the response of the sao paulo continental shelf, brazil, to synoptic winds

Ocean Dynamics (2009) 59:603614DOI 10.1007/s10236-009-0209-2

The response of the Sao Paulo Continental Shelf,Brazil, to synoptic winds

Marcelo Dottori Belmiro Mendes Castro

Received: 29 April 2008 / Accepted: 18 June 2009 / Published online: 8 July 2009 Springer-Verlag 2009

Abstract The response of the Sao Paulo ContinentalShelf (SPCS) to synoptic wind forcing has been ana-lyzed. Two different methods are used for this purpose,one based on hydrographic data, bottom topography,and geographical characteristics, and a second on an-alyzing currentmeter data directly and using empiri-cal orthogonal functions. Both methods show similarresults for an essentially barotropic shelf. The SPCSresponse in the subinertial frequency band appears tobe trapped on the continental shelf. Numerical exper-iments have also been carried out showing results thatqualitatively agree with the observations, including thevelocity component parallel to the coastline.

Keywords Sao Paulo Continental Shelf Subinertial frequencies Empirical orthogonal functions Currents Winds

Responsible Editor: Richard Signell

Supported by CAPES.

M. Dottori (B) B. M. CastroOceanographic Institute, University of Sao Paulo,Praa do Oceanogrfico, 191. Cidade Universitria,Sao Paulo, 05508-120, Brazile-mail: [email protected]

B. M. Castroe-mail: [email protected]

Present Address:M. DottoriLaboratoire dOcanographie de Villefranche,B.P. 8, Quai de la Darse, Villefranche-sur-Mer, CEDEX,06238, France

1 Introduction

The importance of winds in the marine dynamics onthe Sao Paulo Continental Shelf (SPCSFig. 1) is wellknown (Castro and Lee 1995). With an extension ofabout 400 km, the SPCS is the central part of the SouthBrazil Bight, located off the southeastern Braziliancoast. Its topography is very smooth and the shelf breakis located about 150 km from the coast, at depths ofaround 150 m. The lie of the coast is about 50 east ofreal north, and it is typified by several medium-sizedand small bays, especially in the more easterly portion.One of the first physical studies for this region waspublished by Emilsson (1962). He analyzed 13 days ofdata from the Sao Sebastiao Channel (SSC), locatedalong the coast between the continent and the SaoSebastiao Island, showing that southerly winds forcedalongchannel northeastward currents, while northerlywinds forced currents in the opposite direction. Kvinge(1967), analyzing the same data set, showed that themain subinertial period for these currents was 4 days,almost the same as for the synoptic atmospheric pres-sure oscillations (45 days). Castro (1990) analyzedsimultaneous sea level, wind, and current data, showingthat subinertial currents are the most energetic in theSSC, flowing mainly northeastward during winter in re-sponse to southerly atmospheric synoptic systems (coldfronts).

Two meteorological systems are the main forcingmechanisms for currents in the inner and middleSPCS: the large-scale South Atlantic Subtropical High(SASH) and the meso scale cold front systems (CFS)(Castro and Miranda 1998). The former is a permanentfeature located around 30S, 10W, imparting easterlynortheasterly winds on the SPCS, which are intensified

604 Ocean Dynamics (2009) 59:603614

Fig. 1 SPCS and the locationof currentmeter moorings:PIOF project (P1, P2, andP3) and COROAS project(C1, C2, C3, and C4). PG isthe pressure gauge ofCOROAS

49oW 48oW 47oW 46oW 45oW 44oW26oS

25oS

24oS

23oS

100 km

Sao Sebastiao Island

* Sao Paulo

* Santos* Santos P2 and P3

P1

C1 and C2

C3, C4 and PG

Brazil

during summer. The CFS are synoptic low-pressuresystems usually formed south of Brazil with northeast-ward alongcoast displacement. Winds associated withthe CFS usually blow towards northwestnortheast onthe SPCS. The time interval between two consecutivefrontal systems varies typically between 34 to 1215 days, being lower during spring and winter.

