Sub-surface small-scale eddy dynamics from multi-sensor observations and modeling

18
Sub-surface small-scale eddy dynamics from multi-sensor observations and modeling J. Bouffard a,, L. Renault a,b , S. Ruiz a , A. Pascual a , C. Dufau c , J. Tintoré a,b a IMEDEA (CSIC-UIB), TMOOS Dept., Mallorca, Spain b SOCIB, Coastal Ocean Observing and Forecasting System, Spain c CLS Space Oceanography Division, Ramonville, France article info Article history: Received 4 February 2011 Received in revised form 6 June 2012 Accepted 15 June 2012 Available online 20 July 2012 abstract The study of mesoscale and submesoscale [hereafter (sub)mesoscale] hydrodynamic features is essential for understanding thermal and biogeochemical exchanges between coastal areas and the open ocean. In this context, a glider mission was conducted in August 2008, closely co-located and almost simulta- neously launched with a JASON 2 altimetric pass, to fully characterize the currents associated with regio- nal (sub)mesoscale processes regularly observed to the north of Mallorca (Mediterranean Sea). A synoptic view from satellite remote-sensing fields, before and during the glider mission, provided a descriptive picture of the main surface dynamics at the Balearic Basin scale. To quantify the absolute surface geo- strophic currents, the coastal altimetry-derived current computation was improved and cross-compared with its equivalent derived from glider measurements. Model simulations were then validated both qualitatively and statistically with the multi-sensor observations. The combined use of modeling and multi-sensor observational data reveals the baroclinic structure of the Balearic Current and the Northern Current and a small-scale anticyclonic eddy observed northeast of the Mallorca coast (current 15 cm/s, <30 km in extent and >180 m deep). This mesoscale structure, partially intercepted by the glider and along-track altimetric measurements, is marked by relatively strong salinity gradients and not, as is more typical, temperature gradients. Finally, the use of the validated model simulation also shows that the geostrophic component of this small-scale eddy is controlled by sub-surface salinity gradients. We hypothesize that this structure contains recently modified Atlantic water arriving from the strait of Ibiza due to a northerly wind, which strengthens the northward geostrophic circulation. Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction Mesoscale and submesoscale [hereafter (sub)mesoscale] hydro- dynamic features are particularly important in establishing and understanding the horizontal and vertical transport of heat (Volkov et al., 2008) and biogeochemical tracers (McGillicuddy et al., 1998; Lévy, 2008 and references therein). Indeed, 90% of the kinetic en- ergy of ocean circulation is contained in small-scale features (e.g., eddies, fronts and filaments), whereas 50% of the vertical exchange of water mass properties between the upper and the deep ocean may occur at the (sub)mesoscale (Fu et al., 2010a, 2010b). Extend- ing our knowledge of the formation, evolution and dissipation of eddy variability is therefore critical to understanding the ocean’s roles in the earth’s climate. This challenge highlights the impor- tance of describing complex ocean dynamics using both theoretical and modeling approaches (Klein and Lapeyre, 2009; Capet et al., 2008a; 2008b; 2008c; Hu et al., 2009) and observational ap- proaches adapted to the coastal domain (Nencioli et al., 2011; Bouffard et al., 2010; Hu et al., 2011; Garreau et al., 2011). Eddy signatures in terms of sea surface height have revealed that multi-satellite altimetry is highly effective in observing and tracking eddies in the global ocean due to its ability to provide al- most synoptic and periodic measurements of the sea surface topography (Chelton et al., 2011; Morrow and Le Traon, 2011). However, given their relatively limited resolution, the use of stan- dard geostrophic current maps derived from multi-satellite altim- etry alone (Ducet et al., 2000) is not always sufficient to characterize the spatial and temporal variability of (sub)mesoscale 0079-6611/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.pocean.2012.06.007 Abbreviations: BC, Balearic Current; Corr, correlation with statistical signifi- cance > 95%; DH, dynamic height derived from the glider CTD; IGDR, Interim Geophysical Data Record; IMOS, Integrated Marine Observing System; MDT, Mean Dynamic Topography; MOOSE, Mediterranean Ocean Observing Site for Environ- ment; (M)SLA, Map of Sea Level Anomaly; NC, Northern Current; OOI, Ocean Observatories Initiative; SLA, Sea Level Anomaly; SOCIB, Coastal Ocean Observing and Forecasting System (translated from Spanish to English); STD, standard deviation; SWOT, Surface Water and Ocean Topography. Corresponding author. Present addresses: Aix-Marseille Univ., Mediterranean Institute of Oceanography (MIO), UMR 7294, CNRS/INSU, UMR 235, IRD, 13288, Marseille, Cedex 09, France and Université du Sud Toulon-Var (MIO), 83957, La Garde cedex, France. Tel.: +33 491829108. E-mail address: [email protected] (J. Bouffard). Progress in Oceanography 106 (2012) 62–79 Contents lists available at SciVerse ScienceDirect Progress in Oceanography journal homepage: www.elsevier.com/locate/pocean

Transcript of Sub-surface small-scale eddy dynamics from multi-sensor observations and modeling

  • Sub-surface small-scale eddy dynamics fro

    Du

    Article history:Received 4 February 2011Received in revised form 6 June 2012Accepted 15 June 2012Available online 20 July 2012

    Bouffard et al., 2010; Hu et al., 2011; Garreau et al., 2011).Eddy signatures in terms of sea surface height have revealed

    that multi-satellite altimetry is highly effective in observing andtracking eddies in the global ocean due to its ability to provide al-most synoptic and periodic measurements of the sea surfacetopography (Chelton et al., 2011; Morrow and Le Traon, 2011).However, given their relatively limited resolution, the use of stan-dard geostrophic current maps derived from multi-satellite altim-etry alone (Ducet et al., 2000) is not always sufcient tocharacterize the spatial and temporal variability of (sub)mesoscale

    Abbreviations: BC, Balearic Current; Corr, correlation with statistical signi-cance > 95%; DH, dynamic height derived from the glider CTD; IGDR, InterimGeophysical Data Record; IMOS, Integrated Marine Observing System; MDT, MeanDynamic Topography; MOOSE, Mediterranean Ocean Observing Site for Environ-ment; (M)SLA, Map of Sea Level Anomaly; NC, Northern Current; OOI, OceanObservatories Initiative; SLA, Sea Level Anomaly; SOCIB, Coastal Ocean Observingand Forecasting System (translated from Spanish to English); STD, standarddeviation; SWOT, Surface Water and Ocean Topography. Corresponding author. Present addresses: Aix-Marseille Univ., Mediterranean

    Institute of Oceanography (MIO), UMR 7294, CNRS/INSU, UMR 235, IRD, 13288,Marseille, Cedex 09, France and Universit du Sud Toulon-Var (MIO), 83957, LaGarde cedex, France. Tel.: +33 491829108.

    Progress in Oceanography 106 (2012) 6279

    Contents lists available at

    Progress in Oc

    journal homepage: www.elsE-mail address: [email protected] (J. Bouffard).1. Introduction

    Mesoscale and submesoscale [hereafter (sub)mesoscale] hydro-dynamic features are particularly important in establishing andunderstanding the horizontal and vertical transport of heat (Volkovet al., 2008) and biogeochemical tracers (McGillicuddy et al., 1998;Lvy, 2008 and references therein). Indeed, 90% of the kinetic en-ergy of ocean circulation is contained in small-scale features (e.g.,

    eddies, fronts and laments), whereas 50% of the vertical exchangeof water mass properties between the upper and the deep oceanmay occur at the (sub)mesoscale (Fu et al., 2010a, 2010b). Extend-ing our knowledge of the formation, evolution and dissipation ofeddy variability is therefore critical to understanding the oceansroles in the earths climate. This challenge highlights the impor-tance of describing complex ocean dynamics using both theoreticaland modeling approaches (Klein and Lapeyre, 2009; Capet et al.,2008a; 2008b; 2008c; Hu et al., 2009) and observational ap-proaches adapted to the coastal domain (Nencioli et al., 2011;0079-6611/$ - see front matter 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.pocean.2012.06.007The study of mesoscale and submesoscale [hereafter (sub)mesoscale] hydrodynamic features is essentialfor understanding thermal and biogeochemical exchanges between coastal areas and the open ocean. Inthis context, a glider mission was conducted in August 2008, closely co-located and almost simulta-neously launched with a JASON 2 altimetric pass, to fully characterize the currents associated with regio-nal (sub)mesoscale processes regularly observed to the north of Mallorca (Mediterranean Sea). A synopticview from satellite remote-sensing elds, before and during the glider mission, provided a descriptivepicture of the main surface dynamics at the Balearic Basin scale. To quantify the absolute surface geo-strophic currents, the coastal altimetry-derived current computation was improved and cross-comparedwith its equivalent derived from glider measurements. Model simulations were then validated bothqualitatively and statistically with the multi-sensor observations. The combined use of modeling andmulti-sensor observational data reveals the baroclinic structure of the Balearic Current and the NorthernCurrent and a small-scale anticyclonic eddy observed northeast of the Mallorca coast (current 15 cm/s,180 m deep). This mesoscale structure, partially intercepted by the glider andalong-track altimetric measurements, is marked by relatively strong salinity gradients and not, as is moretypical, temperature gradients. Finally, the use of the validated model simulation also shows that thegeostrophic component of this small-scale eddy is controlled by sub-surface salinity gradients. Wehypothesize that this structure contains recently modied Atlantic water arriving from the strait of Ibizadue to a northerly wind, which strengthens the northward geostrophic circulation.

    2012 Elsevier Ltd. All rights reserved.a r t i c l e i n f o a b s t r a c tand modeling

    J. Bouffard a,, L. Renault a,b, S. Ruiz a, A. Pascual a, C.a IMEDEA (CSIC-UIB), TMOOS Dept., Mallorca, Spainb SOCIB, Coastal Ocean Observing and Forecasting System, SpaincCLS Space Oceanography Division, Ramonville, Francell rights reserved.m multi-sensor observations

    fau c, J. Tintor a,b

    SciVerse ScienceDirect

    eanography

    evier .com/locate /pocean

  • features on coastal and regional scales (Dussurget et al., 2011). Seasurface properties related to geostrophic and ageostrophic(sub)mesoscale motions are clearly observed on satellite sea sur-face temperature and ocean color images (Lehahn et al., 2007)

    tered the basin through the channels to the south (La Violette

    to a more robust error budget assessment, using in situ and re-

    J. Bouffard et al. / Progress in Oceabut such surface signatures do not convey much quantitative infor-mation on associated currents and sub-surface structures. In addi-tion, the relatively sparse distribution of conventional in situmeasurements at depth (such as ADCP or Argo oats) and at thesurface (such as drifters) only enable a limited quantication of(sub)mesoscale processes and the associated mechanisms alongthe water column.

