Quasi-synoptic mesoscale variability in the Balearic Sea

~ Pergamon Deep-Sea Research I, Vol. 41, No. 5/6, pp. 897-914, 1994

Copyright ~ 1994 Elsevier Science Ltd Printed in Great Britain. All rights reserved

09674)637/94 $7.00 + 0.00

Quasi-synoptic mesoscale variability in the Balearic Sea

JEAN-MICHEL PINOT,* JOAQUfN TINTORI~,* and DAMI,~ GOMIS*

(Received 18 March 1993; in revised form 16 August 1993; accepted 15 October 1993)

Abstraet--The circulation and dynamics of the Balearic Sea, a sub-basin of the Western Mediterra- nean located between the Iberian peninsula and the Balearic Islands, are characterized by a significant temporal and spatial variability induced by two well defined shelf/slope fronts over the continental and island slopes. To establish the importance of the mesoscale features frequently observed, a quasi-synoptic AXBT survey of the Balearic Sea was carried out in only three days with a horizontal resolution of 18 km. An objective analysis technique for quantitative scale separation was used to investigate the spatial variability of the temperature field. At the sub-basin scale, the primary feature is the Balcaric Front that separates cold Mediterranean waters from warmer waters of Atlantic origin. The front is located over the islands slope and is characterized by a well defined meandering structure that appears to be strongly correlated with bottom topography. Strong mesoscale activity is also found, mostly in the Balearic frontal region, with two small eddies embedded within the meanders and a warm filament overshooting towards the open sea. These energetic mesoscale features appear in the upper 150 m and strongly modify the almost permanent relatively weak sub-basin scale circulation. The relationship between these instabilities and local topography/coast irregularities is also suggested.

1. INTRODUCTION

HISTORICAL o c e a n o g r a p h i c s tudies of the Wes te rn M e d i t e r r a n e a n have been the resul t of coarse da ta sampl ings ( abou t 40 km) that p r o v i d e d a r a the r genera l desc r ip t ion of the c i rcula t ion (e.g. OVCnINNIKOV, 1966). In the Ba lea r i c Sea, l oca ted be tween the Ibe r i an coast and the Ba lea r i c I s lands (Fig. 1), these s tudies showed the ex is tence of a large scale cyclonic c i rcula t ion with a centra l axis of d ive rgence a l igned with the shape of the basin. The recent accumula t ion of m o r e closely spaced ship da ta and sate l l i te image ry f rom in te rna t iona l e x p e r i m e n t s such as the Wes t e rn M e d i t e r r a n e a n Ci rcu la t ion E x p e r i m e n t (LA VIOLEI~rE, 1987) p r o v i d e d a new view of s t ronger mesosca le var iab i l i ty and more complex genera l c i rcula t ion. FONT et al. (1988) desc r ibed two p e r m a n e n t dens i ty f ronts loca ted over the con t inen ta l s lope ( the Con t inen ta l or Catalf in F ron t ) and islands s lope (the Ba lear ic F ron t ) . T h e y showed that these fronts are the main fea tu res cont ro l l ing the sub-bas in c i rcula t ion in the u p p e r 300 m. A D C P da ta analysis ind ica ted that bo th fronts are assoc ia ted with a long-s lope cur ren ts (CASTELL6N et al. , 1990). M a x i m u m veloci t ies of the o r d e r of 30 cm s 1 were d e t e c t e d in the Catalfin and Ba lear ic Cur ren t s .

A l o n g the mos t ly sal ine Con t inen ta l F ron t , rap id ly evolv ing edd ies and f i laments that s t rongly affect the exchange of shelf and s lope waters were desc r ibed by WANC et al. (1988)

*Dept. de Ffsica, Universitat de les llles Balears, E-07071, Palma de Mallorca, Spain.

897

898 J.-M. PINOT et al.

2£ C.q

J

- 1 ° IT' 1 ° 2 ° 3 ° 4 ° 5 ° 6 °

Longitude o E )

Topography of the Balearic Basin. Fig. 1.

and TINTORI~ et al. (1990). These energetic features were associated with a plume of cold water that originated in the northern Gulf of Lion (LA VIOLETTE et al., 1990). The locations of these instabilities often appeared to be dependent on regional submarine canyons (MAst) and TINTORI~, 1991; TINTORI~ et al., 1990). In contrast to the Continental Front, the Balearic Front is more thermal in nature but has been poorly documented. Examination of satellite imagery (LA VIOLE'IffE et al., 1990) presented a limited view of the front having a wavelike structure.

All these studies showed the existence and the importance of mesoscale features in the Balearic basin. However, the lack of closely-spaced synoptic in situ data has not allowed an accurate characterization of the spatial variability in the whole region. The FE91 field experiment was designed to obtain a quasi-synoptic view of the surface circulation in the Balearic Sea and to assess the impact of the mesoscale structures. The quasi-synoptic data set was obtained in an airborne expendable bathythermograph (AXBT) survey carried out during 22-24 May 1991.

