Mercator Ocean newsletter 11

The MERCATOR Quaterly Newsletter N°11 – October 2001 – Page 1GIP Mercator Océan

Oceanic transport through Atlantic straits Romain Bourdallé-Badie, Yann Drillet and Karine Béranger

Introduction

Straits play a special role in ocean circulation. They transform water masses, form currents and allow water to flow from one

basin to another. In the North Atlantic and Mediterranean area simulated using the PSY2V1 analysis and forecasting system,

there are numerous straits more or less determining large-scale circulation throughout the basin. This study focuses on straits in

two areas: the Caribbean Sea, where the Gulf Stream is formed, and the North-East Atlantic where the North Atlantic bottom

waters and Greenland current are formed. The present study is based on the transport of water masses computed by PSY2V1

from October 2001 to June 2003 after having divided up the areas into sections. We take a more detailed look at 2002.

Numerical model

PAM

The PSY2V1 analysis and forecasting system implements Mercator’s PAM ocean model (a prototype covering the North Atlantic

and Mediterranean), described in Mercator Newsletter No. 5 [Siefridt et al., 2001]. The PAM configuration was developed from

version 8.1 of the OPA ocean model [Madec et al, 1998] using the physical parameters for the Clipper project [Tréguier et al.,

2001]. The simulated area ranges from 9°N to 70°N in the Atlantic Ocean and covers all the Mediterranean Sea with a horizontal

resolution of between 5 and 7 km. This study only used the free configuration of the PAM model for comparisons with PSY2V1.

Transport through Caribbean straitsWater transport around the islands is deduced from the stream function, which indicates barotropic transport through the straits.

This study with PSY2v1 focuses on 2002, during the spin up to real time, started in October 2001. The study uses the model’s

daily output.


Circulation in 2002 computed during the spin up to real time

The Florida current is the main contributor to the Gulf Stream, which is why it must be represented accurately. Measurements

give widely varying transport values ranging from 23 Sv in the Yucatan Channel (KANEC campaign [Candela et al, 2003]) to over

33 Sv in the Florida current (determined from voltage measurements using cables between Florida and the Bahamas,

http://www.pmel.noaa.gov/wbcurrents/cabletransport.html).

Past studies have shown that mean transport between Florida and the Bahamas may be estimated at 31.5 ± 2 Sv [Molinari et al.,

1985; Leaman et al., 1987; Schott et al., 1988; Lee et al, 1985; Larsen and Sanford, 1985]. There may be significant daily

fluctuations of up to 10 Sv.

PSY2V1 determined the mean value for 2002 to be 29 Sv

between Florida and the Bahamas (Figure 1), which concords

with mean values found during previous studies. However,

data recorded during the KANEC campaign in the Yucatan

Channel do not fit PSY2V1 and other measurements, including

the voltage measured across the Florida Straits. This raises a

doubt over the validity of these data for it is impossible for 10

Sv to flow between the Bahamas and Cuba.

Compared with the free configuration of PAM, the mean

intensity of flow is similar in both configurations, as it is for

other Caribbean straits which globally concord with

"conventional" data.

Figure 1: PSY2V1 bathymetry and mean

transport in 2002 for A1 analyses

Figure 2 represents the transport time series for 2002. The instantaneous values computed at each time step are in blue and the

values smoothed over seven days are in red. As could be expected, transport varies greatly over time, with occasional

fluctuations over ten Sverdrup.

There is no correlation between the transport computed between the Yucatan Channel ("Cuba" on the graphs) and the entrance

to the Caribbean Sea ("Porto Rico"). The correlation coefficient between these two graphs is 0.12, which is around the same

value as the forced model. This is not therefore an effect of assimilation. We may thus advance two hypotheses on the

representation of circulation in the Caribbean Sea.