    In this context, complementary high-resolution monitoringtechnologies (e.g., autonomous underwater vehicles or gliders)are also being implemented in ocean observatories, such as IMOSin Australia,1 OOI in the USA,2 and SOCIB3 and MOOSE4 in Europe.By collecting high-resolution observations of temperature, salinityand biogeochemical variables both at the surface and through thewater column, gliders have led to major advances in the understand-ing of key scientic questions related to (sub)mesoscale physical andbiogeochemical processes (e.g., Sackmann et al., 2008; Nie-wiadomska et al., 2008; Perry et al., 2008; Martin et al., 2009a;2009b; Hodges and Fratantoni, 2009; Testor et al., 2010).

    Moreover, numerous studies have indicated that the combina-tion of gliders and altimetry yield additional benets when moni-toring transports (Gourdeau et al., 2008) and characterizingmesoscale structures in the upper ocean (Htn et al., 2007; Martinet al., 2009a, 2009b; Ruiz et al., 2009a, 2009b; Pascual et al., 2010).Bouffard et al. (2010) recently developed innovative strategiescombining improved coastal along-track altimetry (rather thanstandard altimetric maps) and glider data to quantify horizontalows more precisely, specically in terms of the current velocityassociated with laments, eddies or shelf-slope ow modicationsin the Balearic Sea. The use of these two datasets improves the sep-aration of small-scale dynamics from noise. This increased clarityhas revealed the presence of relatively intense eddies, as previ-ously observed with satellite infra-red images and in situ data, con-rming that (sub)mesoscale variability is a dominant factoraffecting the local circulation and water exchanges between adja-cent Balearic sub-basins (Pinot et al., 1995). As a step forward, thispaper proposes scientically exploiting such datasets in support ofregional modeling with the main objectives of improving the char-acterization of small mesoscale features and investigating the for-mation process and associated forcing. The study area is theBalearic Basin (Fig. 1), located in the western Mediterranean,where the circulation is rather complex due to the presence ofmultiple interacting scales, including basin scale, sub-basin scaleand mesoscale structures.

    This article is organized as follows: First, we briey present thestudy area characteristics. Second, we describe the experimentalcoastal altimetric data, the glider data and the model conguration.Third, we proceed to a description of the dynamic patterns ob-served in August 2008 from multi-sensor data both at the BalearicBasin scale (with remote-sensed sea surface temperature and alti-metric current maps) and at the coastal scale, north of Mallorca(with glider and along-track altimetry observations). The resultsobtained from the multi-sensor dataset are then compared to arealistic numerical simulation both at the surface and along thewater column. Finally, the validated simulation is exploited toidentify the potential mechanisms associated with the small-scale

    1 http://www.imos.org.au.2 http://www.oceanleadership.org/programs-and-partnerships/ocean-observing/

    ooi/.

    3 http://www.socib.es/.4 http://www.insu.cnrs.fr/environnement/atmosphere/moose-mediterranean-

    ocean-observing-system-on-environment (in French).,mote-sensing observations as well as modeling has the advantageof providing complementary information in terms of resolutionet al., 1990). Both BC and NC have widths on the order of 50 kmand are in good geostrophic balance, as winds only seem to pro-duce transient perturbations in terms of near-inertial oscillations(Font, 1990). In addition to the general basin-scale circulation,the Balearic sub-basin is characterized by frontal dynamics nearthe slope areas and developing between the BC and the NC, suchas mesoscale eddies (Tintor et al., 1990; Pinot et al., 2002;1994; Rubio et al., 2009), laments and shelf-slope ow modica-tions (Wang et al., 1988; La Violette et al., 1990). These dynamicshave been found to modify not only the local dynamics, with asso-ciated signicant vertical motions (Pascual et al., 2004), but alsothe large-scale patterns, as shown by Pascual et al. (2002) in a de-tailed study of the blocking effect of a large anti-cyclonic eddy. Thesubmarine topography associated with these complex interactionsbetween the surface and sub-surface waters plays a key role incontrolling the transport between the northern and southern re-gions (Astraldi et al., 1999) and may also enhance the (sub)meso-scale activity in the Balearic Sea (Alvarez et al., 1996).

    Despite several previous studies, the characterization of(sub)mesoscale dynamics in the Balearic Sea is difcult given thewide spectrum of temporal and spatial variability of the processeswith which they interact. Moreover, the signal-to-noise ratio inobservations is generally low because the eddy kinetic energy overthis area (Pascual et al., 2007) is an order of magnitude weakerthan that observed by altimetry in the global ocean (Pascualet al., 2006).

    3. Materials and methods

    Due to the scales of wavelengths and magnitudes involved (fewcentimeters over few tens of kilometers in terms of sea surfaceslope); it is difcult to differentiate small-scale dynamic featuresfrom noise in sea surface observations, specically over the coastaldomain of the north western Mediterranean Sea (Bouffard et al.,2008, 2011). Cross-comparisons between model and multi-sourcedatasets should therefore increase condence in our dynamicalstructure characterization. Within this framework, a glider missionwas conducted, well co-located and almost simultaneouslylaunched with a JASON 2 satellite pass (see Fig. 1, right), with themain objective being characterizing the 3D horizontal currentsassociated with the oceanic small mesoscale features. In additionstructure simulated north of Mallorca and previously observedalong the altimetric and glider transects.

    2. Study area

    The general surface circulation of the Balearic Sea is mainly con-trolled by the presence of two fronts and their associated currents(Font et al., 1988; Font, 1990). The Catalan front is a shelf/slopefront that separates old Atlantic Water (AW) in the center of theBalearic sub-basin from the less dense water transported by theNorthern Current (NC), which is also old AW but is fed into the Gulfof Lion and the Catalan shelves by continental fresh water (refer toFig. 1). The NC is a density coastal current owing southwestwardfrom the Ligurian Sea to the Balearic Sea. There, it passes the IbizaChannel or retroects cyclonically over the insular slope formingthe Balearic Current (BC). The Balearic front is also a slope front re-lated to the presence of more recently modied AW that has en-

    nography 106 (2012) 6279 63(glider, along-track altimetry) and coverage (gridded altimetry,model). The following sections provide a detailed description ofthe dataset and the model used.

  • ma70r Ba.)

    64 J. Bouffard et al. / Progress in Oceanography 106 (2012) 62793.1. Satellite altimetry

    Radar altimetry is a key tool for observing the complexity of theopen ocean circulation (Le Traon and Morrow, 2001). However,coastal and regional dynamics in the Balearic Sea, where horizontalspatial scales can be on the order of 10 km (Send et al., 1999), aremuch more complex to observe with altimetry. Until recently, itwas difcult to capture the dynamics associated with coastalsmall-scale processes with along-track data because the signal-to-noise ratio in the coastal band is rapidly degraded as the altim-eter and radiometer signals are perturbed at distances of 10 and50 km from the coast, respectively. In addition to land contamina-tion, the data quality in these regions suffered from a lack of coast-al zone algorithms (Anzenhofer et al., 1999; Vignudelli et al., 2005).

    In this respect, new altimetric post-processing methods andquality control procedures have been developed and can now beexploited for regional and coastal applications (Emery et al.,

    Fig. 1. (a) The bathymetry of the western Mediterranean Sea with ROMS and WRF donorthward transect (the southward transect is equivalent) and J2 altimetric trackmesoscale events (dashed white arrow). NC stands for Northern Current, BC stands foto color in this gure legend, the reader is referred to the web version of this article2011, Kouraev et al., 2011; Ginzburg et al., 2011; Lebedev et al.,2011; Bouffard et al., 2008, 2010, 2011; Roblou et al., 2011; Birolet al., 2010). These studies indicate that new strategies, such asthe use of high-frequency, along-track sampling (20 Hz data), com-bined with a coastal-oriented editing strategy, better constrain thesurface geostrophic current computation in the coastal domain andfor dynamical structures smaller than 50 km.

    In the present study, the post-processing of the coastal altime-ter Sea Level Anomaly (SLA) is similar to that used in Bouffard et al.(2010) for ENVISAT but is applied to the JASON 2 experimentalalong-track data provided by the PISTACH project (see Table 1 fora summary of the main characteristics). Across-track altimetricgeostrophic current anomalies are then derived from fully cor-rected altimetric SLA. The along-track altimetric gradient is esti-mated by the optimal lter developed by Powell and Leben

    Table 1Main altimetric data characteristics.

    Products/satellite Resolution/spatialltering

    Mean s

    PISTACH/JASON-2 (track 70)(Coastal and Hydrology Altimetry producthandbook, 2010)

    20 Hz 350 m MSS Clwww.aMDT (R

    1 Hz 7 km/15 km

    Gridded/Multi-missions(SSALTO/DUACS User Handbook)

    1/8/42 km MSS Clwww.aMDT (R(2004) with a spatial window of 15 km, which is about the rstRossby radius in the Mediterranean Sea (Send et al., 1999). Theacross-track surface geostrophic current is then calculated by add-ing the interpolated Mean Dynamic Topography (MDT) of Rio et al.(2007). By construction, this current is perpendicular to the satel-lite track that, in this case, intercepts the main components of thedynamics related to the BC and/or NC systems (see Fig. 1). Thistype of measurement will therefore be used to precisely quantifythe surface geostrophic current intensity associated with the ob-served coastal and (sub)mesoscale patterns.

    Additionally, the 2D surface geostrophic current derived fromregional AVISO Maps of Sea Level Anomalies ((M)SLA on a 1/8 1/8 grid, updated delayed-time product, refer to SSALTO/DUACS User Handbook) added to the MDT will be also used toidentify the main surface dynamic structures and provide a quali-tative assessment of the model simulation at the Balearic Basinscale.

    ins (in red and black, respectively). (b) Magnied image of the ROMS domain, glider(yellow) overlapped by the main permanent currents (white arrows) and typicallearic Current and AW stands for Atlantic water. (For interpretation of the references3.2. The coastal glider

    As a complement to altimetric observations, glider sub-surfacemeasurements will show the baroclinic structure of related surfacepatterns. To this end, the glider transects were almost co-locatedwith the JASON 2 (hereafter J2) altimetric track along a northward(period: 13/08/0820/08/08) and southward (period: 21/08/0827/08/08) transect (see Fig. 1 for the glider transect location atnorthward transect; the southward transect is almost identical).