In this study, we focus on the description of the sub-basin scale and mesoscale three- dimensional temperature fields in the upper 200 m, and we apply an objective analysis technique for scale separation and mesoscale structure detection (MADDOX, 1980; GOMlS and ALONSO, 1990). The sub-basin structures are defined in this paper as those present over the whole Balearic basin (i.e. the two fronts basically), while the mesoscale features are those having a characteristic length scale of the order of a few times the internal Rossby radius of deformation (about 10 km in the Balearic Sea), up to 50-60 km.

Quasi-synoptic mesoscale variability in the Balearic Sea 899

2. DATA AND METHODS

The Balearic Sea, roughly 350 × 180 km 2, extends from 38°40'N to 41°40'N and 0°20'W to 4°00'E. In the aircraft survey, 128 AXBTs were dropped, resulting in a dense, homogeneous sampling of the study area. The mean spacing of 18 km provided a high resolution coverage suitable to capture and analyse the mesoscale patterns of the region (Fig. 2).

AXBTs 1-19 were dropped on 22 May over the northern Balearic Sea (between Barcelona and Menorca), AXBTs 20-76 on 23 May northwest of Mallorca, and AXBTs 77-128 on 24 May in the Gulf of Valencia, northwest of Ibiza. AXBT probes were launched every 7 min, the aircraft position being obtained using the onboard navigation system. The success rate was 84% (108 good drops).

The probes used were Sippican Inc. AXBTs (T-4) that reach depths of 460 m. The probes transmitted on three assigned frequencies: 170.5,172.0 and 173.5 MHz. The data acquisition equipment used to record AXBT data was developed by NASA Wallops Flight Facility. It consisted of a special NASA software, a Zenith 80286 IBM-compatible laptop computer, three IIP Joslyn receivers and a NASA provided low-noise GAS-FET pream- plifier and power divider. Radio frequency interference from local FM radio stations in Mallorca was detected during the first day.

Depth values are calculated as a quadratic function of the time of probe fall by Sippican algorithm. The global system accuracy and resolution are, respectively, 0.2°C and 0.01°C for temperature, and 4.6 m (or 2% of depth if greater) and 0.15 m for depth. Segments of bad data due to radio-interference were systematically discarded, and off scale tempera-

42.8 I

I

J 38.4i @.4

~2 • / +'--~<~,

6 * 10 +

\~7 . "~//+ J \:o. "J..>

/ !

Sect~o~ A

.4

Fig. 2. FE91 experiment: A X B T profiles used for objective analysis. Surface intersections of vertical Sections A and B are shown. Ticks are every 0.4 ° latitude and longitude. The rectangular

line shows the analysis grid perimeter.


ture data were removed by filtering. After this preliminary analysis, we kept 100 (of 108) records that we considered good data (Fig. 2). We then used cubic splines interpolation techniques to proceed to vertical interpolating and smoothing of temperature profiles. Temperature values were thus returned at each meter depth and gaps were automatically filled. More information about the data smoothing method can be found in PINOT et al.

(1992).

3. THE OBJECTIVE ANALYSIS TECHNIQUE

Since we are interested in describing the mesoscale variability, we need an objective analysis technique capable of isolating mesoscale features from the large scale back- ground. Because of the homogeneity of the sampling, a simple, isotropic interpolation can be used to analyse the available data set. We chose the Barnes-Maddox analysis technique for quantitative scale separation, which has proved to be effective in mesoscale diagnosis of meteorological and oceanographic fields (TINTOR~ et a l . , 1991). This technique, developed by BARNES (1964, 1973) and modified by MAooox (1980), essentially consists of a two-step univariate scheme based on a numerical filtering built-in in the spatial interpolation algorithm. Bvzzl et al. (1991) designed an additional correction step to reduce the adverse impact of the inhomogeneity of the station distribution, but we considered its use not justified given the homogeneity of our sampling.

The analysis process first consists of a low-pass filter with large wavelength cut-off (Fig. 3), which smoothes the observed data field to define the macroscale. Then, a second low- pass filter with short wavelength cutoff allows the filtering of short scale noise. The band- pass filter thus obtained (defined as the difference between the two low-pass filters) extracts the mesoscale signal. The total field is recovered as the sum of the macroscale and the mesoscale contributions, the noise being filtered throughout the analysis. The shape of each low-pass filter is designed through the tuning of a pair of analysis parameters. Complete description of the analysis scheme is given in the Appendix.

The scale separation has to be determined to design the filters before applying the

?-

c~

,5 ~r ~z

12 t • i o t ~ , t

] . 0

t i . . . . . . . . . . .

0 . 5

3 5 7 0 1 3 0 200

WAVELENGTH (Kin)

Fig. 3. Theoretical response of the objective analysis scheme.


