Lagrangian evolution of a mid ocean anticyclonic eddy

ELISABET RODRÍGUEZ CRUZ Curso 2014/2015 PABLO SANGRÀ INCIARTE DIANA GRISOLÍA SANTOS Trabajo Fin de Título para la obtención del título en el Grado de Ciencias del Mar LAGRANGIAN EVOLUTION OF A MID OCEAN ANTICYCLONIC EDDY Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz LAGRANGIAN EVOLUTION OF A MIO OCEAN ANTICYCLONIC EDDY •Datos del Estudiante: Nombre: Elisabet Rodríguez Cruz Titulación que cursa: Grado en Ciencias del Mar Curso: 4° Curso Créditos superados: 210 ECTS Institución universitaria: Universidad de Las Palmas de Gran Canaria •Datos del Tutor: Nombre: Pablo Sangra lnciarte Departamento: Física Empresa: Instituto de Oceanografía y Cambio Global (IOCAG) •Datos del Co-Tutor: Nombre: Diana Grisolía Santos Departamento: Física Empresa: Instituto Universitario de Oceanografía y Cambio Global (IOCAG) •Proyecto: PUMP (CTM2012-33355). •Financiado por:. Ministerio de Economía y Competitividad Firma estudiante: Firma Tutor: Firma Co-Tutor: 2 Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 3 Índice 1. Introduction 4 2. Methodology 6 2.1. PUMP eddy survey and buoys deployment. 6 2.2. Estimating the Eddy center trajectory. 7 3. Results 7 3.1. Buoys trajectories and mean periods. 7 3.2. Eddy translation and orbits. 9 3.3. Orbital Radius. 9 3.4. Velocities and periods. 10 4. Discussion 11 5. Summary and conclusions 13 References 14 6. Actividades realizadas 26 7. Formación recibida 27 8. Nivel de integración e implicación dentro del departamento 28 9. Aspectos positivos y negativos más significativos relacionados con el desarrollo del TFT 28 10. Valoración personal del aprendizaje conseguido a lo largo del TFT 28 Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 4 Lagrangian evolution of a mid ocean anticyclonic eddy 1Elisabet Rodríguez-Cruz, 1Pablo Sangrà, 1Diana Grisolía 1 Instituto de Oceanografía y Cambio Global (IOCAG). Universidad de Las Palmas de Gran Canaria Abstract The lagrangian evolution of 4 months old anticyclonic intrathermocline eddy of the Canary Eddy Corridor is investigated from the trajectories of 5 satellites tracked drifting buoys. Buoys were drogued below and above the Ekman depth at 15 m and 100 m, respectively. One buoy remained inside the eddy during almost 4 months being thus a long lived coherent feature with a life span of at least 8 months. The eddy consisted in a central core rotating in solid body rotation with a rather constant periodicity of 4 days and in an outer ring rotating much more slowly with periodicities between 8 and 12 days. It translated westward with a slight equatorward deflection at an average speed of 3.5 km day-1 which is close to the phase speed of nondispersive baroclinic Rossby waves for this latitude, being thus strongly nonlinear. Buoys orbital radius variability analysis indicates that the eddy is elliptical and that its experience successive stages of increasing ellipticity followed by axisymetrization which is indicative of submesoscale filamentation. This analysis also reveals that when a surface drogued buoy converges toward the eddy center, a deep drogued buoy diverges toward the eddy periphery and vice versa. We propose that this is related with the switching of the secondary circulation between upwelling a downwelling modes being its periodicity of c.a. 20 days. This was observed for the first 65 days, in the rest of the period, the buoys orbital radius also fluctuates but in phase, indicating that the eddy also evolves pulsating increasing and decreasing its radius. 1. Introduction Mesoscale eddies are nearly ubiquitous features of the World Ocean (Chelton et al., 2007, 2011). They occupy the 25 % of the ocean surface at any time (Chaigneau et al., 2009). They may be viewed as frontal structures with nearly circular shape. Therefore they isolate physical and biogeochemical properties at their interior and introduce gradients of those properties at their periphery, while they are advected. Their radius varies with the first baroclinic Rossby radius of deformation which is O (100-50 km) for low and mid latitudes and O (10 km) for high latitudes (Chelton et al., 1998). There are mainly three eddy types: cyclones, anticyclones and intrathermocline eddies (e.g. McGillicuddy et al., 2007). Due to the geostrophic