    The coastal glider has provided high-resolution hydrographicdata between depths of 10 and 180 m at a horizontal resolutionof approximately 500 m. The glider data processing includes thethermal lag correction (Garau et al., 2011) for unpumped CTD sen-sors installed on Slocum gliders. Hydrographic proles have been

    ea surface/MDT Quality controlprocedure

    Time sampling/cycles

    s01 (http://viso.oceanobs.com)/io et al., 2007)

    As in Bouffard et al.(2010)

    10 days/cycle 4 andcycle 5

    sO1 (http://viso.oceanobs.com)/io et al., 2007)

    Standard 7 days/time-averaged

  • Oceaaveraged vertically to 1 m bins. Relative geostrophic currents canbe estimated by the thermal wind equation from the glider dy-namic height (DH) obtained from temperature and salinity elds(see Ruiz et al., 2009b) with respect to an arbitrary reference depth,which is here represented by the maximum depth of glider mea-surements (in our case, 180 m).

    The issue of reference level correction has been recently ad-dressed in Bouffard et al. (2010) and Gourdeau et al. (2008) andis applied here by combining the relative geostrophic currents withdepth-averaged currents retrieved from the GPS glider positioning.These depth-averaged currents are obtained using a hydrodynamicmodel and on the basis of the assumption that the main differencebetween glider GPS surfacing points and dead-reckoned positionsis due to horizontal currents. Thus, the difference between the rel-ative depth-average geostrophic currents (from DH) and the abso-lute depth-average currents (from GPS glider positioning) shouldapproximately correspond to the absolute geostrophic current at180 m because horizontal ageostrophic motion, tide currents, andbarotrophic high-frequency contributions can be considered negli-gible (Bouffard et al., 2010). By adding this value to the relativegeostrophic current at each depth level, the absolute geostrophiccurrent, both at the surface and along the entire 180 m water col-umn, can therefore be estimated, as expressed in the followingequations:

    ugz urgz uref 1

    uz urgz uref uagz 2where u corresponds to the total current at depth z (such that uz isthe glider-based depth-average velocity); urg and uag are the relativegeostrophic and ageostrophic currents, respectively; and uref is theunknown current at the reference level depth.

    By vertically averaging and combining Eqs. (1) and (2), we ob-tain the following:

    uref uz urgz uagz 3Because the vertical average horizontal ageostrophic motion is

    considered here to be negligible compared to total horizontal cur-rents, we obtain from Eqs. (1) and (3) the absolute geostrophic cur-rent from the sea surface to the reference level depth:

    ugz urgz uz urgz 4At the surface (z = 0) and setting aside the issue of synopticity,

    this quantity should therefore be directly comparable with theinstantaneous, co-located, absolute geostrophic current derivedfrom altimetric measurements (when an MDT is also used).

    3.3. Model description

    The coastal and (sub)mesoscale processes north of Mallorcashould be partially observed by the glider and along-track altimet-ric measurements, but they are spacetime sub-sampled by thesetwo observing systems. The interpretation of such data does not al-low for a complete identication of the mechanisms and forcingsassociated with the observed patterns. This challenge is the mainreason why a regional oceanic model has also been used to bettercharacterize the water masses dynamical interactions at the Bale-aric Basin scale and their relative positions with respect to the gli-der and altimetric track.

    The oceanic model used is the Regional Ocean Modeling System(ROMS) (Haidvogel et al., 2000; Shchepetkin and McWilliams,2005), a 3D free-surface, sigma coordinate, split-explicit equation

    J. Bouffard et al. / Progress inmodel with Boussinesq and hydrostatic approximations. The read-er is referred to the work of Shchepetkin and McWilliams (2005)for a more complete description of the numerical code. Note thatthe simulation has been implemented over the Balearic Sea anddeveloped within the framework of the Balearic Islands CoastalObserving and Forecasting System (SOCIB, www.socib.eu and Tin-tor et al., 2012) and, therefore, is not specically designed for thisstudy. The model domains extends from 1W to 5E and from 38Nto 44N (see Fig. 1b). The vertical discretization considers 30 sigmalevels, and the horizontal grid is 192 224 points with a resolutionof 1 km, which enables a comprehensive sampling of the rst baro-clinic Rossby radius of deformation throughout the entire area(1015 km, Send et al., 1999). The bottom topography is derivedfrom Smith and Sandwell (1997). The simulation, lasting 2 years,is initialized on 1st May 2007 using data on temperature, salinity,horizontal velocities, and sea surface elevation derived from theMediterranean Forecasting System (MFS, Pinardi et al., 2003). Atthe three lateral open boundaries (north, east and south) an active,implicit, upstream-biased radiation condition connects the modelsolution to the surrounding ocean (Marchesiello et al., 2001). Thedaily MFS elds are used to infer the thermodynamics and the cur-rents at the open-boundary conditions. The connection betweenthe MFS elds and the model did not create articial features;therefore, we can focus on features situated close to theopen-boundary conditions. The regional conguration of theatmospheric model Weather Research and Forecasting (WRF,Skamarock et al., 2008), described in Ruiz et al. (2012), providesROMS with the following atmospheric elds every 1 h: 2 m airtemperature, relative humidity, surface wind vector, net shortwaveand downwelling, long wave uxes, and precipitation. A bulk for-mulate based on Fairall et al. (2003) is used to compute turbulentheat and momentum uxes.

    4. Results

    4.1. Hydrodynamics from observations

    4.1.1. Synoptic view from remote sensingBefore the glider mission, from 01/08/2008 to 12/08/2008

    (Fig. 2a), the AVHRR SST show a warm and relatively homogenoustemperature of approximately 2728 C in the center of the do-main, whereas cooler water, below 26.5 C, is only present to thenorth and south. This conguration is typical, as described in Sec-tion 2. Indeed, a marked temperature front, corresponding to theCatalan front, is positioned to the north of the domain, separatingcolder (by 1.5 C) northern water from the warmer water of theBalearic Sea. This front (indicated by the red dashed line onFig. 2a) is remarkably well co-localized with the main surface abso-lute geostrophic current patterns derived from altimetry sea levelelevation (see vectors on Fig. 2).

    During the glider mission, from 13/08/2008 to 27/08/2008(Fig. 2b), the synoptic situation becomes quite different. The SSTcools throughout the western part of the domain, where a meandecrease of 1.5 C is observed, with reference to the previous15 days. The map also indicates a northward current owingthrough the Ibiza Channel into the Balearic Basin along the temper-ature fronts (red dashed line); one branch of this geostrophic owpropagates north until 41N, whereas another branch generates ameander after owing along the north coast of Ibiza. This meanderexpands at the time of the glider mission and evolves into an anti-cyclonic, mesoscale eddy (hereafter called OBELIX, indicated bythe black arrow in Fig. 2b) with a mean current of 15 cm/s, cen-tered southwest of the glider transect (2E, 40.5N). This mesoscalestructure is located at the interface between the relatively warmwater to the south and the cooler water to the north; it corre-sponds to the northwestern border of the new temperature front

    nography 106 (2012) 6279 65replacing the previously observed Catalan front (Fig. 2a). A branchof the eastern border of this eddy partially feeds the BC, whichows and accelerates northward along the coast of Mallorca until

  • ceanography 106 (2012) 627966 J. Bouffard et al. / Progress in Oreaching the glider transect, at its southern edge (near the Minor-can coast). Farther north (>40.5N along the glider transect), apartfrom the return branch of the NC located at the northern edge ofthe glider transect (also refer to Fig. 1), no clear sea surface signa-ture is observed.

    4.1.2. Surface currents from along-track altimetry and glidermeasurements

    Fig. 3 shows the absolute across-track surface geostrophic cur-rents derived from the glider data, satellite altimetry data (bothgridded and along-track product) and model output (used later,in Section 4.2.2.2) to precisely identify surface signatures associ-ated with dynamical features during the glider mission.

    The northward (southward) glider transects, corresponding tothe 13/08/0820/08/08 (21/08/0827/08/08) period, were com-pared with across-track altimetric measurements of J2 cycle 4:13/08/2008 (J2 cycle 5: 23/08/2008). The two measurements arespatially well co-located (Fig. 1; right), but the time delay betweeninstantaneous altimetry sampling and the glider data may be apotential source of differences and should be considered (not

    Fig. 2. Time-averaged SST (AVHRR) overlapped by the gridded altimetric geostrophic cuthe glider mission. (For interpretation of the references to color in this gure legend, th

    Fig. 3. Comparisons of absolute surface currents (in cm/s) interpolated at the glider locat15-km-smoothed signals). The blue curve corresponds to the glider rebuilt across-track abSection 3.2). The red curve corresponds to the instantaneous altimetric across-track absocycle 4 of JASON2; (b): cycle 5 of JASON2). The gray curve corresponds to the space/timfrom AVISO ((M)SLA + MDT). The large black curve corresponds to the space/time interpothe model across-track total surface current. BC, NC and the associated separations represof the Northern Current, respectively. (For interpretation of the references to color in thspecically discussed here). Nevertheless, the correlations are0.91 and 0.73 for the northward and southward transects, respec-tively (see Table 2). However, even if the spatial variations ofcurrent are well phased, an important bias (8.2 cm/s) is observedon the southward transect. This bias is not found during the north-ward transect, where the mean difference is only 2.3 cm/s (Fig. 3band Table 2). This unrealistic bias is likely due to instrumentalerrors in the glider GPS positioning system and induced glider

    rrent (a) before (01/08/200812/08/2008) and (b) during (13/08/200827/08/2008)e reader is referred to the web version of this article.)

    ion along the (a) northward and (b) southward transects (large curves correspond tosolute surface geostrophic current (using a reference level correction as described inlute surface geostrophic current derived from along-track PISTACH data (+MDT, (a):e interpolated altimetric across-track absolute surface geostrophic current derivedlated model across-track absolute surface geostrophic current and the thin curve toent the supposed mean position of the Balearic Current and of a retroection branchis gure legend, the reader is referred to the web version of this article.)

    Table 2Statistical comparisons between across-track absolute surface geostrophic current(mean in cm/s) from along-track altimetry (PISTACH) and glider observations.

    Mean Altimetry GliderCorrelation Correlation

    AltimetryCycle 4 7.9 1 0.91Cycle 5 6.8 1 0.73

    GliderNorthward 5.6 0.91 1Southward 15.0 0.73 1

  • compass errors (Merckelbach et al., 2008). To correct this error, thespatial average altimetric absolute surface geostrophic currentcould be used as a reference despite the potential inconsistenciesdue to temporal lags between the (instantaneous) altimetric and(non-synoptic) glider measurements.

