrowded withnumbers. (b) Illustration of the flood tide (barotropic tidal current toward the AtlanticOcean). (c) The same for ebb tide (tidal current toward the Mediterranean). See text fora more detailed explanation.

14 A. Sánchez-Román et al. / Journal of Marine Systems 98–99 (2012) 9–17

A.Sánchez-Románetal,JGR2012


8<

:

QA +QM = E � P �R

QASA +QMSM = 0(1)

Strait of Gibraltar Background: Hydraulics

MassConservation

SaltConservation

Knudsenequations(1899)8><

>:

QA = SMEnetSM�SA

QM = �hSAEnetSM�SA

i.

(1)

If the exchange is subject to one hydraulic control in the western part of the Strait, the regime is called submaximal, while if the flow exchange is also controlled in the eastern part of the Strait along TN, the regime is called maximal.

The maximal regime can be expected to have larger heat, salt, and mass fluxes and to respond more slowly to changes in stratification and thermohaline forcing within the Mediterranean Sea and the North Atlantic Ocean.

Strait of Gibraltar Background: Hydraulics


Climate models, both global and regional, represent the

Strait of Gibraltar like a rectangular pipe & notides

NEMERTE project: Questions & Motivation

StraitofGibraltar


• Is it reasonable?

• If NOT, what are these modelsneglecting?• If NOT, what is the minimum resolution to adopt for the Strait?

Climate models, both global and regional, represent the

Strait of Gibraltar like a rectangular pipe & notides

NEMERTE project: Questions & Motivation


Answer: Direct simulation of the Strait at high resolution


Ratio = 3:1

1/8° x1/8°42z-levels

Sannino et al. 2009, “An eddy-permitting model of the Mediterranean Seawith a two-way grid refinement at the Strait of Gibraltar”. Ocean. Modeling

Effects of high resolution at Gibraltar in a 1/8° Mediter. model

1/24° x1/24°

1/8° x1/8°




1/24° x1/24°42z-levels


OceaniccomponentinthePROTHEUS regionalclimatesystem


Salinity field @ 40m



Sannino et al. 2009, “An eddy-permitting model of the Mediterranean Seawith a two-way grid refinement at the Strait of Gibraltar”. Ocean Modeling



Sannino et al. 2009, “An eddy-permitting model of the Mediterranean Seawith a two-way grid refinement at the Strait of Gibraltar”. Ocean Modeling

UnresolvedissuesinMediterraneanSeaLevel



ModifiedPOM

MinimalHor.Resolution:<500m

ExternalTime-Step:0.1sec

O1 K1 diurnaltidalcomponent

M2 S2 diurnaltidalcomponent

•Sannino et al, JGR-Book, 2013

•Sannino et al, JPO, 2009

•Sanchez et al, JGR, 2009

•Garrido et al, JGR, 2008

•Garcia-Lafuente et al, JGR, 2007

•Sannino et al, JGR, 2007

•Sannino et al , NC, 2005



Sub-basin Model: Cadiz – Gibraltar - Alboran


salinity along-strait section



Velocity @ 5 m Velocity @ 250 m



Sánchez-Román etal.,JGR,2009

Max Differences:Amp: 10 cm s-1

Pha: 20°

TidalComponentscomparisonSurfaceelevation

TidalComponentscomparisonAlong-straitvelocity

Max Differences:Amp: 3.6 cmPha: 11°

Sanninoetal.,JPO,2009



G2 = F 21 + F 2

2

In order to determine whether a particular state is subcritical, supercritical, or critical, it is

helpful to rewrite this equation as:

where

Strait of Gibraltar: new 3Layer Hydraulic Theory

Sanninoetal,JPO,2009


Bars indicating the presence of provisional supercritical flowwith respect to one mode (black) and with respect to bothmodes (grey) in the three main regions of the Strait: EspartelSill, Camarinal Sill and Tarifa Narrow. Lower panel indicatestidal elevation at Tarifa.

POM model and hydraulics


Are the hydraulic results model depended?

To answer the question the exchange flow simulated by POM has beencompared with the exchange flow simulated by a very high resolutionnon-hydrostatic model implemented for the Strait region.

POM model and hydraulics


MITgcm vs POM : model grids

POMgridMaxresolution

300m

MITgcmgridMaxresolution

25m(only25%ofthe

actualgridisshown)

Sanninoetal2014


MITgcm vs POM : model bathymetry

Realbathymetry

POMbathymetry32sigma-levels

MITgcmbathymetry53verticalzeta-levels

(partialcell)

Sanninoetal2014


MITgcm model simulation

Interface depth evolution


POM MITgcm

MITgcm vs POM – Internal bore evolution



Horizontal velocity(m/s)

MITgcm POMVertical velocity(m/s)

MITgcm POM

HorizontalcurrentvelocitysimulatedbyMITgcmduringthearrivalofaninternalwavestrainatTN.(b)Sameas(a)simulatedbyPOM.(c)VerticalcurrentsimulatedbyMITgcmduringthesameinstantof(a).(d)Sameasc)simulatedbyPOM.