Barnes-Maddox technique. Namely, we have to determine the macroscale and noise cutoff wavelengths (those with a theoretical response equal to 0.5) and the wavelength at which the mesoscale response has to be centered. This apriori decision can be taken either on the basis of theoretical constraints (characteristic scales, deformation radius) or from a previous knowledge of the structures present in the field. In our case, the choice was made with the aim of separating the sub-basin frontal structures with their large scale meandering shape from the shorter scale eddies that should have a characteristic length up to 60 km (Rd about 10 km). We therefore centered the band-pass response of the filtering at a wavelength of 70 km (Fig. 3). Wavelengths at which mesoscale response is 0.5 are 35 km and 130 km. Thus, taking an eddy diameter as half a wavelength, any eddy with a diameter within the range 20-65 km will be isolated and theoretically considered a mesoscale feature. Macroscale features will be those with wavelengths greater than 130 km. With respect to the shortest noisy scales, we considered they were those that could not be resolved by the survey resolution (18 km), so we designed the corresponding low-pass filter with a cutoff about 35-40 km.

Detailed sensitivity analysis was carried out modifying the low-pass filter parameters and consequently the mesoscale filter central wavelength 2o. For 2 o between 60 and 80 km (eddy diameter between 30 and 40 km), no significant changes were observed in the analysis output. For 2o greater than 80 km (which is numerically interesting but physically unrealistic), the mesoscale band was shifted towards sub-basin scale wavelengths and the entire large scale meandering of the fronts was captured in the mesoscale.

The data were objectively analysed onto a 10 km grid. The largest axis of the grid is rotated 62 ° clockwise from north (Fig. 2). Regions of the grid without sampling have been blanked.

4. RESULTS

4.1. The temperature field structure

4.1.1. Basic hydrography. Whereas the baroclinic circulation over the Continental slope of the Iberian peninsula is largely controlled by salinity gradients between fresh Continen- tal waters over the shelf and salty Mediterranean waters offshore, temperature mostly drives the density gradients over the Balearic Islands slope (FONT et al., 1988; ESTRAOA et al., 1989) where the Balearic front can be easily detected in winter by infra-red satellite imagery. Thus, proper interpretation of the thermal patterns obtained from the AXBTs can be used to infer the variability and kinematics of the flow along the Balearic Islands, while we expect a weaker thermal signature of the circulation on the Continental side. The horizontal temperature distributions at different depths, surface, 50 and 125 m (Fig. 4a-c) show the main patterns present in the Balearic Sea. Different water masses can therefore be described:

• Warm waters of Atlantic origin (MAW; 18-20°C at the surface, 13.0-13.5°C at 125 m) that flow northward through the Ibiza Channel (between the peninsula and Ibiza Island) and along the western Balearic Islands shelf/slope region.

• Cooler Mediterranean waters (MW; 16.5-17.5°C at the surface, 12.7-13.0°C at 125 m) that lie in the whole central part of the region. A fraction of these waters flow southward out of the basin, along the Spanish coast, through the Ibiza Channel.

• Continental waters (CW; 15.5-16.0°C at the surface), off Barcelona, spreading

9 0 2 J.-M. PINOT et al.

(a) 4 2 .

38. -0.4

i val = 0 . 2 5 ° C [

i T I 4 . 4

(b) 4 2

Fig. 4.

e r v a l = 0 . 2 * C 3 8

- 0 . 4 4 . 4

Tempera ture field at (a) sea-surface, (b) 50 m, (c) 125 m. In (c), 200 and 1000 m isobath~ are plotted.

(c)

38. -0.4

42.


4 . 4

Fig. 4. (continued)

southward from the northern part of the study area. They originate mainly in the northern Gulf of Lion from river runoff but also from local river discharge.

The MAW are found in the upper 200 m while between 200 and 460 m (not shown) temperature distribution is fairly uniform (12.7-13.0°C) and is characteristic of the denser Intermediate Mediterranean Water. To the north, the thermal signature of CW is only found above the thermocline in the upper 25 m.

4.1.2. Horizontal structures. From the sub-basin scale point of view, a unique coherent pattern is the Balearic Front fully developed over the islands slope and detected at depths between 30 and 150 m (Fig. 4b,c), with a cross-front width of approximately 50 km. The Balearic Front can be defined at 50 m (Fig. 4b) by temperatures higher than 13.8°C. At this depth, it is particularly noticeable that the front presents a sinuous structure and appears as a boundary between the cold MW and the warm MAW. The front has no clear signature in the upper 30 m. The seasonal formation of an upper mixed layer (25 m) due to enhanced seasonal solar heating (end of spring) is responsible for the erosion of the top of the front, which thus becomes less visible at the surface in summer (Fig. 4a). The Balearic Front sharply meanders (M) anticyclonically northwest of Ibiza and this is particularly noticeable at 50 m (Fig. 4b). The length of this meander can be roughly estimated as 120 km and its width as 50 km. This meander presents the strongest temperature gradients observed in the region (I°C in 40 km). Cold MW (13.2-13.6°C at 50 m) is found on the outer edge while warm MAW (13.8-14.8°C at 50 m) is located inshore. This MAW clearly enters the basin from the south through the Ibiza Channel. At the limit of the sampled zone, over the Mallorca-Ibiza sill, the front is deflected back to the north in a cyclonic path. Southwest of


Mallorca, the Balearic Front encounters the island slope and extends northeastward along it. Temperature gradients in this part of the front are similar to those off Ibiza (I°C in 40 kin).