    As in Bouffard et al. (2010) (for ENVISAT), comparisons alsomade with standard 1 Hz altimetry (not shown here) conrm thatthe PISTACH 20 Hz along-track sampling using the new editingstrategy (described in details in Bouffard et al., 2010) improvesthe altimetryglider statistical consistency: along the northwardtransect, the correlations are 0.90 (for edited 20 Hz data) and 0.78(for standard 1 Hz data), with 55% of STD explained for the edited20 Hz data and 35% for the standard 1 Hz data. The same conclu-sions are obtained when the glider surface absolute currents arecompared with corresponding currents derived from standard AVI-SO (M)SLA. Despite weaker current amplitudes, this product (seethe gray curves in Fig. 3) however provides realistic surface currentswhen qualitatively compared to glider and along-track altimetry.

    The previous comparisons show a relatively good agreementbetween the glider and PISTACH altimetry (see Fig. 3 and Table 2),particularly at the southern edge of the track, where the currentintensity is approximately 15 cm/s. This nding corresponds withthe BC position (see Figs. 1 and 2), whereas at the opposite edge,the two datasets also captured major dynamical features (cur-rents > 10 cm/s), which may correspond to a return branch of theNC (see Figs. 1 and 2). Between the potential positions of the BCand NC, a small-scale oscillation is also revealed (hereafter calledASTERIX), whereas this oscillation was not clearly observed inthe previous SST and gridded altimetry-derived map (Fig. 2b). Toinvestigate the associated baroclinic structure, we now analyze

    4.1.3. Hydrography from glider measurementsFig. 4 shows the potential temperature (a), salinity (b) and den-

    sity (c) proles captured by the glider for the northward transect,co-located at the surface with the altimetric track (the resultsalong the southward transect, not shown here, are equivalent).

    Concerning the potential temperature (Fig. 4a), sub-mesoscaleoscillations, less than 10 km in extent, can be observed in the rst50 m layer, whereas no marked signals appeared below 50 m.Examining the salinity proles (Fig. 4b), small-scale features(marked by relatively low salinity values of 38.038.1 psu) are alsoobserved in the top 50 m of the water column. These structures arevery different from the signals at greater depth (>60 m), wheremarked horizontal salinity gradients appear. These (sub)surfacesalinity gradients are well phased with the surface altimetric andglider currents previously analyzed. Table 3 shows that below50 m, the mean correlations between these salinity gradients andthe across-track surface absolute geostrophic currents are 0.67and 0.81 for altimetry and glider data, respectively (0.13 and0.06, respectively, in the rst 50 m). Because such signicant(anti)correlations are not obtained with temperature gradients,this nding may indicate that local geostrophic currents are drivenmainly by salinity and not temperature gradients.

    To quantitatively conrm or reject this assessment, two virtualdensity elds have been computed in two different ways (Fig. 4dand e) and compared to the density eld derived from the realtemperature (T) and salinity (S) proles (real, Fig. 4c). The rstdensity eld (Fig. 4d) was built using the real potential tempera-ture measurements but by considering, at each depth level, a con-stant salinity value (the horizontal spatial mean from the glider). Incontrast, the second density eld (Fig. 4e) was built using the real

    the ne sa

    J. Bouffard et al. / Progress in Oceanography 106 (2012) 6279 67the hydrographic elds from the glider measurements. This analy-sis should provide insight into the potential forcings associatedwith the observed surface geostrophic ow.

    Fig. 4. (a) The potential temperature, (b) salinity and (c) density glider proles alongof depth (m) and latitude (N). (d) The potential density computed by removing th

    gradients. The white dashed lines in (a)(e) correspond to 60 m depth. (f) Black curve: tcurve: the percentage of the STD explained by (e) in (c) as a function of depth (in m). (Forthe web version of this article.)salinity eld but considering at each depth a constant temperaturevalue (the spatial mean from glider). These approaches imply thatthe rst density eld (S xed, Fig. 4d) does not consider effects

    orthward transect (the result for the southward transect is equivalent) as a functionlinity gradients. (e) The potential density computed by removing the temperature

    he percentage of the STD explained by (d) in (c) as a function of depth (in m). Blueinterpretation of the references to color in this gure legend, the reader is referred to

  • caused by the salinity gradients, whereas the second density eld(T xed, Fig. 4e) does not consider effects related to the temper-ature gradients. By comparing the horizontal gradients of thesetwo virtual density elds to the observed density eld (Fig. 4c), itshould be possible to evaluate the respective contributions ofsalinity and temperature to the density gradient (and thereforeon the across-track geostrophic ow).

    The results are reported in Fig. 4f, which shows that for depthsless than 50 m, temperature gradients dominate because the per-centage of standard deviation (STD) explained by the S xed(T xed) eld in the real density eld is greater than 80%(approximately 0%). Below 50 m, the percentage of STD explainedby the S xed eld progressively decreases, whereas the percent-age of STD explained by the T xed eld increases. Below 60 m,the percentage of STD explained by the T xed eld (>70% below100 m) is greater than that explained by the S xed eld (

  • remote-sensing data over the JulyAugust 2008 period. In thisanalysis, the surface geostrophic current from altimetric griddedabsolute dynamic topography (ADT = (M)SLA + MDT) can be com-pared to the 2D model geostrophic currents derived from the mod-el sea level height (interpolated on the same grid). Thus, the kineticenergy (KE) has been calculated from these two consistent currentelds and time-averaged over the period JulyAugust 2008 (seeFig. 5b and d). To complement this analysis, we also computed,during the same period, the time-averaged SST from both simula-tion and satellite AVHRR data (see Fig. 5a and c). The temporal evo-lution of these two variables, spatially averaged over the entiremodel domain, is also used for quantitative comparison (Fig. 6).The corresponding statistical results are reported in Table 4.

    Fig. 5 illustrates the relatively close general agreement betweenthe model and satellite data. Concerning SST (Fig. 5a and c), theCatalan front is clearly observed in the two SST elds in the north-ern part of the model domain, separating relatively cold water(26 C) located in the Balearic

    curve in Fig. 6b). When the eddy kinetic energy (EKE) is computedby removing the 2-month mean current both in the model andaltimetry (and therefore the impact of MDT in the altimetric cur-

    Table 4Statistical comparisons between model and AVHRR SST (in C), KE and EKE (byremoving the 2-month mean current) (in cm2/s2).

    Area Mean STD Absolute difference

    SSTModel 26.0 0.8 0.5Satellite 26.0 1.0

    KEModel 170 35 52Satellite 221 20

    EKEModel 58 16 18Satellite 41 18

    J. Bouffard et al. / Progress in Oceanography 106 (2012) 6279 69Sea. In the southern part of the domain, colder water patches ob-served using remote sensing are also reproduced by the model.The time series of the spatial-averaged SST (Fig. 6a) also showthe models ability to correctly simulate the SST variability ob-served by satellite measurements in the period JulyAugust2008. The curves show a general SST decrease of 1 C between01/07/08 and 20/07/08, an increase of 2.5 C between 21/07/08and 06/08/08 and a decrease of 1 C until 31/08/08. The statisticalresults conrm this good agreement (see Table 4): the model andsatellite SST are characterized by the same mean temperature(26 C), an absolute mean difference of 0.5 C, a correlation greaterthan 0.9 and similar standard deviations of 0.8 C and 1 C for themodel and the satellite observations, respectively.

    As expected, the agreement between the model and remote-sensing data in terms of KE is not as clear because the model isnot constrained using data assimilation, and, as noted previously,altimetric (M)SLA are not always adapted for regional studies. Rel-atively high mean KE values (>500 cm2/s2) are observed in the twomaps, especially in the southern part of the domain, but they arenot co-located (cf. Fig. 5b and d). The temporal evolution of thespatially averaged KE, however, exhibits very closely aligned ten-dencies with a general decrease of 20% between 01/07/08 and06/08/08 followed by an equivalent increase between 06/08/08and 27/08/08, exactly coincident with the SST decrease previouslyobserved (see Fig. 6a and b). Statistical results show a mean abso-lute difference between the model and altimetry data, which rep-resents 20% of the mean altimetric KE (see Table 4 and the redFig. 6. A time series of spatially averaged (over the model domain) SST and KE derived froto the absolute difference between model and satellite KE. (For interpretation of the refearticle.)results are reported in Appendix A and appear to conrm that themodel is able to satisfactorily reproduce the surface ocean dynam-ics at the Balearic regional scale for the period JulyAugust 2008.

    4.2.1.2. Surface changes during the glider mission. We now focus ontwo distinct periods of August 2008 to describe potential changesat the Balearic Basin scale immediately before (01/08/200812/08/2008) and during (from 13/08/2008 to 27/08/2008) the glidermission. These two periods correspond to a relatively low and highKE, respectively (Fig. 6b), in addition to a signicant change in thetrend of SST (Fig. 6a), which tends to decrease after 06/08/2008(both observed and simulated).

    As previously indicated, a comparison between Figs. 2 and 7conrms the good agreement between the ROMS simulation andremote-sensing data, specically over the two periods in Augustrents) data, the results are quite different (see Table 4; gure notshown). In that case, the mean spatial EKE is 66% and 80% lessfor the model and altimetry data, respectively (refer to Table 4for the associated values), the STD is of the same order of magni-tude, and the mean absolute difference is only 18 cm2/s2.

    These initial assessments show that the model is able to ade-quately reproduce the general surface dynamics for the periodJulyAugust 2008, both for the spatial patterns and the associatedtemporal variability. In addition, we subdivided the model domaininto four zones of equal surface (as shown in Fig. 5) to discriminatethe relative errors and the contributions of each sub-region interms of KE and SST mean and variability. The obtained statisticalmmodel and satellite (AVHRR and AVISO (M)SLA + MDT). The red curve correspondsrences to color in this gure legend, the reader is referred to the web version of this

  • 08/2acte

    70 J. Bouffard et al. / Progress in Oceanography 106 (2012) 62792008. The description of dynamical patterns is similar to that ob-served using remote sensing, described in Section 4.1.1. In particu-lar, the temperature appears cooler over the western part of thedomain during the glider mission. An analysis of the models atmo-spheric forcing (not shown) indicates this cooler temperature islikely due to a strong cold wind event that increased local heat lossby turbulent heat uxes and vertical mixing. The local SST cooling,in turn, generates a zonal temperature gradient that reinforces thenorthward geostrophic circulation and also leads to an increase inthe mesoscale activity (as observed in the KE; see the previous sec-tion and Appendix A).