Although the SPCS subinertial response to synopticwinds has been studied before, it is not very clearwhether this response is mainly barotropic or baro-clinic, especially during summer when high stratifica-tion is observed due to a seasonal thermocline. Themain purpose of this study is to analyze if the SPCSsubinertial currents, in response to synoptic winds, areprimarily barotropic or baroclinic. Two different meth-ods are used: one applying the model proposed byClarke and Brink (1985) (hereafter CB-1985), whichrequires the BruntVaisala frequency and topographydata, and another analyzing currentmeter data, includ-ing empirical orthogonal functions (EOF). The semi-

Table 1 Characteristics of the currentmeter data from the PIOFproject

Local Currentmeter Length Perioddepth depth (m) (hours)(m)

P1 40 10, 20, and 30 2,048 06/Jul/1989 to 29/Sep/1989P2 70 10, 35, and 60 2,043 16/Dec/1988 to 11/Mar/1989P3 70 10, 35, and 60 2,048 04/Jul/1989 to 27/Sep/1989

P1, P2, and P3 are mooring positions indicated in Fig. 1. Samplingrate was 1 h

analytical numerical model proposed in CB-1985 hasalso been applied to average water properties.

The data set used in this study will be described inSection 2. Section 3 gives a brief explanation of thetwo methods, CB-1985s model and EOF applied tovectors. Data analysis results are given in Section 4and numerical modeling will be presented in Section 5.Discussion and conclusions appear in Section 6.

2 Data set

Two different projects, Integrated Project/PhysicalOceanography (PIOF) and Oceanic Circulation in theWestern Part of the South Atlantic (COROAS), pro-vided data for this study. PIOF field work took place

Table 2 Characteristics of the currentmeter data from theCOROAS project

Local depth Currentmeter Length Period(m) depth (m) (hours)

C1 100 30, 58, and 91 1,736 11/Jul/1993to 21/Sep/1993

C2 100 30, 58, and 91 2,041 22/Sep/1993to 20/Dec/1993

C3 200 31, 74, 127, 2,178 15/Dec/1992and 190 to 20/Mar/1993

C4 200 31, 74, 127, 2,160 21/Mar/1993and 190 to 20/Jun/1993

C1, C2, C3, and C4 are mooring positions indicated in Fig. 1.Sampling rate was 1 h

Ocean Dynamics (2009) 59:603614 605

Fig. 2 Hydrographic stationsof the PIOF project. Themain characteristics of thesurveys are shown in Table 3.The island labeled ISS is SaoSebastiao, shown in Fig. 1

H1

SPRJ

ISS

200

m

H2

SPRJ

ISS

200

m

H3

SPRJ

ISS

200

m

H4

SPRJ

ISS

200

m

H5

SPRJ

ISS

200

m

H6

SPRJ

ISS

200

m

H7

SPRJ

ISS

200 m

from October 1985 to March 1989, providing closelyspaced hydrographic data, as well as the currentmeterdata for the 40-m (inner shelf) and 70 m (middle shelf)

Table 3 Characteristics of the hydrographic surveys from theproject PIOF

Survey Period Number Numberof stations of radials

H1 December 1985, 1114 64 6H2 July 1986, 2024 70 6H3 December 1986, 1116 61 6H4 July 1987, 0914 72 6H5 December 1987 1116 60 5H6 July 1988, 0206 70 5H7 December 1988, 1519 59 5

Station locations are shown in Fig. 2

isobaths. COROAS provided currentmeter data for the100-m (middle shelf) and 200-m (outer shelf) isobaths,besides sea level data for the coast and shelf break.