Focusing on the mesoscale, we observe that the open sea MW in the vicinity of the front presents a high mesoscale activity. Eddies of cold water on the open sea edge of the meandering structure (Fig. 4b,c) are correlated with deflections of the front. It is important to note the existence of a cold (12.7°C) cyclonic eddy (El ) , well defined at 125 m (Fig. 4c) that interacts with the Balearic Front. It is centered near 39.6°N-1.8°E and its diameter is roughly 60 km. Another eddy of cold water (E2) is observed northward in a trough of the front off Mallorca. This eddy is also found at 125 m, centered at 40.2°N-2.4°E with temperature in its core about 12.8°C and a diameter near 60 kin.

A westward filament (F) of warm MAW to the open sea off Mallorca (Fig. 4b) seems to be the upper signature of a vertically coherent feature that appears at all depths below the mixed layer. At 25 m (not shown), it can be described as a tongue of warm water (15.0- 15.5°C), 50 km wide, that reaches the center of the basin, 100 km off Mallorca. At 50 m, it has a filament-like narrower shape that penetrates 120 km offshore. At 125 m, we still observe a westward tongue of warm waters (>13.0°C) in the same location, between the two eddies we have just described.

The correlation between the observed structures and the bottom topography is generally high. The meanders of the Balearic Front generally conform to the underlying slope of the Balearic Islands (Fig. 4c). Namely, the sharp anticyclonic meander off Ibiza follows the Ibiza slope bathymetry (200-1000 m); the following cyclonic path corresponds to the eastward penetration of the front over the Mallorca-lbiza sill and its northward deviation over the Mallorca slope, while the above mentioned eddy (E l ) is located directly above the steep topography trough associated with this sill. Finally, the Balearic Front detaches from the slope right to the northwest of Mallorca, in a location where the slope drastically increases (note the convergence of the 200 and 1000 m isobaths). Eddy 2 and the filament are associated with this behaviour of the front.

The reader will note that the AXBT survey apparently did not resolve the Continental Front. As noted in section 4.1.1., this front is of a more saline nature than the Balearic Front (FONT et al., 1988) and horizontal temperature gradients are expected to be much weaker on the continental side. To have a complementary view of the temperature field structure in the western part of the Balearic basin, we plotted the depth of the 13.0°C isotherm (Fig. 5). The Continental Front remarkably appears through the depression of this isotherm that sinks abruptly shoreward below 100 m down to 240 m depth. The front is located over the continental slope (200 and 1000 m isobaths are shown) of the Gulf of Valencia, although it seems to detach in the southern basin.

4.1.3. Vertical structures. Section A (Fig. 6) considerably enlightens the previous observations. It gives a vertical section of the temperature field in the Balearic Front region. Stations 1-4 nearly span the whole Ibiza Channel. The Balearic Front extends vertically from the surface down to 150 m. Near the surface (upper 5 m) sharp gradients coincide with the warmest water (17.5-19.5°C) of the region.

The Ibiza meander (M) is clearly identified from Stas 4-8 where the isotherms are strongly tilted downward (anticyclonic motion). The 13.25°C isotherm penetrates more than 100 m deeper in the core of the meander than in surrounding waters. The sharpest temperature gradients are found between Stas 7 and 8, to the north of the meander. These


42.

I W r :J

38.4-- , ", -0.4

. ~ . . . : :

" i " ~ " " "

~ . o n t o u r i n t e r v a l = 2 0 m ] ' t -

4 . 4

Fig. 5. 13.0°C isotherm depth. Isobaths (200 and 1000 m) are represented.

strong gradients are the expression of the intensification of the Balearic Front due to the presence of a cold cyclonic eddy (El , Stas 7-10) embedded in the front. The vertical temperature distribution shows this eddy extends from the surface down to more than 150 m and that it is much more pronounced below the base of the thermocline (50 m). In its core, at Sta. 9, between 100 and 150 m, a temperature minimum (<13°C) is observed. Its diameter is roughly 60 km as observed in horizontal distributions.