    A major difference between the model and satellite observa-tions, however, occurs near the Iberian coast and more generallyin zone 3 (see Figs. 2 and 7), where a southward coastal currentnot reproduced by the model is observed. In addition to potentialboundary condition issues, this disagreement may be due to stan-dard altimetry limitations in the coastal zone (refer to the intro-duction) and a lack of the data required to constrain the MDTcomputation (refer to Rio et al., 2007). Close to this area, a perma-nent and clearly unrealistic cyclonic eddy, centered at (1E,40.5N), is also seen (present in the MDT and not in the (M)SLA).This structure, not reproduced by the ROMS simulation before orduring the glider mission, is relatively far from the glider transect

    Fig. 7. The time-averaged SST overlapped by the geostrophic current (a) before (01/mission from the ROMS oceanic model. Boxes used to monitor the water mass charand altimetric track location (>200 km) and should not affect thefollowing interpretations. Farther east, the modeled and altimetricsurface geostrophic circulations are quite similar (as will also be

    Fig. 8. The temporal difference in the surface geostrophic current (vectors) and relative vglider mission from (a) altimetric data from AVISO and (b) ROMS model in the closeinterpretation of the references to color in this gure legend, the reader is referred to thconrmed in Section 4.2.2), which conrms the previous valuesof KE, showing relatively good results (except in zone 3; refer toAppendix A) when the damaging impact of the MDT is notremoved.

    With regard to the absolute geostrophic current, using a mapof differences between two time periods allows us to precisely re-move potential issues related to the MDT by dealing directly withthe oceanic signal variability. Fig. 8 shows that the changes dur-ing the glider mission in the model and gridded altimetry spatialpatterns are quite similar. In particular, both the model and satel-lite observations are marked by a reinforcement of the currentintensity and a negative vorticity associated with an anticycloniceddy acceleration (OBELIX, as noted previously in Section 4.1.1).Fig. 8 also shows that the BC intensity decreases by approxi-mately 5 cm/s, whereas the northward geostrophic current cross-ing the glider transect in the north tends to increase. The middlepart of the glider transect is characterized by low surface vorticityvariability (and KE; see also Fig. 5b and d) both in the model andgridded altimetry. However, weak geostrophic surface currentsignatures may hide more intense sub-surface geostrophic cur-rents (conrmed in the following section). Additionally, even ifcurrents derived from the altimetric map provide a qualitativeassessment of general geostrophic patterns at the Balearic Basin

    00812/08/2008, top) and (b) during (13/08/200827/08/2008, bottom) the gliderristics (in Figs. 1315) have also been added.scale, the synoptic representation of the surface circulation at asub-regional scale is compromised by the limitations arising fromthe space/time smoothing effect from the optimal interpolation

    orticity before (01/08/200812/08/2008) and during (13/08/200827/08/2008) thevicinity of the glider transect (the red line indicates the northward transect) (Fore web version of this article.).

  • required to merge data from multiple altimeters (Dussurget et al.,2011).

    4.2.2. Comparison with glider data4.2.2.1. Hydrography. A comparison of Figs. 9 and 4 shows that theROMS simulation and glider hydrographic proles are in goodagreement despite some differences (including a weak salinity gra-dient in the model) and spatial lags of approximately 10 km (theBC is farther offshore in glider data). Moreover, the glider sectionexhibits wiggles in the rst layer, which may be due to internalwave heaving, that are not reproduced by the model.

    Concerning the potential temperature (Fig. 9a), the modelshows less spatial variability in the rst 50 m layer, where sub-mesoscale oscillations less than 10 km in extent were observedin the glider prole. As a result, in this area, the temperature STDis 30% lower in the model and low correlations between modeland glider are observed (Fig. 10, top). Outside of the BC location(latitude > 40.2N) and below a depth of 50 m, the temperaturehorizontal gradients are very weak (STD < 0.5 C in both the gliderand model), implying that correlations are not highly signicant.Despite this weak correlation, the glider and model data showclose general statistical proles (STD and mean; Fig. 10, top) withdepth-averaged mean temperatures of 15.8 C and 16.1 C andSTDs of 0.31 C and 0.25 C, respectively.

    Concerning salinity, the mean model and glider prole are sim-ilar, with correlations below 50 m of between 0.8 and 0.9 (Fig. 10,middle). However, even if the glider and model salinity proles arequite clearly phased, the sub-surface salinity STD in the glider datais twice that in the model (see Fig. 10). Indeed, as suggested by thesalinity prole (see Figs. 4 and 9b), the sub-surface salinity gradi-ents are stronger in the glider data than in the model. This discrep-

    surface currents with respect to the observed currents (as will benoted in Section 4.2.2.2; see also Fig. 2). As previously observedwith glider measurements, these model salinity gradients are alsowell correlated with the surface absolute geostrophic current (cor-relations > 0.5, see Tables 3 and 5), unlike the temperaturegradient.

    To precisely evaluate the relative contributions of salinity andtemperature gradients to the model density gradient computation,the same experiment as that performed with the glider has beenperformed. The results from the model (Fig. 6f) conrm that thesub-surface geostrophic currents are also mainly driven by salinitygradients. Indeed, as with the glider data, the inuence of modeltemperature gradients decreases for depths below 50 m. Fig. 6falso shows that the impact of salinity gradients exceeds the impactof temperature gradients at depths above 100 m (whereas thethreshold was at approximately 60 m in depth according to the gli-der data). This nding may explain why the correlations betweenthe model and glider density proles become signicantly betterbelow 100 m in depth, where the salinity gradients dominate (cor-relation > 0.8; refer to Fig. 10, bottom).

    Moreover, the water masses identied in Figs. 4 and 9 between40.1N and 40.4N and between 40.6N and 40.8N have the samehydrographic properties in both the model and glider measure-ments (a low salinity of less than 38.1 psu and temperature of14 C at 70 m), which suggests that they have the same originbut have followed two different trajectories. This nding is consis-tent with the 2D surface analysis described in Section 4.1.2 (alsoconrmed in Section 4.2.1.2 with model outputs), in which theOBELIX eddy was shown to partially feed the BC in the south,whereas its northern part interacted with the NC return loop andadvected water northward. The origin of this water mass, charac-

    cetens

    J. Bouffard et al. / Progress in Oceanography 106 (2012) 6279 71ancy may be the source of an underestimation of the modeled

    Fig. 9. (a) The potential temperature, (b) salinity and (c) density model proles spatransect is equivalent) as a function of depth (m) and latitude (N). (d) The potential d

    by removing the temperature gradients. The white dashed lines in (a)(e) correspond tofunction of depth (in m). Blue curve: the percentage of the STD explained by (e) in (c) as alegend, the reader is referred to the web version of this article.)terized by a lower salinity value at the source of local salinity

    ime interpolated along the northward glider transect (the result for the southwardity computed by removing the salinity gradients. (e) The potential density computed

    60 m depth. (f) Black curve: the percentage of the STD explained by (d) in (c) as afunction of depth (in m). (For interpretation of the references to color in this gure

  • 72 J. Bouffard et al. / Progress in Oceanography 106 (2012) 6279gradients (and therefore geostrophic ows), will be discussed inSection 4.3.

    Fig. 10. Statistical comparisons between model and glider hydrographic measurementscorrespond to statistics from the glider (both STD and mean); black curves correspondmodel and glider measurements. (For interpretation of the references to color in this g

    Table 5Mean spatial correlations (function of depth) between across-track surface absolutegeostrophic currents and temperature and salinity gradient (from model).

    Temperature(from model)

    Salinity(from model)

    Correlation with surface absolute geostrophic current (from model)0 to 50 m 50 to 180 m 0 to 50 m 50 to 180 m0.35 0.30 0.64 0.53

    Fig. 11. Comparisons between glider (blue curves, raw and 15-km-smoothed data) and spcurrents as a function of latitude (reference level at 180 m) at the time of the (a) northwa2008). (For interpretation of the references to color in this gure legend, the reader is r4.2.2.2. Currents. The model surface relative geostrophic current

    (potential temperature, salinity and density) as a function of depth (m). Blue curvesto statistics from the model. The red curve corresponds to the correlation betweenure legend, the reader is referred to the web version of this article.)has been computed with respect to the 180 m reference depth(as in glider) and interpolated to the glider track in space and time.Fig. 11 shows comparisons with the glider hydrographic measure-ments along the northward (13/08/0820/08/08, Fig. 11a) andsouthward transects (21/08/0827/08/08, Fig. 11b).

    Despite spatial lags of a few kilometers, Fig. 11 shows that themodeled and glider relative surface geostrophic currents are con-sistent, with a mean difference less than 1 cm/s and correlationsof 0.91 and 0.52, respectively, for the northward and southwardtransects (see Table 6 for statistics). For the northward glider tran-sect (Fig. 5a), the model and glider data show a decrease in the cur-rent intensity from 5 cm/s to 0 cm/s in the cross-shore direction at

    acetime interpolated model (black curve) relative across-track surface geostrophicrd (13/08/200820/08/2008) and (b) southward transects (right, 21/08/200827/08/eferred to the web version of this article.)

  • the BC mean location (40.140.5N). The ndings differ for thesouthward transect (Fig. 5b), with a BC relative surface geostrophiccurrent that is larger but half as intense. In the northern BC, the rel-ative geostrophic current is alternatively negative (southward) and

    return-branch, a sub-surface feature is observed by the gliderand is partially reproduced by the model (see the red square inFig. 12b and d). Although the model structure shows a spatial lagof approximately 10 km toward the south, it shows an alterna-tively positive (northward) and negative (southward) current withmaximum intensity at depths above 50 m for the positive currentvalues in the model and the negative values in glider data (at lati-tude 40.6N).

    The next challenge will be to use the validated model to de-scribe the characteristics and the potential processes associatedwith this structure, intercepted north of Mallorca (previouslycalled ASTERIX in Section 4.1.2).

    4.3. Spatio-temporal variability from numerical modeling

    Table 6Statistical comparisons between across-track relative surface geostrophic current (/180 m) from model and glider observations (mean is in cm/s).

    Mean Correlation with glider

    GliderNorthward 2.2 1Southward 2.7 1

    ModelNorthward 1.6 0.91Southward 1.3 0.52

    J. Bouffard et al. / Progress in Oceanography 106 (2012) 6279 73positive (northward), with a current intensity of 5 cm at thenorthward and southward transects for both the glider and modeldata.

    The comparisons between the absolute surface geostrophic cur-rents from the model and observed data show relatively goodagreement along the northward transect (signicant correlationsof 0.78 and 0.50 with respect to glider and along-track altimetry),which is not the case for the southward transect, marked by a20 km spatial lag (see Fig. 3). Moreover, the amplitude of absolutecurrents (both geostrophic and total) in themodel is much less thanin the observations despite the good agreement previously ob-served in terms of surface relative geostrophic currents. Therefore,this decreased amplitude of absolute currents should also corre-spond to an underestimation of sub-surface currents in the model.