adjustment, cyclonic eddies are recognizable by the upward doming of the isopycnals, while anticyclones are recognizable by their downward displacement. Intrathermocline eddies, also often referred as mode-water eddies, are a particular type of anticyclonic eddies being their Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 5 most remarkable feature the associated dome shape of the isopycnals in the shallower layers and bowl shape in the deeper layers forming thus lens-like structures overall. A 10 percent of the global eddy population are long-lived structures with lifetimes >4 months (Chelton et al., 2007, 2011; Sangrà et al., 2009). They have an average lifetime of 32 weeks and an average propagation distance of 550 km (Chelton et al., 2011). They propagate nearly due westward at approximately the phase speed of nondispersive baroclinic Rossby waves with preferences for slight poleward and equatorward deflection of cyclonic and anticyclonic eddies, respectively (Chelton et al., 2011). Long lived anticyclones slightly dominate over cyclones. The Canary Archipelago is a continuously source of mesoscale eddies as consequence of prevailing currents and winds perturbation by the islands topography (Arístegui et al., 1994; Sangrà et al., 2005, 2007, 2009). Sangrà et al. (2009) observed that Canary Island induced eddies contribute to a zonally oriented long lived eddy corridor that can extend as far as the Mid Atlantic Ridge; this was named as the Canary Eddy Corridor. It constitutes a major pathway for long lived eddies in the northeastern subtropical Atlantic. Long lived anticyclonic eddies clearly dominate over cyclonic eddies. As observed by Chelton et al (2011) for global eddies those anticyclones propagates westward with a slight equatorward deflection. It is well known that mesoscale eddies modulate biological production and related biogeochemical fluxes (Benitez-Nelson et al., 2007; McGillicuddy et al., 2007; Lévy, 2008). In particular, long lived anticyclonic eddies may exert a profound influence on phytoplankton. In this regard there are growing evidences that point out that anticyclonic eddies may be more productive than cyclones (McGillicuddy et al., 2007; Gaube et al., 2013, 2014). Therefore the classical view that anticyclones are oligotrophic structures while cyclones are productive structures is being abandoned. Self-induced Ekman pumping, as a consequence of eddy-wind interaction, may lead to upwelling in the interiors of anticyclonic eddies (Martin and Richards, 2001; McGillicuddy et al., 2007; Gaube et al., 2013). This will lead to an increase of primary production through nutrient injection to the euphotic zone at eddies centers. Eddy-wind interaction through nonlinear Ekman pumping may also enhance primary production at eddy peripheries (Mahadevan et al., 2008). Other mechanisms that can enhance primary production in anticyclones are the upwelling generated during the decay of anticyclones, often referred to as ‘‘eddy pumping’’ (Falkowski et al., 1991), the stirring of the ambient chlorophyll field by advecting phytoplankton around eddy peripheries (Siegel et al., 2007, 2011; Chelton et al., 2011b), and the eddy trapping of high productive waters during their generation (Lehahn et al., 2011; Early et al., 2011). Sangrà et al. 2005 suggested that anticyclones inside of the Canary Eddy Corridor may account for a total primary production as high as the Northwest African Upwelling system at the same latitude range. There are very few studies on the lagrangian evolution of the Canary Eddy Corridor anticyclones. In an earlier study Pingree (1996) surveyed an anticyclonic eddy located in the middle of the Canary Eddy Corridor near 27ºN, 22º W and deployed three Argos buoys drogued at ca. 200 m. One buoy tracked the eddy during ca. 16 months along a 1650 km westward trip. The initial rotating period was of 8 days then increasing to 5 Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 6 days due to its subducting character. This author described the eddy as a shallow subtropical subducting westward propagating eddy and named it as Swesty. Sweesties are mode water of intrathermocline eddy type. Sangrà et al. (2005) described during 7 months the life story of an anticyclonic eddy shed by Gran Canaria Island from the trajectories