POM MITgcm



Observed and models interface layer thickness

a) Locationsofhistoricalconductivity-temperature-depthdata(CTD,blackdots)collectedintheStrait.

b) InterfacelayerthicknesscomputedfromCTDdata.

c) BottomtopographyalongthecentralaxisoftheStrait.

POMMIT


MITgcm vs POM alongstrait hydraulics

Frequencyofoccurrence,overthetropicalmonthperiod,ofsupercriticalflowwithrespecttoonemode(a)andbothmodes(b)alongtheStraitasobtainedbyPOM(dashedline)andMITgcm (solidline).

MITgcm displaysamarkedalong-straitvariabilityassociatedtothefinerdescriptionofthebathymetry.Moreoverwhenbothmodesaresupercritical(a),MITgcm predictslowervaluesallalongtheStraitwithrespecttoPOM,exceptforESwhereonthecontraryMITgcm exceedsPOM.Whentheflowissupercriticalwithrespecttojustonemode,themajordifferencesareconfinedalongTN.InparticularPOMpredictshigherfrequencieswithrespecttoMITgcm.

(a)

(b)

(c)


MITgcm vs POM alongstrait hydraulics

POM

MITgcm


EvolutionofthehorizontalvelocityfieldalonglongitudinalSectionin themiddleoftheStraitduringthearrivalofanintervalwavetraintoTN.Elapsetimebetweenframesis1.33hours.Panelsonthetopofeachframeindicatetheflowcriticality;zero:subcriticalflow;one:onlyoneinternalmodecontrolled;two:bothinternalmodescontrolled.

MITgcm vs POM alongstrait hydraulics & bore propagation


MITgcm sensitivity to non-hydrostaticity

MITgcmNONHydrostatic

MITgcmHydrostatic


MITgcm alongstrait hydraulics & bore propagation-Hydrostatic

Hydrostatic NON-Hydrostatic


MITgcm alongstrait hydraulics & bore propagation-Hydrostatic

POMOriginal simulation MITgcm

NON-HydrostaticOriginal grid

MITgcmhydrostatic

Model grid as POM


•Variable horizontal resolution

(1/16° up to 1/200°)

•72 vertical levels

•Tidal forcing (main 4 components)

•Surface atmospheric pressure

MITgcm

New modeling strategy for the Mediterranean

X - 6 SANNINO ET AL.: SHORT TITLE

6oW 3oW 0o 3oE

34oN

36oN

38oN

40oN

42oN

6oW 3oW 0o 3oE

34oN

36oN

38oN

40oN

42oN

Figure 2: Model grid over the Western Mediterranean Sea.

10 20 30 40 50 60Time (Days)

−3−2

−1

0

1

23

Tran

spor

t (Sv

)

10 20 30 40 50 60Time (Days)

−3−2

−1

0

1

23

Tran

spor

t (Sv

)

10 20 30 40 50 60Time (Days)

−3−2

−1

0

1

23

Tran

spor

t (Sv

)

10 20 30 40 50 60Time (Days)

−3−2

−1

0

1

23

Tran

spor

t (Sv

)

Figure 3: Atlantic (black) and Mediterranean (gray) transports computed from the tidal simulation for section at 6.22◦W(upper panel), 5.76◦W (middle panel) and 5.33◦W (lower panel).


New modeling strategy for the Mediterranean: NEMERTE Results

New modeling strategy for the Mediterranean: NEMERTE Results

SurfaceSalinityfield


Atlantic(black)andMediterranean(grey)transportscomputedfromExpT simulationforsectionsat6:22W(upperpanel),5:76W(middlepanel)and5:33W(lowerpanel).Positiveforeastwarddirection.

NEMERTE Results: volume transport through the Strait

10 20 30 40 50 60Time (Days)

−3−2−1

0123

Tran

spor

t (Sv

)

10 20 30 40 50 60Time (Days)

−3−2−1

0123

Tran

spor

t (Sv

)

10 20 30 40 50 60Time (Days)

−3−2−1

0123

Tran

spor

t (Sv

)

10 20 30 40 50 60Time (Days)

−3−2−1

0123

Tran

spor

t (Sv

)


NEMERTE Results: temperature in the Strait

Timeseriesoftheverticalminimumtemperature fortheyears1963-1967.UpperpanelfortheCSandlowerpanelforES.ExpT ingrey,ExpNT inblack

Jan3 Apr3 Jul3 Oct3 Jan4 Apr4 Jul4 Oct4 Jan5 Apr5 Jul5 Oct5 Jan6 Apr6 Jul6 Oct6 Jan7Time (Month Year)

12.913.013.113.213.313.413.5

Tem

pera

ture

(C0 )


12.913.013.113.213.313.413.5

Tem

pera

ture

(C0 )


12.913.013.113.213.313.413.5

Tem

pera

ture

(C0 )

TimeseriesofthedifferencebetweendailyminimumtemperatureinCSandinES filteredofffrequencieshigherthan30days.ExpT ingrey,ExpNT inblack.


−0.2

−0.1

0.0

0.1


−0.2

−0.1

0.0

0.1

tide

no-tide

Sannino et al. 2015 Progress in Ocean.