    Based on the analysis in Section 4.1.3, it followed that the sur-face altimetric and glider absolute geostrophic current an inte-grated value reected the salinity gradient signals of depth andnot the surface layer, where small-spatial-scale disturbances con-ceal more stable deep geostrophic signals. Therefore, we investi-gate and compare the patterns in the distribution of the currentsalong the water column from both the model and the glider mea-surements (Fig. 12).

    When the absolute geostrophic current (Fig. 12b and d) and therelative geostrophic current (Fig. 12a and c) are compared, the fea-tures are quite different, both in terms of magnitude and spatialdistribution. In both cases, the BC is located between 40.1N and40.3N, with a maximum intensity at a depth of 50 m, but the abso-lute current intensity is twice larger than the relative geostrophiccurrent (1020 cm/s as compared to 58 cm/s). The BC observedby the gliders was wider (up to 40.4N) and stronger than thecurrent simulated by the model. Between the BC and NCFig. 12. Comparisons between glider and ROMS across-track currents (cm/s) as a functcurrent derived from the DH calculated with respect to 180 m depth from (a) the ROMS musing the methodology described in Section 2.2 from (b) ROMS and (d) glider measuremeand a red square, respectively. (For interpretation of the references to color in this gurion of latitude (N) through the water column (0180 m). The relative geostrophicThree virtual boxes are located at critical sites chosen with re-spect to the previous analyses conducted with the multi-sensordata and oceanic model outputs. The rst box (the red box inFig. 7b) is aimed at monitoring the time variability of hydrologicaland dynamical properties of the southern water entering throughthe Ibiza Strait. The two other boxes are located on each side ofthe glider transect to assess the hydrological and dynamicalchanges both near the ASTERIX site (the black box in Fig. 7b) andaround the NC retroection (the blue box in Fig. 7b).

    Fig. 13 shows the T/S diagrams, as a function of depth and hor-izontally space-averaged, at the three locations during the JulyAugust 2008 period. This gure clearly demonstrates that waterin the south (Box 1) shows hydrographical properties very differentfrom those of the water in the north (Box 2 and Box 3). Indeed, thewater mass here appears to be mainly composed of recently mod-ied AW characterized by a salinity of approximately 0.5 psu lessthan the older, more modied AW of northern origin (see Sec-tion 2). Moreover, the T/S diagram in Box 1 (recent AW) is moredispersed than that in Boxes 2 and 3 (older AW), which suggestsa progressive change of water mass properties during the 2-monthstudy period.

    To quantitatively assess the temporal evolution of the watermasses characteristics in the three virtual boxes and their poten-tial interactions, we now use a Hovmller diagram of horizontalspace-averaged key variables salinity, current magnitude and rel-ative vorticity as a function of time and depth (see Fig. 14).

    Fig. 14 shows that the salinity anomaly (removing the 2-monthmean value) in the Ibiza Strait becomes signicantly negative frommid-July to mid-August. This strong decrease in the salinity anom-aly coincides with an increase in the total current at depth (5 cm/sat 50 m depth). This change corresponds to an entrance of less saltyodel and from (c) glider measurements. The rebuilt absolute geostrophic current bynts. The BC and a sub-surface mesoscale structure are highlighted by a black squaree legend, the reader is referred to the web version of this article.)

  • cea74 J. Bouffard et al. / Progress in OAW, which is progressively modied and advected northward bythe general circulation until it reaches the OBELIX location in Au-gust 2008. At the northwest side of the glider transect (Box 2),the conditions in July show a strong current magnitude (+15 cm/s) and alternating relative vorticity (2.105 s1). This trend corre-sponded to the cyclonic NC position, which moved southward ornorthward as a function of its interaction with southern currents(as also observed in Fig. 16). In August 2008, a positive salinityanomaly of 0.050.1 psu is observed, which is associated with a de-crease of current intensity. This nding corresponds with theperiod during which OBELIX progressively expands to the north

    Fig. 13. A T/S diagram from the model (in C and psu) as a function of depth (m) from Jthree boxes (refer to Fig. 15 for the location of the boxes).

    Fig. 14. (a) The modeled spatially averaged salinity anomaly, (b) total absolute current nothe Ibiza Channel neighborhood (Box 1); middle: the NC retroection neighborhood (Boxto the glider mission period.nography 106 (2012) 6279and interacts with the NC return loop. Consequently, the less saltyAW entering through the Ibiza Strait is advected northward until itreaches salty water from the NC system at the time of the glidermission (see dashed line on Fig. 14). Indeed, Fig. 14 (Box 3) showsa negative salinity anomaly of 0.1 psu in August 2008 in conjunc-tion with a relatively strong, deep current (approximately 5 cm/sat the surface and 1015 cm/s from 30 m to 200 m). At the sametime, a negative sub-surface relative vorticity is generated at depthin the neighborhood of the glider transect, with maximum valuesof 3 105 s1. This relatively strong negative vorticity istime-correlated (>0.9, see Fig. 15) with the sub-surface salinity

    uly 2008 to August 2008, spatially averaged (but not time averaged) for each of the

    rm and (c) relative vorticity as a function of time and depth in the three boxes. Top:2); bottom: the glider east-side neighborhood (Box 3). The dashed lines correspond

  • difference observed between Box 2 and Box 3 (0.2 psu over a fewkm). We will now try to identify the 2D horizontal structure asso-ciated with this sub-surface negative vorticity signature, generatedclose to Box 3, in the neighborhood of the glider transect.

    Fig. 16 shows the relative vorticity temporal evolution (fromJuly 2008 to August 2008) to characterize the 2D horizontal shear

    hypothesis is that the same type of structure has been interceptedby the glider and along-track altimetry measurements. Given itssmall-scale extension (

  • worldwide challenge of coastal (sub)mesoscale dynamic character-ization must be addressed locally through such an integrated ap-proach, combining both observations and free numerical runs.This is especially true in areas where small mesoscale structuresare characterized by a low signal-to-noise ratio. The surface signa-tures are particularly weak in the northwestern Mediterranean andare therefore difcult to interpret with the use of isolated measure-ments alone.

    Although the obtained results are quite encouraging, our com-parisons have also shown the limitations of these two observationsystems in terms of both coverage and accuracy. Focused on theglider platform, the observed bias at the time of back transectcould be further reduced by applying additional corrections tothe glider depth-averaged GPS currents. In particular, we wouldexpect to achieve better results with a more accurate estimationof the error in the glider displacement assumptions, which wouldentail a more accurate measurement of glider heading. With regardto the gridded altimetric product and the MDT used, we have alsoconrmed the relatively poor resolution of existing products,which employ both a more comprehensive satellite constellationcoverage and a regional approach to better constrain the represen-tativeness of local mesoscale features (refer to Dussurget et al.,2011). Furthermore, experiments based on glider versus along-track altimetry cross comparisons need to be repeated to improveand consistently assess the impact of new methods dedicated tocoastal zone applications. The promising early results in coastalaltimetry (refer to Vignudelli et al., 2011, for a review) support

    the need for continued research and scientic applications withthe opportunity to providing inputs and recommendations to fu-ture satellite missions (refer to the SWOT satellite and last CoastalAltimetry Workshop, San Diego 2011).

    In this study, in addition to multi-sensor cross comparisons,particular attention has been paid to the model validation andthe characterization of a small-scale (

  • J. Bouffard et al. / Progress in Oceanography 106 (2012) 6279 77Acknowledgements

    We gratefully acknowledge the contribution of M. Martnez-Ledesma and B. Casas during the data acquisition phase of the gli-der missions. Special thanks are also due to B. Buongiorno Nardelli,R. Escudier, F. Nencioli, A. Petrenko and A. Doglioli for fruitful dis-cussions concerning the interpretations of data and simulations.The authors are also grateful to B. Garau for processing the gliderdata, E. Heslop and R. Campbell for highly valuable editing com-ments and G. Vizoso for the ROMS numerical setup and data man-agement. The altimeter (M)SLA was produced by SSALTO/DUACSand distributed by AVISO with support from the Centre National

    Fig. A2. A time series of the spatially averaged (over the four subdivided domains; refsatellite altimetry (AVISO (M)SLA + MDT). The red curve corresponds to the absolute diffein this gure legend, the reader is referred to the web version of this article.)

    Table A1Statistical comparisons between model and AVHRR SST (in C, refer to Fig. 5 for theZone locations).

    Area Mean STD Absolute difference

    ZONE 1 Model 26.1 0.8 0.4Satellite 26.3 1.1

    ZONE 2 Model 26.0 0.8 0.4Satellite 25.9 1.0

    ZONE 3 Model 25.8 0.8 0.3Satellite 25.9 1.0

    ZONE 4 Model 26.0 0.8 0.4Satellite 26.0 1.0dEtude Spatiale (CNES, France). The coastal high-resolution altim-eter products were produced by CLS and distributed by AVISOwithin the PISTACH project supported by CNES. Sea surface tem-perature images were acquired and processed by EUMETSAT. Thiswork was partially conducted during a CNES Postdoctoral Fellow-ship and supported by PERSEUS FP7-OCEAN-2011 (GA 287600)and the European Commission MyOcean Project (SPA.2007.1.1.01 development of upgrade capabilities for existing GMES fast-trackservices and related operational services; Grant Agreement218812-1-FP7-SPACE 2007-1).

    er to Fig. 5) KE associated with the geostrophic currents derived from model andrence between model and satellite KE. (For interpretation of the references to color

    Table A2Statistical comparisons between model and AVISO KE (in cm2/s2, ref to Fig. 5 for theZone locations).

    Area Mean STD Absolute difference

    ZONE1 Model 170 20 21Satellite 169 30

    ZONE 2 Model 230 65 53Satellite 270 44

    ZONE 3 Model 105 52 135Satellite 240 25

    ZONE 3 (EKE) Model 53 27 18Satellite 41 20

    ZONE 4 Model 182 41 40Satellite 178 40

  • Garau, B., Ruiz, S., Zhang, W.G., Pascual, A., Heslop, E., Kerfoot, J., Tintor, Joaqun.,2011. Thermal lag correction on Slocum CTD Glider Data. Journal of

    ceaAtmospheric and Oceanic Technology 28, 10651071, http://dx.doi.org/10.1175/JTECH-D-10-05030.1.