To the north of this eddy (Stas 9-11), isotherms are deflected downward below the thermocline (<13.5°C). This pattern is the expression of the filament (F) previously described, presumably the surface signature of a strong coherent vertical structure. This tongue of warm M A W is embedded between the two cold eddies, E1 to the south and E2 to the north. Eddy 2 is found between Stas 10 and 13 and is similar to eddy 1. As previously, it extends down to 150 m. At the surface, at Sta. 12, relatively cold water (<16.5°C) is apparently upwelled above the core of eddy 2 so as isotherms intersect vertically the surface. The 13.0°C isotherm is also upwelled up to the base of the thermocline. Stations 12-14 intersect again the Balearic Front across the Mallorca slope. Horizontal temperature gradients are weaker than those observed in the Ibiza Channel.

We plotted the same section to 400 m depth (not shown). A constant observation on this section is that the various structures extend down to 300 m but not deeper.

Section B (Fig. 7) gives the tempera ture field structure across the basin. The Balearic Front exhibits the strongest gradients and is found from the surface down to 200 m. On the continental side, the Continental Front is observed only below the thermocline. Although horizontal tempera ture gradients are lower than those on the Balearic slope, the 13.0°C


7

lbiza Channel Ibiza meander eddy filament eddy 2

0 2 3 t 4 5 t 6 7 8 i 10 I1 12 i3 14

" - ~ - - - - ~ , % : , \ ~ " H//!I,. ~ L ~'~./\~,k. ? '

, < ' 7 7i 7i ZOO. ; ' ; : ) ' '

i?. ] :: ', "~ -" 13¢5

] ~ : 5 /,, t .} j.XI,~. i ~- 1~7, v / k t i t/,,/ t / I \ i

i t 2 }Tfl 1,375 1 ":75 1 3 0 )

Fig. 6.

ALONG-SECTION DISTANCE ( Km )

Sect ion A: ver t ical sect ion of the t e m p e r a t u r e field in the Ba lea r i c f ronta l region. C o n t o u r

in terval is 0.25°C. Loca t ion in Fig. 2.

isotherm sinks much more vertically towards the continental shelf, in agreement with the result described in the previous section about the Continental Front.

4.2. Data objective analysis and scale separation

Most of the studies of the circulation in the Balearic Sea have concentrated either on the large scale sub-basin circulation or on the analysis of mesoscale features. We therefore have applied the Barnes-Maddox objective analysis technique to separate the sub-basin scale and mesoscale contributions. In this section, we describe the spatial scales of the structures present in the Balearic Sea with particular emphasis on the interaction between sub-basin scale and mesoscale patterns.

Figure 8 gives the total field, macroscale and mesoscale temperature distributions at 50 m. The analysed total field is almost superposable to the observed field (Fig. 4b), and this confirms the applicability of this type of objective analysis. The standard deviation between observations and analysed total field values at station locations is 0.04°C which implies that the analysis process (and in particular in the noise filtering operation with a cutoff at 40 km) did not remove any structure from the dataset. This is consistent with the fact that the AXBT survey, with its 18-km resolution, could not resolve wavelengths shorter than, say, 30-40 km. The macroscale is clearly identified as the Balearic Front and the associated meandering structure with relatively strong uniform gradients. Note that excluding the Balearic Front region, the temperature distribution is fairly homogeneous in

Q u a s i - s y n o p t i c m e s o s c a l e va r i ab i l i ty in the B a l e a r i c S e a 907

Fig. 7.

ua

15 16 17 18 5

: ~0~

5 0 . , - : '

\ \

\ 150. h ~k\

200 . - - 0.0 5().0

F

h I

i00.0

ALONG-SECTION DISTANCE ( K m )

Sec t ion B: c ross -bas in ver t ica l sec t ion o f the t e m p e r a t u r e field. C o n t o u r in te rva l is 0 .25°C.

L o c a t i o n in Fig. 2.

the rest of the basin. This is typically the kind of picture that was obtained during the coarse resolution (40 km) early cruises in the Balearic Sea. It confirms that these studies could not evidence any mesoscale variability because of their poor resolution. The macroscale distribution stresses the continuity of the front along the islands slope and indicates that at 50 m, warm waters of Atlantic origin flow northward into the Balearic Sea through the whole Ibiza Channel. The sharp meander off Ibiza has been mostly captured in the macroscale though highly smoothed through the analysis.

The mesoscale distribution (Fig. 8c) gives a synoptic view of the small-scale features. It quantifies the mesoscale variability that can be extracted from the total field and gives valuable information on how mesoscale motions can modulate the sub-basin circulation. Mesoscale features have large amplitudes (<0.6°C). The two cold cyclonic eddies E1 and E2 described in Sections 4.1.2. and 4.1.3. appear clearly isolated as the major mesoscale features associated with the Balearic Front. Eddy 2 (40.2°N, 2.4°E), elongated in the N-S direction, exhibits gradients up to I°C over 50 km. Eddy 1 (39.6°N, 1.8°E) has slightly smoother gradients. Both eddies have 50-60 km diameters. Between them lies a filament (F) that extends westward across the basin. This feature seems to be associated with an anticyclonic eddy located over the continental slope. It should be noted that all the mesoscale patterns exhibit gradients larger than those of the macroscale. This implies that the dynamics associated with these features should be more energetic than those of the sub-basin circulation. Another noteworthy pattern is the ribbon of small eddies (30-40 km diameters) located off the continental slope and aligned with it, with the same SW-NE orientation. They are the signature of the Continental Front variability. Note that the

908 J.-M. P1NOT et al.

(a) 4 2 L ,

k

C o n t o u r i n t e rva l = 0.2°C 3 8 . 4 . . . . . . . .