Table 4 Mean BruntVaisala frequency squared of the surveysshown in Fig. 2

Survey BruntVaisala Seasonfrequency squared(105s2)

H1 1.65 0.12 Summer 1985/1986H2 0.37 0.02 Winter 1986H3 1.37 0.07 Summer 1986/1987H4 0.63 0.03 Winter 1987H5 1.43 0.14 Summer 1987/1988H6 1.10 0.24 Winter 1988H7 1.46 0.18 Summer 1988/1989

606 Ocean Dynamics (2009) 59:603614

Table 5 Average values obtained for each season for the BruntVaisala frequency squared and Eqs. 1 and 2

Season N2(105s2) N22f 2

Nf

Winter 0.70 0.05 (3.6 0.6) 104 0.019Summer 1.48 0.15 (1.69 0.27) 104 0.013Average w/s 1.09 0.10 (2.6 0.4) 104 0.016

2.1 PIOF currentmeter data

The PIOF currentmeter data were collected from sur-face (40 m) and subsurface (70 m) moorings instru-mented with three currentmeters each: close to themixing layer bottom, within the thermocline, and closeto the bottom boundary layer top. Five consecutiveseasons, three winters and two summers, were surveyedhourly, but only the three seasons with higher datareturn are used in this study. Currentmeters used wereSensordata, model SD2000, with a precision of 2 102 m/s for the amplitude and resolution of 15 fordirection. Table 1 shows the main features and Fig. 1the locations of the PIOF moorings.

2.2 COROAS currentmeter data

Four moorings from COROAS, located in the middleshelf (100 m) and outer shelf (200 m, near the shelfbreak), are used in this study. The 100-m moorings wereinstrumented with three Sensordata SD2000 current-meters, with positions along the water column similarto the 70-m moorings from PIOF. The 200-m moor-ings had four Sensordata SD6000 currentmeters, theadditional currentmeter being located at mid-depths.Sampling interval was 1 h. Sensordata SD6000 cur-rentmeters have the same precision as the SD2000 forspeeds, but a 7.5 resolution for directions. Table 2shows the main features and Fig. 1 the locations of theCOROAS moorings.

2.3 Hydrographic data

Hydrographic data were collected during seven con-secutive summer and winter cruises, being four inDecember (austral summer) and three in July (austral

Table 6 Variance for the sea level data, in meters squared, at thecoast and at the shelf break

Location Original series Filtered series

Coast 0.1376 0.032Shelf break 0.0895 0.002

Note that the time series used to compute the values in the thirdcolumn were filtered using a low-pass Lanczos filter with a cuttingfrequency at subinertial period

Table 7 Variance of the EOFs at each depth of moorings P1, P2,and P3 and mean variance at each mooring

Mooring Empirical mode Surface Mid-depth Bottom Total

P1 1st 93% 96% 86% 93%2nd 7% 2% 12% 6%3rd

Ocean Dynamics (2009) 59:603614 607

calculating N. The currentmeter data have been low-pass filtered using a Lanczos squared filter for removingtidal oscillations and preserving the subinertial oscil-lations (Walters and Heston 1982). Velocity vectorshave been decomposed in the alongshelf and cross-shelf directions using the local isobath orientation. EOFanalysis in the time domain has also been applied to thecurrentmeter velocity vectors.

4 Data analysis

4.1 Hydrographic dataClarke and Brink (1985)model

The BruntVaisala frequency squared, N2, was calcu-lated using the discretized form:

N2 g

z= g

(2 1)(z2 z1) , (3)

where is the vertically averaged density, g is the grav-ity acceleration, and 1 and 2 are densities at specificdepths z1 and z2, respectively. For each continentalshelf station (depth less than or equal to 200 m), the

BruntVaisala frequency was calculated for adjacentdepths. All the calculated values were used to estimatean average BruntVaisala frequency for each cruise.Results shown in Table 4 are consistently and consid-erably higher for summer than for winter, the summeraverage being about twice as high as that for the winter(Table 5). Although the stratification presents smallvariability over summer, it is clear that, for winter, thestratification increases considerably from 1986 to 1988.Castro (1996) has shown that the summer stratificationincrease is due mostly to buoyancy fluxes associated tothe seasonal subsurface South Atlantic Central Water(SACW) intrusions towards the coast. SACW intru-sions are forced by the easterlynortheasterly winds,varying from year to year.