NEMERTE Results: salinity in the Strait

Timeseriesoftheverticalmaximumsalinity fortheyears1963-1967.UpperpanelfortheCSandlowerpanelforES.ExpT ingrey,ExpNT inblack


0.00

0.05

0.10

0.15

0.20


0.00

0.05

0.10

0.15

0.20TimeseriesofthedifferencebetweendailymaximumsalinityinCSandinES filtered offfrequencieshigherthan30days.ExpT ingrey,ExpNT inblack.


38.038.1

38.2

38.3

38.4

38.5

Salin

ity (P

su)


38.038.1

38.2

38.3

38.4

38.5

Salin

ity (P

su)


38.038.1

38.2

38.3

38.4

38.5

Salin

ity (P

su)

tide

no-tide


Absolutedifference intheSoGbetweentheverticalmixinginExpT andExpNT.Thepositionofthesectionintheupperpanelisshownbytheblacklineinthelowerpanel.Lowerpanelmapshowsthemaximumcomputedovertheentiredepth.Valuesaretime-averagedoverthelastyearofthesimulation.Logarithmicvaluesareshown.

NEMERTE Results: vertical mixing in the Strait of Gibraltar


Salinity along-straitsection.UpperpanelExpT,lowerpaneldifferenceExpT- ExpNT.Time-averageovertheentiresimulatedperiod

NEMERTE Results: temperature & salinity along-strait section

Temperature along-straitsection.UpperpanelExpT,lowerpaneldifferenceExpT-ExpNT.Time-averagedovertheentiresimulatedperiod.


Temperaturedifference betweentheExpT andExpNT.Dataareverticallyaveragedbetweentheupperlayerand150m.Time-averagedovertheentiresimulatedperiod

TemperaturedifferencebetweentheExpT andExpNT.Dataareverticallyaveragedbetween150mand500m.Time-averagedovertheentiresimulatedperiod

NEMERTE Results: temperature difference


Salinitydifference betweentheExpTandExpNT.Dataareverticallyaveragedbetweentheupperlayerand150m.Time-averagedovertheentiresimulatedperiod

SalinitydifferencebetweentheExpTandExpNT.Dataareverticallyaveragedbetween150mand500m.Time-averagedovertheentiresimulatedperiod

NEMERTE Results: salinity difference


NEMERTE ResultsFirstthreeEOFfortemperatureverticallyaveragedovertheupper150m

EOF1 EOF2 EOF3

NO-TIDE

TIDE


NEMERTE Results: surface circulation

Temperatureat50mtime-averagedovertheperiodOctober-December1962

2 4 6 8 10 12 14

2 4 6 8 10 12 14

3638

4042

44

3638

4042

44

Temperature

14.00 14.75 15.50 16.25 17.00 17.75 18.50 19.25 20.00

0.50 (m/s)

2 4 6 8 10 12 14

2 4 6 8 10 12 14

3638

4042

44

3638

4042

44

Temperature

14.00 14.75 15.50 16.25 17.00 17.75 18.50 19.25 20.00

0.50 (m/s)

ExpTD ExpNoTD


2 4 6 8 10 12 14

2 4 6 8 10 12 14

3638

4042

44

3638

4042

44

Salinity (Psu)

36.60 36.90 37.20 37.50 37.80 38.10 38.40 38.70 39.00

0.50 (m/s)

2 4 6 8 10 12 14

2 4 6 8 10 12 14

3638

4042

44

3638

4042

44

Salinity (Psu)

36.60 36.90 37.20 37.50 37.80 38.10 38.40 38.70 39.00

0.50 (m/s)

2 4 6 8 10 12 14

2 4 6 8 10 12 14

3638

4042

4436

3840

4244

Salinity (Psu)

36.60 36.90 37.20 37.50 37.80 38.10 38.40 38.70 39.00

0.50 (m/s)

2 4 6 8 10 12 14

2 4 6 8 10 12 14

3638

4042

4436

3840

4244

Salinity (Psu)

36.60 36.90 37.20 37.50 37.80 38.10 38.40 38.70 39.00

0.50 (m/s)

2 4 6 8 10 12 14

2 4 6 8 10 12 14

3638

4042

44

3638

4042

44

Salinity (Psu)

36.60 36.90 37.20 37.50 37.80 38.10 38.40 38.70 39.00

0.50 (m/s)

2 4 6 8 10 12 14

2 4 6 8 10 12 14

3638

4042

44

3638

4042

44

Salinity (Psu)

36.60 36.90 37.20 37.50 37.80 38.10 38.40 38.70 39.00

0.50 (m/s)

SalinityfieldsaveragedoverOctoberandDecember@50m

1962 1963 1964

TIDE

NO-TIDE

NEMERTE Results: surface circulation


Mixedlayerdepth(m)

NO-TIDE

TIDE

NEMERTE Results: Mixed Layer Depth


NEMERTE Results


NEMERTE Results: deep convection


Is it still reasonable?

RegionalClimatemodelsrepresentthe

StraitofGibraltarlikeasimplepipewithouttides

Conclusions

Documents

How much do tides affect the simulated climate of the ... · How much do tides affect the simulated climate of the Mediterranean Sea? The Nemerte Project ... PROTHEUS regional climate