- 0 . 4 .4 (b) 42.

38.4~ -0.4 C o n t o u r in te rva l = 0 .1°C

t ~ T r

4 . 4

Fig. 8. (a) Total, (b) macroscale and (c) mesoscale temperature objective analysis at 50 m. In the mesoscale map, negative contributions are shown by a thin line (zero value contour has been

removed).


(c) 4 2 .

Contour interval = O.I*C 3 8 . 4 - , , , ,

- 0 . 4 4 . 4

F i g . 8. (continued)

characteristic lengths of the mesoscale structures are larger for the Balearic Front (60 km) than for the Continental Front (30 km). Finally, the pattern off Ibiza is the core of the meander which has been extracted through the mesoscale filtering.

The results of the objective analysis at 125 m show that the analysed total field (Fig. 9a) is in good agreement (S. D. = 0.02°C) with the observed field (Fig. 4c); Fig. 9b gives the macroscale contribution. We observe that the global circulation is still represented by the large cyclonic gyre, with a core of cold waters (< 13.0°C) described in earlier studies (FONT and MIRALLES, 1978). The sharp meander off Ibiza also appears while the rest of the Balearic Front has been dislocated northwest of Mallorca. It is interesting to note that the large gyre does not extend to the southern Balearic basin. In fact, in the Gulf of Valencia, the 13.0°C detaches offshore from the continental slope at the same location where the Continental Front was observed to leave the isobaths (Fig. 5). Actually, we do not know which is the partition of the flow in this region.

The mesoscale distribution (Fig. 9c) reveals less patchiness than at 50 m. Also, mesoscale patterns amplitudes (0.25°C) are smaller. The two cold eddies E1 and E2 on the outer edge of the Balearic Front are still detected though gradients are weaker than at 50 m. Between them, the region with anticyclonic pattern (F) corresponds to the filament- like structure described before. In essence, the basin can be divided in two sub-regions of mesoscale activity separated by the central SW-NE axis. The region off the Balearic Islands exhibits the more intense mesoscale activity while we observe a lower number of mesoscale patterns off the peninsula. The previous ribbon of small eddies along the continental slope does not appear, suggesting they were upper layer features only.

910 J.-M. PmOT et al.

(a) 42.

38.4~ -0.4 C o n t o u r i n t e r v a l = O . I ° C

r , i

. 4

(b) 42

38.4~ -0.4

f

! I

!

P i

C o n t o u r i n t e r v a l = O . I ° C

,

4 . 4

Fig. 9. (a) Total, (b) macroscale and (c) mesoscale temperature objective analysis at 125 m. In the mesoscale map, negative contributions are shown by a thin line (zero value contour has been

removed).


(c) 42.

C o n t o u r in te rva l = 0 .05°C 38.4 , ,

-0.4

Fig. 9. (continued)

.4

5. DISCUSSION

5.1. Sub-basin scale dynamics

Over the Balearic slope, a well defined density front separates the warm (and fresh) M A W from the colder (and saltier) MW. A major characteristic that is reinforced here is the shelf-break nature of the front. As shown in previous sections, the mean position of the Balearic Front is generally over the 200-1000 m isobaths. Our observations suggest that the abrupt change of bathymetry associated with the shelf-break along the Balearic Islands exerts a determinant control on the sub-basin circulation trapping the front over the slope.

Several mechanisms are known to be involved in the formation and maintenance of fronts over sloping bot tom though no general agreement has been reached even in a well studied area as the Middle Atlantic Bight slope front. Ou (1983) used a two-dimensional (vertical/cross-shelf) model and suggested that frontogenesis was taking place as a result of geostrophic adjustment. CHAPMAN (1986) used a steady barotropic coastal model and showed how bot tom friction could lead to enhancement of cross-shelf gradients over the slope. These results were more recently extended to a stratified ocean by GAWARKIEWICZ and CHAPMAN (1992) who used a primitive equation model to show that cross-shore bot tom Ekman fluxes could be responsible for formation and maintenance of a front over a shelf-break. In the Balearic basin, as in most oceanic slope regions, it is not yet well established which physical mechanism is responsible for the front formation and maintenance although we have shown that a clear relationship exists between the front location and bot tom topography.