of 3 Argos buoys drogued at 100 m depth. It drifted southwestward up to 500 km with the mean Canary Current. The eddy evolved from a young stage, where the core retains its vorticity and occupies most of the eddy being in solid body rotation, through a mature stage, where the eddy has a reduced inner core and a slowly revolving outer ring, to a decay stage, where the vorticity maximum is substantially reduced. Rotating rates varies from 3 days for the young stage to 6 days for the mature and decay stages. The eddy evolved pulsating decreasing and increasing its radius. A typical anticyclonic eddy of the Canary Eddy Corridor named as the eddy PUMP, located 300 nautical miles southwest of the Canary Islands, was interdisciplinary surveyed on September 2014 in the framework of the PUMP project (ref: CTM201233355). This project aims to investigate the modulation of the biogeochemical fluxes by the ageostrophic secondary circulation (SAC) and mixing in anticyclonic mesoscale eddies. The eddy, 4 months old, was an intrathermocline type eddy characterized by a dome shape of the isopycnals in the sallower layers and bowl shape in the deeper layers. It was elliptical, 110 km diameter and 400 m deep. This study aims to describe lagrangian evolution of this eddy from 6 drifting buoys trajectories. As it will be showed it will provide news insight on this evolution such as the occurrence of axisymetrization episodes or the switching between upwelling and downwelling modes. 2. Methodology 2.1. PUMP eddy survey and buoys deployment In the framework of the PUMP project an anticyclonic eddy was surveyed from 04 to 20 of September 2014. Figure 1 depicts the eddy location in September 13, as obtained from merged altimetry AVISO data. It was generated by Tenerife Island four months before the survey, hence being four months old. Its signal is easily recognizable by a strong positive sea level anomaly (SLA) 300 nautical miles (nm) southwest of the Canary Island inside the Canary Eddy Corridor (CEC; Sangrà et al., 2009). As already introduced, the CEC is built up by the Canary Islands induced eddies as consequence of prevailing wind (Trades winds) and currents (Canary Current) perturbation. Long lived (life span > 4 months) anticyclonic eddies are more frequent than long lived cyclones due to their higher initial rotating rate (Sangrà et al., 2007). The PUMP eddy can be viewed as a typical anticyclonic eddy of the CEC. As part of the cruise strategy we conducted a high resolution meridional transect named as “Le Tourmalet”, crossing the eddy center (Figure 2). As indicated in Table 1 and illustrated in Figure 2 we deployed 6 buoys along this transect at different distances from the eddy center. Three buoys (buoys 01, 02 and 06) were drogued inside the Ekman Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 7 layer (15 m) and three (buoys 03, 04 and 05) below the Ekman layer (100 m). Drogues were a holey sock type with 1 m of diameter and 15 m of longitude. Position data were transmitted through the SPOT satellite messenger system which is a low cost system with a yearly flat rate. Averaged time resolution was about 1 hour. 2.2. Estimating the eddy center trajectory Eddy center trajectory is calculated using a method proposed by Brassington, G. B. (2009) based on that, the lagrangian trajectory given by a buoy inside the eddy exhibits an oscillatory motion. In that way, plotting each orthogonal coordinate of the time series position will provide two extremun points per completed orbit that will correspond to the northern and southern points for the meridional coordinate and the eastern and western points for the zonal one (Figure 6). The maxima and minima for each coordinate will be interpolated in order to get the extrema for each instant of time. Its averaging will give us the eddy center trajectory. To apply this method to our data, we selected buoys 5 and 6 as their remained the longest time inside the eddy and in addition one was drogued at 100 m (buoy 5) and other at 15 m (buoy 6). Only data while the buoys remained inside the eddy were considered. Buoys position will be treated separately for each coordinate. Firstly, we worked with RAW data, taking the deployment date, 14 of September, as the day zero (t=0) at 02:00 local hour of Madrid and we