A cross section running from So Sebastio Islandto the shelf break was selected as representative forthe main bathymetric features of the region. The shelfslope calculated using a linear regression of thebathymetry is 0.0014 0.0002. The estimated Coriolisparameter at 23.5S was 5.8 105 s1. Using those lasttwo values and seasonal averages of N2, the products inEqs. 1 and 2 were computed. The results are shown inTable 5.

Fig. 3 EOF for the PIOFmoorings. From left to right,the first, second, and thirdmodes and, from top tobottom, moorings P1, P2,and P3 (Fig. 1 and Table 1).Vector orientations arerelative and not necessarilyassociated to a geographicalorientation. The darker thearrow, the deeper thecomponent. Components arenormalized by the biggestamong them

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

P1

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

P2

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

P3

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

608 Ocean Dynamics (2009) 59:603614

The smallness of the products indicates a continentalshelf with a barotropic response to the subinertial windforcing for both winter and summer, even the strati-fication being higher in the warm season. The morerestrictive condition shows that subinertial motions aretrapped on the continental shelf, which is confirmedby the sea level data analysis. Sea level variances cal-culated for the low-pass filtered series from both mea-surement points show that, at the coast, the variance isabout 16 times greater than at the shelf break (Table 6).

4.2 Currentmeter dataEOF analysis

4.2.1 PIOF data

Results from PIOF data show a strong predominanceof the first empirical mode (Table 7) with a total vari-ance superior to 90%. The first EOF at all depthsand moorings (Fig. 3) do not show much variationin directions and in amplitudes, thus resembling thebarotropic mode. It is important to notice that themethod allows determining only the relative vector di-rection for each EOF, and not the geographical vectororientation. In other words, it is possible to charac-

3

2

1

0

1

2

no

rma

lized

am

plitu

de

190 200 210 220 230 240 250 260 270day from January 1st 1989

3

2

1

0

1

2


3

2

1

0

1

2


parallel wind30m current1st PC

40

30

20

10

0

10

20

30

alo

ngsh

ore

curre

nt (c

m/s)

40

30

20

10

0

10

20

30

40

30

20

10

0

10

20

30 10 meters20 meters30 meters

Fig. 4 Low-passed alongshelf current time series at P1(top panel). Low-passed wind stress from NCEP/NCAR(http://www.cdc.noaa.gov), low-passed alongshelf current at P1near bottom currentmeter, and the first principal component(PC) obtained with the EOF analysis (bottom panel) associatedwith the first EOF (Fig. 3). All time series in the bottom panel arenormalized by the standard deviation. The correlation betweenthe first PC and the alongshelf current is 0.96 and above 99% ofsignificance, and the correlation between the wind stress and thefirst PC is 0.42, above 95% of significance

terize a barotropic response but not the direction ofthe barotropic currents. Figures 4, 5, and 6 show time-series for the alongshelf low-passed currents from themooring data, for the first principal component, andfor the average alongshelf wind-stress obtained fromNCEP/NCAR (Kalnay et al. 1996) at the grid points25S and 42.5W and 25S and 45W near to themooring sites. The EOF analysis produces a principalcomponent time series that is complex with the real partcorresponding to the alongshelf velocity and the imag-inary part rotating the EOF in 90. However, the realpart variability corresponds to more than 80% of thetotal variability of the first EOF at all PIOF mooringsand, then, we will neglect the imaginary part associatedwith the cross-shelf component. Correlations betweenthe real part of the first principal component and thealongshelf velocity is high, above 0.95, for all moorings.Therefore, directions of the first EOF are almost par-allel to the alongshelf directions. Correlations betweenthe real part of the first principal component and thelow-passed alongshelf wind-stress is also relatively high,showing that the subinertial alongshelf variabilities areforced by the alongshelf winds.