5.2. Mesoscale dynamics

In the Balearic Sea, the circulation at different scales (sub-basin and mesoscale) is clearly coupled, since the highest mesoscale activity is found along the Balearic Front. The mesoscale eddies and filament observed in May 1991 were found near significant local changes of the shelf/slope bathymetry and coastal geometry. Therefore, it is suggested that the local topography and coastline irregularities might be responsible for triggering and/or enhancing the growth of instabilities. Examples of topographically-induced filaments and eddies were given in TINTOR~ et al. (1990) from satellite imagery in the Balearic Sea and in the recent dynamical simulations of HAIDVOGEL et al. (1991) for the California Current filaments. However, the quantitative role of changes in topography or coastline triggering the development of mesoscale filaments remains an unresolved question.

6. S U M M A R Y

The Balearic Sea was surveyed from 22 to 24 May 1991, during the FE91 AXBT survey. The Balearic Front was clearly observed in the upper 150 m. The temperature field was analyzed through an objective technique with scale separation. This separation between mesoscale and macroscale features allowed us to filter the sub-basin scale circulation found in previous studies of the Balearic Sea and to quantify the new information gained by the fine resolution quasi-synoptic AXBT mesoscale survey. Thus, we explicitly showed that a high mesoscale variability is associated with the front and that the mesoscale features strongly modify the relatively weak sub-basin scale circulation. The main mesoscale features detected were two small cold eddies and a well-defined warm filament which might affect the shelf/slope water exchanges. Diameters of the eddies were estimated to be about 50-60 km and the filament length was about 100 km. Also important is that the Balearic Front was observed to be generally located over the Balearic Islands slope, stressing the shelf-break nature of this front. Finally, the location of the filament downstream, a local irregularity of the Mallorca slope, suggests a relation between instabilities and local topographic changes. In any case, it appears that further modeliza- tion of the Balearic Sea should resolve mesoscale motions and take into account the interaction of the circulation with bottom topography.

Acknowledgements--This work is a UIB contribution to the EUROMODEL project funded by thc European Community (DG XII) program MAST (0043-C) and MAS2-CT92-0041. NASA Wallops Flight Facility provided AXBT data acquisition equipment. Expertise of Charles W. Wright and Paul E. La Violette is gratefully acknowledged. We are also particularly thankful to the Spanish Search and Rescue (SAR) 801 Squadron from Son Sant Joan (Palma) for their help and enthusiasm during all the field experiments. Useful discussion with Dong-Ping Wang and Christine Provost strongly improved the quality of the manuscript. We are also thankful to anonymous reviewers for their comments and suggestions. Additional financial support was obtained from CICYT, MAR90-1229-E and CICYT, MAR91-1492-CE. Jean-Michel Pinot acknowledges an E.C. fellowship from the MAST program, S/MAST-913017.

R E F E R E N C E S

BARNES S. L. (1964) A technique for maximizing details in numerical weather map analysis. Journal of Applied Meteorology, 3,396--409.

BARNES S. L. (1973) Mesoscale objective map analysis using weighted timeseries of observations. NOAA Technical Memorandum, ERL NSSL-62, 60 pp.

Buzzl A., D. GOMIS, M. A. PEDDER and S. ALONSO (1991) A method to reduce the inverse impact that


inhomogeneous station distributions have on spatial interpolation, Monthly Weather Review, 119, 2465-

2491. CASTELLON A., J. FONT and E. GAROA (1990) The Liguro-Provenqal-Balearic Current (NW Mediterranean)

observed by Doppler profiling in the Balearic Sea. Sciencias Marinas, 54, 269-276. CHAPMAN D. C. (1986) A simple model of the formation and maintenance of the shelf/slope front in the Middle

Atlantic Bight. Journal of Physical Oceanography, 16, 1273-1279. DOSWELL C. A. (1977) Obtaining meteorologically significant surface divergence fields through the filtering

property of objective analysis. Monthly Weather Review, 105,885-892. ESTRADA M. and J. SALAX (1989) Phytoplankton assemblages of deep and surface water layers in a Mediterranean

frontal zone. Topics in Marine Biology, Sciencias Marinas, 53,203-214. FONT J. and L. MIRALLES (1978) Circulaci6n geostr6fica en el Mar Catalan. Resultados de la Expedicion Cientifica

del Buque Oceanografico Cornide de Saavedra, 7,155-162. FONT J., J. SALAT and J. TINTORE (1988) Permanent features of the circulation in the Catalan Sea. Oceanologica

Acta, S-9, 51-57. GAWARKIEWICZ G. and D. C. CHAPMAN (1992) The role of stratification in the formation and maintenance of

shelf-break fronts. Journal of Physical Oceanography, 22,753-772. GOMtS D. and S. ALONSO (1990) Diagnosis of a cyclogenetic event in the western Mediterranean using an

objective technique for scale separation. Monthly Weather Review, 118,723-736. HAIDVOGEL D. B., A BECKMANN and K. S. HEDSTROM (1991) Dynamical simulations of filament formation and

evolution in the Coastal Transition Zone. Journal of Geophysical Research, 96, 15017-15040. LAVIOLE/~fE P. E. (1987) Portion of Western Mediterranean Circulation Experiment completed. Eos, 68, 123-