removed outlayers. Data had to be filtered using a low pass butterworth filter to remove inertial oscillations with a cut-off frequency of two days. The next step consists in interpolating the extrema for each coordinate. Once the extrema are identified from the filtered time series, as shown (Figure 4), it is applied a cubic spline interpolation to maxima and minima points separately. Figure 5 shows the interpolated curves obtained for each coordinate. As it can be seen from this Figure, the beginning and the end of the interpolated curves for the extrema do not fit in time, given that the maxima and minima of the series start at different times. Therefore, due to both a maximum and a minimum is necessary to compute the eddy center position at a given time, the interpolated data which start in the first minima to last maxima will be used corresponding to 16 of September of 2014 at 20:00 to 13 of February of 2015 at 07:00 for buoy 5 and to 18 of September of 2014 at 08:00 to 29 of January of 2015 at 20:00 for buoy 6. Figure 7 shows the eddy center trajectory and the buoy position for buoy 6 and buoy 5. 3. Results 3.1. Buoys trajectories and mean periods Figure 3 illustrated buoy´s trajectories while they remain inside the eddy. Buoy 1 only transmitted its position during 6 days hence it was discarded for the analysis. Buoy 2 drogued at 15 m depth remained inside the eddy during ca. 32 days tracing 4 loops (Table II). Its trajectory indicates that during these first 25 days the eddy translates Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 8 westward. Eddies self-translates westward due to the gradient of planetary vorticity (βeffect) along its radius (Cushman-Roisin, 1994; Van Leeuwen, 2007). This induces a net Coriolis force that must be compensated by a gradient pressure force that forces the eddy westward displacement. Eddies trajectory will also rely on their interaction with the mean flow (Cushman-Roisin, 1994). In this regard, the westward propagation of the eddy suggests that during this first period self-westward translation dominates over the background flow advection indicating that this later must be relatively low. From a detailed inspection of the buoy 2 trajectory, we can notice smalls near closed loops superposed to the mains loops. These are the signature of near inertial waves (NIW). Trapping of NIW by anticyclonic eddies has been predicted theoretically and observed for anticyclonic eddies close to the Canary Islands (Kunze, 1985; Lueck and Thomas, 1986; Martínez-Marrero et al., 2014). Inside the eddy, the frequency of the NIW is reduced by the negative value of the eddy relative vorticity. Then the resulting frequency value for NIW inside the eddy will be below the inertial value, determined by the latitude. This will cause that NIW reflect as they reach the eddy boundary resulting on its trapping by the eddy. Buoy 6, also drogued at 15 m, was the one that remained more time inside the eddy tracking, 13 loops during 145 days (Figure 3, Table II). Its trajectory indicates that after the first 25 days the eddy translates southwestward until day 110 and then, it moves again westward. This suggests the occurrence of an intensification of the southward flowing Canary Current during days 25-110. Superposed small near closed loops associated to NIW are also noticeable all along this buoy trajectory. Buoy 3 and 4 drogued at 100 m remained 32 and 46 days inside the eddy, tracking 6 and 13 loops respectively (Table II). Buoy 5, also drogued at 100 m, together with buoy 6 remained more than three months inside the eddy drawing 36 loops along 123 days. When the buoys where deployed the eddy was 4 months old. As buoy 6 remained inside the eddy during near four more months the eddy life span is at least 8 months being thus a long lived coherent structure. This indicates that this structure is stable to inertial perturbations. The inertial stability criteria depends on the eddy type, Rankine or Gaussian, and on its initial rotating period (Sangrà et al., 2007). We may obtain an approach to the eddy mean period dividing the number of buoys loops by the time-span while they remained inside the eddy (Table II). This period varies from 3.5 days for buoy 5 closer to the eddy center, to 11 days for buoys 6 orbiting at the eddy periphery. This indicates that the eddy was not in solid body rotation with an inner core rotating faster than the periphery, being thus a Gaussian type eddy. As we will discuss in section 4, these rotating rates for a Gaussian vortex render the eddy stable to inertial perturbation being thus able to evolve as a long lived coherent structure. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 9 3.2. Eddy translation and orbits The next parameters that we are going to describe have been only calculated for buoys 5 and 6, being those that remained more time inside the eddy as already mentioned. Figure 7a and b show the whole trajectory and the center eddy trajectories for these two buoys. Remember that centroids locations give us an approach to the eddy center displacement/trajectory or eddy translation. In both cases centroids positions cross the loops centers indicating that the methodology was well approached to the eddy center trajectory (Figure 7). Although this trajectory is similar for both buoys, there is some mismatch. There are some loops like displacements on buoy´s 5 eddy center trajectory that are not presented in buoy 6. This is probably related to high rotating rate of buoy 5, while the filter to remove high frequency variability was the same as for buoy 6. As already detailed from the whole trajectories analysis, eddy center trajectory indicate that initially it translates westward, then it moves southwestward probably due to its interaction with the Canary Current, and finally self-translates again westward. Although mean eddy center speed for both buoys are similar, there is clearly mismatch for the instantaneous speed as shown is Figure 8. Time series for buoy 5 show a high frequency variability about 10 days of period which is not present in buoy 6 time series. This high frequency variability is related with the loops observed in the trajectory of the eddy center for buoy 5 mentioned above, being thus probably a methodological artifact. When this variability is filter out, both times series resembles with a rather constant velocity although with small amplitude low frequency variability. We may obtain the buoys orbits around the eddy center subtracting the eddy center trajectory (eddy drift) to the whole trajectory (Figure 9). This will allow us to describe the lagrangian properties of the eddy removing the translation effects. Orbits for the interior buoy (buoy 5) show a near circular shape with maximum radius of ca. 10 km indicating that this buoy remained very close to the eddy center for the 123 days period (Figure 9a). It described 36 clockwise near circular revolutions around the eddy center being thus its mean period about 3.4 days (Table II). Buoy 6, located at the eddy periphery, describes only 13 revolutions along a 145 days period being thus the mean period much lower, ca. 11 days (Figure 9b). This confirm that the eddy is a Gaussian type, where the periphery rotates much slower than the eddy core. Outer orbits of buoy 6 are elliptical with the major axis oriented zonally, being their maximum radius about 50 km. 3.3. Orbital Radius Figure 10 displays the instantaneous and mean (9 days) orbital radius for the two buoys. This mean radius has been obtained with a low-pass filter of the instantaneous values (1 hour data) using a Lanczos filter with a 9 days cutoff period. In all cases the instantaneous radius fluctuates around a slowly changing mean value, which indicates that the buoy orbits are elliptical. Notice that those elliptical related fluctuations are of higher frequency and smaller amplitude for the inner buoy (buoy 5). Notice also, that Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 16 Journal of Physical Oceanography, 9(6), 1155–1182 Melander, M. V., Zabusky, N. J. & McWilliams, J. C., (1988). Symmetric vortex merger in two dimensions: causes and conditions. Journal of Fluid Mechanics, 195, 305– 340. Percival, D. B., Walden, A., T., (2000). Wavelet Methods for Time Series Analysis, Cambridge University Press, New York. Piedeleu, M., (2014). Remolinos oceánicos de las Islas Canarias: generación, características y evolución. PhD thesis, Universidad de Las Palmas de Gran Canaria, Spain. Sangrà, P., Pelegrí, J. L., Hernández-Guerra, A., Arregui, I., Martín, J. M., Marrero-Díaz, A., Martínez, A., Ratsimandresy, A. W., Rodríguez-Santana, A., (2005). Life