    Garreau, P., Garnier, V., Schaeffer, A., 2011. Eddy resolving modelling of the Gulf ofLions and Catalan Sea. Ocean Dynamics 61 (7), 9911003.

    Ginzburg, A.I., Kostianoy, A.G., Sheremet, N.A., Lebedev, S.A., 2011. SatelliteAppendix A

    See Figs. A1 and A2 and Tables A1 and A2.

    References

    Alvarez, A., Tintor, J., Sabats, A., 1996. Flow modication and shelfslopeexchange induced by a submarine canyon off the northeast Spanish coast.Journal of Geophysical Reseach 101, 1204312055.

    Anzenhofer, M., Shum, C.K., Rentsh, M., 1999. Coastal altimetry and applications.Tech. Rep. No. 464, Geodetic Science and Surveying, The Ohio State UniversityColumbus, USA.

    Astraldi, M., Balopoulos, S., Candela, J., Font, J., Gacic, M., Gasparini, G.P., Manca, B.,Theocharis, A., Tintore, J., 1999. The role of straits and channels inunderstanding the characteristics of Mediterranean circulation. Progress inOceanography 44, 65108.

    Birol, F., Cancet, M., Estournel, C., 2010. Aspects of the seasonal variability of theNorthern Current (NW Mediterranean Sea) observed by altimetry. Journal ofMarine Systems 81, 297311. http://dx.doi.org/10.1016/j.jmarsys.2010.01.005.

    Bouffard, J., Vignudelli, S., Cipollini, P., Menard, Y., 2008. Exploiting the potential ofan improved multimission altimetric data set over the coastal ocean.Geophysical Research Letters 35, L10601. http://dx.doi.org/10.1029/2008GL033488.

    Bouffard, J., Pascual, A., Ruiz, S., Faugre, Y., Tintor, J., 2010. Coastal and mesoscaledynamics characterization using altimetry and gliders: a case study in theBalearic Sea. Journal of Geophysical Reseach 115, C10029. http://dx.doi.org/10.1029/2009JC006087.

    Bouffard, J., Roblou, L., Birol, F., Pascual, A., Fenoglio-Marc, L., Cancet, M., Morrow, R.,Mnard, Y., 2011. Introduction and assessment of improved coastal altimetrystrategies: case study over the North Western Mediterranean Sea. In:Vignudelli, S., Kostianoy, A.G., Cipollini, P., Benveniste, J. (Eds.), CoastalAltimetry. Springer-Verlag, Berlin Heidelberg, http://dx.doi.org/10.1007/978-3-642-12796-0_12.

    Capet, X., Klein, P., Hua, B.L., Lapeyre, G., McWilliams, J.C., 2008a. Surface kinetic andpotential energy transfer in SQG dynamics. Journal of Fluid Mechanics 604,165174.

    Capet, X., McWilliams, J.C., Molemaker, M.J., Shchepetkin, A.F., 2008b. Mesoscale tosub mesoscale transition in the California current system. Part 2. Dynamicalprocesses and observational tests. Journal of Physical Oceanography 38, 4464.

    Capet, X., McWilliams, J.C., Molemaker, M.J., Shchepetkin, A.F., 2008c. Mesoscale tosubmesoscale transition in the California current system. Part I: Flow structure,eddy ux, and observational tests. Journal of Physical Oceanography 38, 2943.

    Chelton, D.B., Schlax, M.G., Samelson, R.M., 2011. Global observations of nonlinearmesoscale eddies. Progress in Oceanography 91, 167216.

    Coastal and Hydrology Altimetry Product (PISTACH) Handbook, SALP-MU-P-OP-16031-CN 01/00, edition 1.0, October 2010.

    Ducet, N., Le Traon, P.Y., Reverdin, G., 2000. Global high-resolution mapping ofocean circulation from TOPEX/Poseidon and ERS-1 and-2. Journal ofGeophysical Research Oceans 105, 1947719498.

    Dussurget, R., Birol, F., Morrow, R.A., De Mey, P., 2011. Fine resolution altimetry datafor a regional application in the Bay of Biscay. Marine Geodesy 2 (34), 130.

    Emery, W.J., Strub, T., Leben, R., Foreman, M., McWilliams, J.C., Han, G., Ladd, C.,Ueno, H., 2011. Satellite altimetry applications off the coasts of North America.In: Vignudelli, S., Kostianoy, A.G., Cipollini, P., Benveniste, J. (Eds.), CoastalAltimetry. Springer-Verlag, Berlin Heidelberg, http://dx.doi.org/10.1007/978-3-642-12796-0_16.

    Fairall, C.F., Bradley, E.F., Hare, J.E., Grachev, A.A., Edson, J.B., 2003. Bulkparameterization of airsea uxes: updates and verication for the COAREalgorithm. Journal of Climate 16, 571591.

    Font, J., 1990. A comparison of seasonal winds with currents on the continentalslope of the Catalan Sea (Northwestern Mediterranean). Journal of GeophysicalReseach 95 (C2), 15371546.

    Font, J., Salat, J., Tintor, J., 1988. Permanent features of the circulation in theCatalan Sea. Oceanologica Acta, 5157.

    Fu, L.L., Chelton, D.B., Le Traon, Y., Morrow, R., 2010a. Eddy dynamics from Satellitealtimetry. Oceanography 23 (4) (A Quarterly Journal of the OceanographySociety).

    Fu, L.-L., Alsdorf, D., Rodriguez, E., Morrow, R., Mognard, N., Lambin, J., Vaze, P.,Lafon, T., 2010b. The SWOT (Surface Water and Ocean Topography) Mission:Spaceborne Radar Interferometry for Oceanographic and HydrologicalApplications. In: Hall, J., Harrison, D.E., Stammer, D. (Eds.), Proceedings of theOceanObs09: Sustained Ocean Observations and Information for SocietyConference, vol. 2, Venice, Italy, 2125 September 2009, ESA Publication WPP-306.

    78 J. Bouffard et al. / Progress in Oaltimetry applications in the Black Sea. In: Vignudelli, S., Kostianoy, A.G.,Cipollini, P., Benveniste, J. (Eds.), Coastal Altimetry. Springer-Verlag, BerlinHeidelberg, http://dx.doi.org/10.1007/978-3-642-12796-0_14.Gourdeau, L., Kessler, W.S., Davis, R.E., Sherman, J., Maes, C., Kestenare, E., 2008.Zonal jets entering the Coral Sea. Journal of Physical Oceanography 38, 715725. http://dx.doi.org/10.1175/2007JPO3780.1.

    Haidvogel, D.B., Arango, H., Hestrom, K., Beckmann, A., Malanotte-Rizzoli, P.,Shchepetkin, A., 2000. Model evaluation experiments in the North Atlanticbasin: simulations in nonlinear terrain-following coordinates. Dynamics ofAtmospheres and Oceans 32, 239381.

    Htn, H., Eriksen, C.C., Rhines, P.B., 2007. Buoyant eddies entering the Labrador Seaobserved with gliders and altimetry. Journal of Physical Oceanography 37,28382854.

    Hodges, B.A., Fratantoni, D.M., 2009. A thin layer of phytoplankton observed in thePhilippine Sea with a synthetic moored array of autonomous gliders. Journal ofGeophysical Reseach 114, C10020. http://dx.doi.org/10.1029/2009JC005317.

    Hu, Z.H., Doglioli, A., Petrenko, A., Marsaleix, P., Dekeyser, I., 2009. Numericalsimulation of mesoscale eddies in the Gulf of Lion. Ocean Modelling 28 (4), 203208. http://dx.doi.org/10.1016/j.ocemod.2009.02.004.

    Hu, Z.Y., Petrenko, A.A., Doglioli, A.M., Dekeyser, I., 2011. Study of coastal eddies:application in the Gulf of Lion. Journal of Marine System 88 (1), 311.

    Klein, P., Lapeyre, G., 2009. The oceanic vertical pump induced by mesoscale andsubmesoscale turbulence. Annual Review of Marine Science 1, 351375.

    Kouraev, A.V., Crtaux, J.-F., Lebedev, S.A., Kostianoy, A.G., Ginzburg, A.I., Sheremet,N.A., Mamedov, R., Zakharova, E.A., Roblou, L., Lyard, F., Calmant, S., Berg-Nguyen, M., 2011. Satellite altimetry applications in the Caspian Sea. In:Vignudelli, S., Kostianoy, A.G., Cipollini, P., Benveniste, J. (Eds.), CoastalAltimetry. Springer-Verlag, Berlin Heidelberg, http://dx.doi.org/10.1007/978-3-642-12796-0_15.

    La Violette, P.E., Tintor, J., Font, J., 1990. The surface circulation of the Balearic Sea.Journal of Geophysical Reseach 95, 15591568.

    Le Traon, P.-Y., Morrow, R., 2001. Ocean currents and mesoscale eddies. In: Fu, L.-L.,Cazenave, A. (Eds.), Satellite Altimetry and Earth Sciences: A Handbook ofTechniques and Applications. Academic, New York, pp. 171215.

    Lebedev, S.A., Kostianoy, A.G., Ginzburg, A.I., Medvedev, D.P., Sheremet, N.A.,Shauro, S.N., 2011. Satellite altimetry applications in the Barents and WhiteSeas. In: Vignudelli, S., Kostianoy, A.G., Cipollini, P., Benveniste, J. (Eds.), CoastalAltimetry. Springer-Verlag, Berlin Heidelberg. http://dx.doi.org/10.1007/978-3-642-12796-0_15.

    Lehahn, Y., dOvidio, F., Levy, M., Heifetz, E., 2007. Stirring of the Northeast Atlanticspring bloom: a Lagrangian analysis based on multisatellite data. Journal ofGeophysical Reseach 112, C08005.

    Lvy, M., 2008. The modulation of biological production by oceanic mesoscaleturbulence. In: Weiss, J.B., Provenzale, A., Transport in Geophysical Flow: TenYears After, vol. 744, Lect. Notes Phys. Springer, Berlin, pp. 21961.

    Marchesiello, P., McWilliams, J., Shchepetkin, A., 2001. Open boundary conditionsfor longterm integration of regional oceanic models. Ocean Modelling 3 (1), 20.

    Martin, J.P., Lee, C.M., Eriksen, C., Ladd, C., Kachel, N.B., 2009a. Glider observations ofkinematics in a Gulf of Alaska eddy. Journal of Geophysical Reseach 114,C12021. http://dx.doi.org/10.1029/2008JC005231.

    Martin, J., Lee, C., Eriksen, C., Ladd, C., Kachel, N., 2009b. A two-month seaglidersurvey of a Gulf of Alaska eddy: how dynamical structure evolves temporally.Ser. EOS Transactions AGU, 52(87) (Fall Meeting 2006, abstract OS34A-02).