Either the Caribbean Sea’s impact on the Florida current has no barotropic element, which could be explained by the size of the

Jamaica Ridge lying between Jamaica and Mexico, or the Florida current is mainly influenced by forcing, and more especially

wind forcing which limits circulation in the Caribbean sea to the East. (In 2002, PSY2V1 computed a correlation of 0.76 between

the entrance to the Caribbean Sea and the Windward Passage).

This lack of correlation can be seen in the smoothed curves of figure 2, which reveals a drop in intensity of transport (around 10

Sv) in the third month for islands in the eastern part of the Caribbean sea and the Windward Passage. The drop in intensity in the

Yucatan and Florida/Bahamas straits is only around 3 Sv.


Figure 2: Time series of instantaneous transport for analysis A1

(blue curve) and smoothed over 7 days (red curve).Transport is

estimated either between an island and the continent or between two

islands.


Variations in the Florida current do not therefore appear in the model as a direct consequence of variations in water transport

from the Caribbean Sea. The process is far more complex due to barocline or vorticity effects [Candela et al., 2003] or even the

major influence of forcing.

Conclusion

Mean transport in the Caribbean in 2003 agrees with past research. Variations also correspond. On the other hand, there is no

apparent correlation in barotropic transport between water flowing from the Caribbean Sea and the intensity of the Florida

current.

The PSY2V1 system predicts the Florida current reasonably accurately. There is quite a significant difference, however, between

the analysis and forecasts for the entrance to the Caribbean Sea. This may be due to proximity to the buffer zone and a badly

represented Brazil current.

Transport through North East Atlantic straits

Circulation through the straits of Iceland and Denmark

There are two straits in the North East Atlantic: the (Denmark

Strait, DKS) and the (Iceland-Scotland Ridge, ISR). These

form the boundary between the North Atlantic subpolar gyre

and the polar gyre of the Greenland and Norwegian Seas

(Figure 3). At the surface, Norway’s warm current—which

extends the North Atlantic drift—flows northward across the

ISR towards the Norwegian coastline. One warm branch also

flows northward through the DKS and reaches the Norwegian

Sea.

The southward circulation of cold water through these straits

causes the North Atlantic Deep Water (NADW) sub-current to

form. The NADW flows from Greenland down along the

American coastline to the South Atlantic.

In PSY2V1, the northern boundary is relatively close to these

sills. Its temperature and salinity are mainly dictated by

seasonal data (Reynaud climatology [1998]). The model’s

representation of circulation North of these sills is not therefore

very realistic. It is, however, useful to analyse the realism of

circulation downstream of these sills and in the Greenland

current that forms the northern part of the subpolar gyre.

To study this circulation in PSY2V1, we selected four sections

(Figure 4), each of which is divided up into two density classes

0<27.8 kg/m³ and 0>27.8 kg/m³ (surface and deep waters

respectively).

Figure 3: Circulation in the North Atlantic. Warm currents are

shown in yellow, orange and red and cold currents in blue,

green and black.

Figure 4: PSY2V1 bathymetry in the North East Atlantic.

Position of the four sections studied in this region: the

Denmark Strait (DKS), Iceland-Scotland Ridge (ISR), East

Greenland (EGD) and South Greenland (SGD).


One-year study of the PSY2V1 system (May 2002-April 2003)

Iceland-Scotland Ridge

In the Iceland-Scotland Ridge, the mean transport taken throughout the depth is a mere 2.1 Sv flowing northward. The North

Atlantic Drift (DNA), a northward flow of warmer surface waters, compensates for the southward transport of dense water from

Northern regions.

The mean annual flow of surface waters over the period in question is estimated at 7.8 Sv with a standard deviation of 4.1. This

mean transport is higher than the 4 Sv estimated by Schmitz and McCartney [1993] but similar to the 7 Sv measured during the

"Student Cruise in Faroe Waters" measurement campaign in July 2001.

PSY2V1 also computes a greater flow of bottom waters across this ridge than that actually measured. Schmitz and McCartney

and the "Student Cruise in Faroe Waters" campaign estimate a bottom water transport value of 3 Sv whereas PSY2V1 considers

there to be 5.6 Sv. The mean annual transport varies between 2.5 and 3.6 Sv, depending on the simulations carried out using the

PAM model without any assimilation.