The first empirical mode explains more of the cur-rent variability at intermediate depths than at other

2

1

0

1

2

3

4

norm

aliz

ed a

mpl

itude

10 0 10 20 30 40 5 0day from January 1st 1989

2

1

0

1

2

3

4


2

1

0

1

2

3

4



40

30

20

10

0

10

20

alo

ngsh

ore

curre

nt (c

m/s)

40

30

20

10

0

10

20

40

30

20

10

0

10

20 10 meters35 meters60 meters

Fig. 5 Low-passed alongshelf current time series at P2(top panel). Low-passed wind stress from NCEP/NCAR(http://www.cdc.noaa.gov), low-passed alongshelf current at P2near bottom currentmeter, and the first principal component(PC) obtained with the EOF analysis (bottom panel) associatedwith the first EOF (Fig. 3). All time series in the bottom panel arenormalized by the standard deviation. The correlation betweenthe first PC and the alongshelf current is 0.92 and between thefirst PC and the wind stress is 0.62, both above 99% of significance

Ocean Dynamics (2009) 59:603614 609

2

1

0

1

2

3

norm

aliz

ed a

mpl

itude

190 200 210 220 230 240 250 260 270

day from January 1st 1989

2

1

0

1

2

3

norm

aliz

ed a

mpl

itude

190 200 210 220 230 240 250 260 270


2

1

0

1

2

3

norm

aliz

ed a

mpl

itude

190 200 210 220 230 240 250 260 270



40 30 20 10

01020304050607080

alo

ngsh

ore

curre

nt (c

m/s)

40 30 20 10

01020304050607080

alo

ngsh

ore

curre

nt (c

m/s)

40 30 20 10

01020304050607080

alo

ngsh

ore

curre

nt (c

m/s)

10 meters35 meters60 meters

Fig. 6 Low-passed alongshelf current time series at P3(top panel). Low-passed wind stress from NCEP/NCAR(http://www.cdc.noaa.gov), low-passed alongshelf current at P3near bottom currentmeter and the first principal component (PC)obtained with the EOF analysis (bottom panel) associated withthe first EOF (Fig. 3). All time series in the bottom panel arenormalized by the standard deviation. The correlation betweenthe first PC and the alongshelf current is 0.97 and between thefirst PC and the wind stress is 0.7, both above 99% of significance

depths, since the former position is more representativeof the geostrophic interior, being far from the top andbottom Ekman layers. Table 7 also suggests that thesecond empirical mode corresponds to the first baro-

clinic mode, as confirmed by Fig. 3. This second modeis more important for the near surface and near bottomcurrents, but it is of almost no significance at intermedi-ate depth. The third empirical mode corresponds to thesecond baroclinic mode, but its total variance is almostzero.

4.2.2 COROAS data

General characteristics of COROAS and PIOF cur-rentmeter data set are similar. COROAS first empiricalmode accounts for most of the total variance and alsocorresponds to the barotropic mode, as seen in Figs. 7and 8. It is also clear that the barotropic mode is almostparallel to the alongshelf velocity. For moorings C1and C2, there is also high correlation between the low-passed wind-stress and the barotropic mode (Figs. 9, 10,11, and 12). There are also some differences betweenthe properties of the COROAS and the PIOF data.First, the variance of the first EOF at C1 is the smallestof all the moorings, and this EOF accounts for less than90% of the variability, although being the predominantmode (Table 8). Second, amplitudes of near bottomcurrents at C3 and C4 are considerably smaller butwith the same direction as those of the other threecurrentmeters (Table 9). Also at C3 and C4, the firstEOF corresponds to the barotropic mode, but the cor-relation between the low-passed alongshelf wind stressand current is either low (moorings C3 and C4) and/orinsignificant (mooring C3), showing that the wind stressrole in forcing alongshelf currents is reduced. It isimportant to remember that those two moorings are