History of an anticyclonic eddy. Journal of Geophysical Research, 11, C03021. Sangrà, P., Auladell, M., Marrero-Díaz, A., Pelegrí, J.L., Fraile-Nuez, E., RodríguezSantana, A., Martín, J. M., Mason, E. and Hernández-Guerra, A., (2007). On the nature of oceanic eddies shed by the Island of Gran Canaria. Deep-Sea Research I, 54, 687-709. Sangrà, P., Pascual, A., Rodríguez-Santana, A., Machín, F., Mason, E., McWilliams, J. C., Pelegrí, J. L., Dong, C., Rubio, A., Arístegui, J., Marrero-Díaz, A., HernándezGuerra, A., Martínez-Marrero, A. and Auladell, M., (2009). The Canary Eddy Corridor: a major pathway for long-lived eddies in the subtropical North Atlantic. Deep-Sea Research I, 56, 2100-2114. Siegel, D., Court, D., Menzies, D., Peterson, P., Maritoena, S., Nelson, N., (2007). Satellite and in situ observation of the bio-optical signatures of two mesoscale eddies in the Sargasso Sea. Deep Sea Research, Part II, 55, 1218–1230. Siegel, D., Peterson, P., McGillicuddy Jr., D. J., Maritorena, S., Nelson, N., (2011). Biooptical footprints created by mesoscale eddies in the Sargasso Sea. Geophysical Research Letters, 38, L13608 Van Leeuwen, P. J., (2007). The propagation mechanism of a vortex on the plane. Journal of Physical Oceanography, Vol. 37, pp. 2316-2330. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 17 BUOY ID DROUGUE DEPTH (meters) DATE TIME (GMT) LATITUDE LONGITUDE PUMP 03 PUMP 01 100 15 13/09/14 03:17 03:37 26º 23.019’N 26º 22.014’W 20º 20.067’W 20º 19.980’W PUMP 05 PUMP 02 100 15 13/09/14 08:30 03:37 26º 13.025’N 26º 11.975’W 20º 20.018’W 20º 19.976’W PUMP 04 PUMP 06 100 15 13/09/14 13:57 14:13 26º 07.980’N 26º 08.736’W 20º 19.845’W 20º 19.950’W Table I. Deployment characteristics of the drifting buoys. Buoy Loops Time span (days) Mean period (days) 02 4 32.5 8.1 06 13 145 11.1 03 6 32.5 5.4 04 13 46 3.5 05 36 123 3.4 Table II. Buoys trajectory inferred parameters. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 18 Figure 1. Sea Level Anomaly (SLA) image from aviso, showing the location of the anticyclonic eddy PUMP. Black dots indicate CTD stations along a transect crossing the eddy, which was named as “Le Tourmalet”. Station numbers at the transect end are also indicated. Figure 2. Buoys deployments location (yellow dots) superposed to 13 of September of 2014 SLA image. “Le Tourmalet” transect CTD stations are indicating by black dots. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 19 Figure 3. Full buoys trajectories within the eddy. The numbers indicate the days (multiples of 10) elapsed after the buoys deployment. Dots are drawn every 10 days. Drogue depth is also indicated. Day 0 corresponds to 14 of September 2014. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 20 Figure 4. (a),(c) Maxima and minima points of longitude and (b),(d) latitude trajectory for buoys 5 and 6. Figure 5. (a),(c) Extrema interpolation of latitude and (b),(d) longitude for buoys 5 and 6. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 21 Figure 6. Buoys trajectories with maxima and minima points for longitude and latitude. (a) Buoy5. (b) Buoy6. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 22 Figure 7. Temporal evolution of the eddy center as traced (red line) by (a) buoy 5 and (b) buoy 6 superposed to full trajectory (black line). The numbers indicate the days (multiples of 10) elapsed after the buoys deployment. Dots are drawn every 10 days. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 23 Figure 8. Speeds of the eddy center unfiltered (red line) and 10 days filtered (blue line) for (a) buoy 5 and (b) buoy 6. Figure 9. Buoys’ orbits corresponding to the complete buoys’ time series. (a) Buoy 5. (b) Buoy 6. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 24 Figure 10. Instantaneous (red line) and mean (blue line) orbital radius for buoy 5 and buoy 6. Figure 11. Upper panels, zonal velocities (eastward positives) for buoys (a) 5 and (b) 6. Lowers panels, kinetic energy spectra as a function of time as obtained from the wavelet analysis of the zonal velocities for buoys (a) 5 and (b) 6. Lagrangian evolution of a mid ocean anticyclonic eddy Elisabet Rodríguez Cruz 25 Figure 12.Schemtics of the secondary circulation for (a) upwelling mode and (b) downwelling mode.