Denmark Strait

Water masses also flow through the Denmark Strait from the North Atlantic to the Nordic Seas and vice versa. The integrated

transport of the surface layer (density below 27.8 kg/m³) is always northward. Its mean value in our study came to 6.2 Sv for a

standard deviation of 3.9. Different estimations give very different transport statistics for surface waters flowing through the

Denmark Strait. Schmitz and McCartney [1993] consider that mean transport is northward and around 2 Sv, whereas "Student

Cruise in Faroe Waters" campaign measurements indicate a southward flow of 2 Sv.

The PSY2V1 value thus appears higher than these estimations, which would mean a greater flow of warm waters north of the

Denmark Strait. In various simulations using the free PAM model, the mean transport of surface waters is also northward but with

lower values, between 3 and 4 Sv.

For bottom waters, the PSY2V1 model indicates a southward flow of 8.5 Sv for a standard deviation of 3.7 through this strait. This

value is also higher than various observations that generally agree on a mean value of around 3 Sv (Schmitz and McCartney,

1993: 3 Sv, Student Cruise in Faroe Waters, 2001: 3 Sv, Macrander et al., 2003: 3.06±0.10 Sv). This tendency to overestimate

dense water transport also occurs, to a lesser extent, in simulations carried out using PAM, a free model that generates transport

values of around 6 to 7 Sv.

ADCP measurements through this strait [Macrander et al., 2003] indicate that PSY2V1 overestimates variability. The lowest and

highest values in the simulated time series are given as 5 and 14 Sv whereas measurements taken between 1999 and 2002

indicate variations between -2 and 8 Sv.

Greenland current

South of the two straits studied above, the two branches along the Greenland coast provide information on the intensity of the

subpolar gyre. As far as surface waters are concerned, the gyre is formed when the East Greenland current meets the returning

branch of the North Atlantic Drift, but the dense waters flowing through the Denmark Strait and Iceland-Scotland Ridge also

contribute to its formation.

The mean transport in the East Greenland (EGD) current is of 33.5 Sv in all, 24.5 Sv of which is accounted for by surface waters

and 9.1 Sv by denser bottom waters. During this year of simulation, it varied between 13 and 37 Sv. These values fit well with

various estimations of EGD current intensity, such as Aagard and Cochen’s estimation [1968], which limited the EGD current to

isotherm 0°C and gave a total transport of 35 Sv. More recently, Hopkins [1991] estimated a current of between 2 Sv and 32 Sv

in the first 500 metres.

Further south, the SGD section indicates a mean transport of 35.8 Sv, 26 Sv of which is accounted for by surface waters and 9.7

Sv by bottom waters. We cannot account for all the dense waters that flowed through the ISR and DKS (14.1 Sv in all) in this

section. Some of the bottom waters do not follow the coastal current but flow further south and are not counted in the transport

through section SGD, whose boundary is 47°N.


Figure 5: Time series smoothed over seven days for water transport through the four sections

(May 2002 to April 2003). The mean and standard deviation in 2002 are based on the time

series without smoothing.


Conclusion

The PSY2V1 system greatly underestimates the intensity of currents flowing over the Iceland-Scotland Ridge and through the

Denmark Strait compared with actual measurements. This is true for both surface waters—defined herein as having a density

below 27.8 kg/m³—and bottom waters, which have a density above 27.8 kg/m³. This bias already exists in simulations without

data assimilation using PAM, but it is even stronger in analyses using PSY2V1. The PSY2V1 model also exaggerates transport

variability compared with actual measurements. The quantity of dense water coming through the two straits is 14.1 Sv (annual

mean), which indicates an intense bottom water current from the Atlantic along the American coastline. The subpolar gyre formed

by the Greenland current has intense but realistic transport values.

Mercator Ocean newsletter 11

Environment