124. LAVIOLETrE P. E., J. TINTORE and J. FONT (1990) The surface circulation of thc Balearic Sea. Journal of

Geophysical Research, 95, 1559-1568. MADDOX R. A. (1980) An objective technique for separating macroscale and mesoscale featurcs in mctcorologi-

cal data. Monthly Weather Review, 108, 1108-1163. MASO M. and J. TINTORE (1991) Variability of the shelf water off thc northeast Spanish coast. Journal of Marine

Systems, 1,441-450. Ov H. W. (1983) Some two-laycr models of the shelf-slope front: geostrophic adjustment and its maintenance.

Journal of Physical Oceanography, 13, 1798-1808. OVCrnNNIKOV I. M. (1966) Circulation on the surface and intermediatc layers of the Mediterranean, Oceanology,

6, 48-59. PINOT J. M., J. TINTORE, C. W. WRIGHT and P. E. LA VIOLEa~rE (1992) AXBT data from the Balcaric Sea

experiment, May-June 1991, UIB Technical Report, 60 pp. TINTOR~ J., D. P. WANG and P. E. LA VIOLETrE (1990) Eddies and thcrmohalinc intrusions of thc shclf/slopc front

off the northeast Spanish coast. Journal of Geophysical Research, 95, 1627-1633. TINTORI~ J., D. GOM1S, S. ALONSO and G. PARR1LLA (1991) Mesoscale dynamics and vcrtical motions in thc

Alborfin Sea. Journal of Physical Oceanography, 21, 811-823. WANG D. P., M. VIEIRA, J. SALAT, J. TINTORE and P. E. LA VIOLE~E (1988) A shelf/slope filament off the

northeast Spanish coast. Journal of Marine Research, 46,321-332.

A P P E N D I X ~ T H E B A R N E S - - M A D D O X A N A L Y S I S S C H E M E

The scale separation technique developed by DOSWELL (1977) and MADDOX (198/)) is based on the filtering properties of the objective analysis scheme developed by BARNES (1964, 1973). A brief description is given here for completeness.

A first analysis, f l , operates as a tow-pass filter, smoothing the observed data field to define thc macroscale. A second analysis, f2, with cutoff located near thc smaller resolvable scale is then carried out. The normalized difference between these two low-pass filters (a band-pass filter) is assumed to extract the mesoscale signal. The total field is recovered as the sum of the macroscale and the mesoscale contributions, the short wavelength noise being filtered out by the analysis. Therefore, in addition to the total analysis, two different contributions are obtained, which allows independent analysis of each scale. This scale separation has been used successfully to isolate different sub-synoptic meteorological phenomena (MADDOX, 1980; GOMIS and ALONSO 1990) and mesoscale oceanographic structures (TINTORE, et al., 1991).

Each Barnes' analysis consists of two steps. The grid point values of a first guessf°(i,j) are initially computed from the total N station values f,, as

914 J.-M. PINOT et aL

where weight functions are

N

Z W,~(i,j)fn n=l (A1) f lO( i 'J) N

~ w.(i#) n = l

W,,(i,j) = expl-d,,(i,j)2/4C], (A2)

dn(i,]) is the distance between the grid point (i,j) and the nth station, and C is an analysis parameter. Next, the first guess is evaluated at each station as in (1), givingf~, and the difference Af~ = fn - f c! is then used to obtain the final grid point values:

N

2 w*( i , j )a f .

f(i,j) = f°(i , j) + "=~v , (A3)

Z W~,(i,j) n I

where the modified weight functions are now:

Wn*(i,j) = exp[-d~(i,j)2/4CG], 0 < G < 1, (A4)

G being another analysis parameter. BARNES (1973) showed that the theoretical frequency response of his scheme is a low-pass filter with cutoff location depending on the parameters C and G. Therefore we can choose the values for C and G as those giving a convenient smoothing of the original field.

The mesoscale analysis f•(i,j) is defined as the difference between the two low-pass analysesf2(i,j) andfl(i,j):

fR(i,j) = r l[f2(i,j ) - fj(i,j)], (A5)

where the transmission factor off2 is obviously greater than that off1 at all wavelengths. In the present study fl was obtained with C l = 600 and G 1 = 0.6 and f2 with C 2 = 100 and G 2 = 0.3. The factor r -1 , here r = 0.82, is necessary to ensure a response unity at the wavelength at which the band-pass is centered (MADDOX, 1980). An objective analysis of the actual field (total analysis) is obtained by adding the macroscale contribution J~ (i,j) to the mesoscale contribution fB(i,j). The total field obtained then differs slightly from the field j~ (i#) because of the normalization.

Quasi-synoptic mesoscale variability in the Balearic Sea

Documents