Fig. 7 EOF for COROASmoorings. From left to right,the first, second, and thirdmodes and, from top tobottom, moorings C1 and C2(Fig. 1 and Table 2). Vectororientations are relative andnot necessarily associated to ageographical orientation. Thedarker the arrow, the deeperthe component. Componentsare normalized by the biggestamong them

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

C1

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

C2

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

610 Ocean Dynamics (2009) 59:603614

Fig. 8 EOF for COROASmoorings. From left to right,the first, second, third, andfourth modes and, from top tobottom, moorings C3 and C4(Fig. 1 and Table 2). Vectororientations are relative andnot necessarily associated to ageographical orientation. Thedarker the arrow, the deeperthe component. Componentsare normalized by the biggestamong them

-1

-0.50

0.51

-1 -0.5 0 0.5 1-1

-0.50

0.51

-1 -0.5 0 0.5 1-1

-0.50

0.51

-1 -0.5 0 0.5 1-1

-0.50

0.51

-1 -0.5 0 0.5 1

-1

-0.50

0.51

-1 -0.5 0 0.5 1-1

-0.50

0.51

-1 -0.5 0 0.5 1-1

-0.50

0.51

-1 -0.5 0 0.5 1-1

-0.50

0.51

-1 -0.5 0 0.5 1

located near the shelf break, where the influence of thewestern boundary Brazil Current is determinant andthe CB-1985 theory does not necessarily apply.

The second empirical mode explains between 4%and 5% of the total variance, with the exception ofmooring C1, where it explains 13%. At the near bottomC3 and C4 data, the second empirical mode accountsfor 58% of the variance. Even with this inversion, thefirst empirical mode still accounts for most (95%) of the

2

1

0

1

2

no

rma

lized

am

plitu

de


2

1

0

1

2


2

1

0

1

2



40

30

20

10

0

10

20

30

40

alo

ngsh

ore

curre

nt (c

m/s)

40

30

20

10

0

10

20

30

40

40

30

20

10

0

10

20

30


Fig. 9 Low-passed alongshelf current time series at C1(top panel). Low-passed wind stress from NCEP/NCAR(http://www.cdc.noaa.gov), low-passed alongshelf current at C1near bottom currentmeter and the first principal component (PC)obtained with the EOF analysis (bottom panel) associated withthe first EOF (Fig. 7). All time series in the bottom panel arenormalized by the standard deviation. The correlation betweenthe first PC and the alongshelf current is 0.93 and between thewind stress and the first PC is 0.65, both above 99% of significance

total variance at those moorings (Table 9). The otherempirical modes explain only a minor portion of thevariance.

5 Numerical modelsome experiments

The CB-1985 model has been applied to the SPCS.Model inputs are: BruntVaisala frequency vertical

2

1

0

1

2

3

no

rma

lized

am

plitu

de


2

1

0

1

2

3

no

rma

lized

am

plitu

de


2

1

0

1

2

3

no

rma

lized

am

plitu

de


2

1

0

1

2

3

no

rma

lized

am

plitu

de



50

40

30

20

10

0

10

20

30

alo

ngsh

ore

curre

nt (c

m/s)

50

40

30

20

10

0

10

20

30

alo

ngsh

ore

curre

nt (c

m/s)

50

40

30

20

10

0

10

20

30

alo

ngsh

ore

curre

nt (c

m/s)

50

40

30

20

10

0

10

20

30

alo

ngsh

ore

curre

nt (c

m/s)

50

40

30

20

10

0

10

20

30

alo

ngsh

ore

curre

nt (c

m/s)

50

40

30

20

10

0

10

20

30

alo

ngsh

ore

curre

nt (c

m/s)


Fig. 10 Low-passed alongshelf current time series at C2(top panel). Low-passed wind stress from NCEP/NCAR(http://www.cdc.noaa.gov), low-passed alongshelf current at C2near bottom currentmeter and the first principal component (PC)obtained with the EOF analysis (bottom panel) associated withthe first EOF (Fig. 7). All time series in the bottom panel arenormalized by the standard deviation. The correlation betweenthe first PC and the alongshelf current is 0.96 and between thewind stress and the first PC is 0.57, both above 99% of significance