    McGillicuddy Jr., D.J., Robinson, A.R., Siegel, D.A., Jannasch, H.W., Johnson, R., Dickey,T.D., McNeil, J., Michaels, A.F., Knap, A.H., 1998. Inuence of mesoscale eddieson new production in the Sargasso Sea. Nature 394, 263265.

    Merckelbach, L.M., Briggs, R.D., Smeed, D.A., Grifths, G., 2008. Current measure-ments from autonomous underwater gliders. In: 9th Working Conferenceon Current Measurement Technology. IEEE, Charleston, SC, pp. 6167.

    Morrow, R., Le Traon, P.-Y., 2011. Recent advances in observing mesoscale oceandynamics with satellite altimetry. Journal of Advances in Space Research. http://dx.doi.org/10.1016/j.asr.2011.09.03.

    Nencioli, F., dOvidio, F., Doglioli, A.M., Petrenko, A.A., 2011. Surface coastalcirculation patterns by in-situ detection of Lagrangian coherent structures.Geophysical Research Letters 38, L17604. http://dx.doi.org/10.1029/2011GL048815.

    Niewiadomska, K., Claustre, H., Prieur, L., DOrtenzio, F., 2008. Submesoscalephysicalbiogeochemical coupling across the Ligurian current (northwesternMediterranean) using a bio-optical glider. Limnology and Oceanography 53,22102225.

    Pascual, A., Buongiorno Nardelli, B., Larnicol, G., Emelianov, M., Gomis, D., 2002. Acase of an intense anticyclonic eddy in the Balearic Sea (westernMediterranean). Journal of Geophysical Reseach 107 (C11), 3183. http://dx.doi.org/10.1029/2001JC000913.

    Pascual, A., Gomis, D., Haney, R.L., Ruiz, S., 2004. A quasigeostrophic analysis of ameander in the Palams canyon: vertical velocity, geopotential tendency and arelocation technique. Journal of Physical Oceanography 34, 22742287.

    Pascual, A., Faugre, Y., Larnicol, G., Le Traon, P.-Y., 2006. Improved description ofthe ocean mesoscale variability by combining four satellite altimeters.Geophysical Research Letters 33, L02611. http://dx.doi.org/10.1029/2005GL024633.

    Pascual, A., Pujol, M.I., Larnicol, G., Le Traon, P.Y., Rio, M.H., 2007. Mesoscalemapping capabilities of multisatellite altimeter missions: rst results with realdata in the Mediterranean Sea. Journal of Marine Systems 65 (14), 190211.

    Pascual, A., Ruiz, S., Tintor, J., 2010. Combining new and conventional sensors tostudy the Balearic Current. Sea Technology 51 (7), 3236.

    nography 106 (2012) 6279Perry, M.J., Sackmann, B.S., Eriksen, C.C., Lee, C.M., 2008. Seaglider observations ofblooms and subsurface chlorophyll maxima off the Washington coast.Limnology and Oceanography 53, 21692179.

  • Pinardi, N., Allen, I., De Mey, P., Korres, G., Lascaratos, A., Le Traon, P.Y., Maillard, C.,Manzella, G., Tziavos, C., 2003. The Mediterranean Ocean Forecasting System:rst phase of implementation (19982001). Annales Geophysicae 21 (1), 320.

    Pinot, J.M., Tintor, J., Gomis, D., 1994. Quasi-synoptic mesoscale variability in theBalearic Sea. Deep Sea Research 41 (56), 897914, http://dx.doi.org/10.1016/0967-0637(94)90082-5.

    Pinot, J.M., Tintore, J., Gomis, D., 1995. Multivariate analysis of the surfacecirculation in the Balearic Sea. Progress in Oceanography 36, 343376.

    Pinot, J.M., Lpez-Jurado, J.L., Riera, M., 2002. The CANALES experiment (199698).Interannual, seasonal, and mesoscale variability of the circulation in theBalearic Channels. Progress in Oceanography 55, 335370.

    Powell, B.S., Leben, R.R., 2004. An optimal lter for geostrophic mesoscale currentsfrom along-track satellite altimetry. Journal of Atmospheric and OceanicTechnology 21, 16331642.

    Rio, M.H., Poulain, P.M., Pascual, A., Mauri, E., Larnicol, G., Santoleri, R., 2007. A meandynamic topography of the Mediterranean Sea computed from altimetric data,in-situ measurements and a general circulation model. Journal of MarineSystems 65, 484508. http://dx.doi.org/10.1016/j.jmarsys.2005.02.006.

    Roblou, L., Lamouroux, J., Bouffard, J., Le Henaff, M., Lombard, A., Marsaleix, P., DeMey, P., 2011. Post-processing altimeter data toward coastal applications andintegration into coastal models. In: Vignudelli, S., Kostianoy, A.G., Cipollini, P.,Benveniste, J. (Eds.), Coastal Altimetry. Springer-Verlag, Berlin Heidelberg,http://dx.doi.org/10.1007/978-3-642-12796-0_9.

    Rubio, A., Barnier, B., Jord, G., Espino, M., Marsaleix, P., 2009. Origin and dynamicsof mesoscale eddies in the Catalan Sea (NW Mediterranean): insight from anumerical model study. Journal of Geophysical Reseach 114, C06009. http://dx.doi.org/10.1029/2007JC004245.

    Ruiz, S., Pascual, A., Garau, B., Pujol, I., Tintore, J., 2009a. Vertical motion in the upperocean from glider and altimetry data. Geophysical Research Letters 36, L14607.http://dx.doi.org/10.1029/2009GL038569.

    Ruiz, S., Pascual, A., Garau, B., Faugere, Y., Alvarez, A., Tintor, J., 2009b. Mesoscaledynamics of the Balearic Front, integrating glider, ship and satellite data.Journal of Marine Systems 78, S3S16, http://dx.doi.org/10.1016/j.jmarsys.2009.01.007.

    Ruiz, S., Renault, L., Garau, B., Tintor, J., 2012. Underwater glider observations andmodeling of an abrupt mixing event in the upper ocean. Geophysical ResearchLetters 39, L01603. http://dx.doi.org/10.1029/2011GL050078.

    Sackmann, B.S., Perry, M.J., Ericksen, C.C., 2008. Seaglider observations of variabilityin daytime uorescence quenching of chlorophyll-a in Northeastern Paciccoastal waters. Biogiosciences Discuss 5, 28392865.

    Send, U., Font, J., Krahmann, G., Millot, C., Rhein, M., Tintore, J., 1999. Recentadvances in observing the physical oceanography of the western MediterraneanSea. Progress in Oceanography 5 (44), 3764, 587, 591.

    Shchepetkin, A.F., McWilliams, J.C., 2005. The regional oceanic modeling system(ROMS): a split-explicit, free-surface, topography-following-coordinate oceanicmodel. Ocean Modelling 9, 347404.

    Skamarock, W.C., Klemp, J.B., Dudhia, J., Gill, D.O., Barker, D.M., Duda, M.G., Huang,X.-Y., Wang, W., Powers, J.G., 2008. A Description of the Advanced ResearchWRF Version 3, NCAR/TN475+STR NCAR TECHNICAL NOTE, June 2008. .

    Smith, W.H.F., Sandwell, D.T., 1997. Global sea oor topography from satellitealtimetry and ship depth soundings. Science 277, 19561962.

    SSALTO/DUACS User Handbook: (M)SLA and (M)ADT Near-Real Time and DelayedTime Products. SALP-MU-P-EA-21065-CLS.

    Testor, P., et al., 2010. Gliders as a component of future observing systems. In: Hall,J., Harrison D.E., Stammer, D. (Eds.), Proceedings of the OceanObs09: SustainedOcean Observations and Information for Society Conference, vol. 2, Venice,Italy, 2125 September 2009, ESA Publication WPP-306.

    Tintor, J., Wang, D.P., La Violette, P.E., 1990. Eddies and thermohaline intrusions ofthe shelf/slope front off Northeast Spain. Journal of Geophysical Reseach 95,16271633.

    Tintor, J., G. Vizoso, B. Casas, L. Renault, S. Ruiz, B. Garau, A. Pascual, M. Martnez-Ledesma, LL. Gomez-Pujol, A. Orla (2012), SOCIB: the impact of new marineinfrastructures in understanding and forecasting the coastal oceans: someexamples from the Balearic Islands in the Mediterranean Sea. In: Briand, F. (Ed.),Planning repeated basin-wide surveys for climatic studies in the MediterraneanSea. Supetar (Island of Brac, Croatia), 1114 May 2011. CIESM WorkshopsMonographs, Monaco, 164 p.

    Vignudelli, S., Cipollini, P., Roblou, L., Lyard, F., Gasparini, G.P., Manzella, G., Astraldi,M., 2005. Improved satellite altimetry in coastal systems: case study of theCorsica Channel (Mediterranean Sea). Geophysical Research Letters 32, L07608.http://dx.doi.org/10.1029/2005GL022602.

    Vignudelli, S., Kostianoy, A.G., Cipollini, P., Benveniste, J. (Eds.), 2011. CoastalAltimetry. Springer-Verlag, Berlin Heidelberg. http://dx.doi.org/10.1007/978-3-642-12796-0.

    Volkov, D.L., Lee, T., Fu, L.-L., 2008. Eddy-induced meridional heat transport in theocean. Geophysical Research Letters 35, L20601. http://dx.doi.org/10.1029/2008GL035490.

    Wang, F., Vieira, M., Salat, J., Tintor, J., La Violette, P.E., 1988. A shelf/slope lamentoff the Northeast Spanish Coast. Journal of Marine Research 46, 321332.

    J. Bouffard et al. / Progress in Oceanography 106 (2012) 6279 79

    Sub-surface small-scale eddy dynamics from multi-sensor observations and modeling1 Introduction2 Study area3 Materials and methods3.1 Satellite altimetry3.2 The coastal glider3.3 Model description

    4 Results4.1 Hydrodynamics from observations4.1.1 Synoptic view from remote sensing4.1.2 Surface currents from along-track altimetry and glider measurements4.1.3 Hydrography from glider measurements

    4.2 Model validation4.2.1 Surface comparison with remote sensing4.2.1.1 The synoptic surface view for the JulyAugust 2008 period4.2.1.2 Surface changes during the glider mission

    4.2.2 Comparison with glider data4.2.2.1 Hydrography4.2.2.2 Currents

    4.3 Spatio-temporal variability from numerical modeling

    5 Discussion and conclusionsAcknowledgementsAppendix AReferences