Ocean Dynamics (2009) 59:603614 611

2

1

0

1

2

3

4

norm

aliz

ed a

mpl

itude

20 10 0 10 20 30 40 50 60 7day from January 1st 1993

2

1

0

1

2

3

4

norm

aliz

ed a

mpl

itude

20 10 0 10 20 30 40 50 60 0day from January 1st 1993

2

1

0

1

2

3

4

norm

aliz

ed a

mpl

itude



8070605040302010

0102030

alo

ngsh

ore

curre

nt (c

m/s)

8070605040302010

0102030

8070605040302010

0102030

8070605040302010

0102030

31 meters74 meters127 meters190 meters

80

Fig. 11 Low-passed alongshelf current time series at C3(top panel). Low-passed wind stress from NCEP/NCAR(http://www.cdc.noaa.gov), low-passed alongshelf current at C3near bottom currentmeter and the first principal component (PC)obtained with the EOF analysis (bottom panel) associated withthe first EOF (Fig. 8). All time series in the bottom panel arenormalized by the standard deviation. The correlation betweenthe first PC and the alongshelf current is 0.98 above 99% ofsignificance and between the wind stress and the first PC is 0.29and below of 90% of significance

profiles, Coriolis parameter, bottom topography, windstress intensity, direction and frequency, and bottomfriction coefficient. Wind stress was parameterizedusing:

= airCDV2, (4)where is the wind stress, air is the air density, CD isthe drag coefficient 10 m above the sea surface, set as1.2 103, and V is the wind speed. The BruntVaisalafrequency profile was determined using the PIOF hy-drographic data set and sigma coordinates to smoothdepth variations (Fig. 13). The cross shelf bathymetric

Table 8 Variance of the EOFs at each depth of moorings C1 andC2 and mean variance at each mooring

Mooring Empirical 30 m 58 m 91 m Totalmode

C1 1st 77% 94% 82% 84%2nd 21%

612 Ocean Dynamics (2009) 59:603614

-1.0

-0.5

0.0

0 1 2-1.0

-0.5

0.0

0 1 2Brunt-Vaisala frequency (X10-5 s-2)

dept

h (si

gma c

oordi

nates

)

Fig. 13 Profile of BruntVaisala frequency used as input for thenumerical model

5.1 Experiments

The first three experiments were forced by alongshelfwind stress and the fourth by cross-shelf wind stress(Table 10). Initially, we ran the model for two typicalCFS situations at the SPCS: winds almost parallel to thecoastline and amplitude varying between 8 and 11 m/s.The third experiment is forced by a relatively weakwind stress, as does the first experiment, but the bottomfriction is slightly less than the first two experiments inorder to estimate how important the friction is. Thefourth experiment uses a relatively strong cross-shelfwind stress and relatively low bottom friction. The pur-pose of this last experiment is to verify the influence ofcross-shelf winds, although this is not the typical regimein the SPCS.

Table 10 Main characteristics of the 4 performed numericalexperiments: wind direction and speed, wind stress and bottomfriction coefficient

Experiment Wind Speed Stress (m/s)direction (m/s) (N/m2)

1 Parallel 8 0.096 0.0012 Parallel 11 0.1815 0.0013 Parallel 8 0.096 0.00094 Perpendicular 11 0.1815 0.0009

5.2 Model results

Experiment 1 shows weak cross-shelf currents ( /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 150 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 150 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org?) /PDFXTrapped /False

/SyntheticBoldness 1.000000 /Description >>> setdistillerparams> setpagedevice

the response of the sao paulo continental shelf, brazil, to synoptic winds

Documents

university of sao paulo

barotropic shelf

shelf breakis

sao sebastiao channel

current data

data set

hydrographic data

synoptic wind forcing