News-Blog‎ > ‎

Cross-frontal exchanges of salt detected by SMOS in the Gulf Stream

posted Mar 28, 2013, 7:11 AM by Salinity CERSAT   [ updated Mar 28, 2013, 10:30 AM ]
                                                           by Nicolas Reul and Bertrand Chapron
with contributions from G. Alory and J. Boutin

Sketch explaining the ocean circulation features in this area. A SMOS 10-days averaged SSS map is shown in the background with vectors representing surface currents derived from altimer and surface wind products (OSCAR).

The Gulf stream is probably the most intensely studied ocean current. Over the past 70 years, many hydrographic sections have been taken accross the stream between Florida straits and the Grand Banks. These sections always reveal strong property gradients aligned with the current, indicating that the stream is a boundary between cold, fresher slope and shelf waters and warm, more saline central waters (Bower and Rossby, 1989). Biologists have also recognized the role of the Gulf stream as a boundray between two environments, each with its own unique set of flora and fauna.

The GulfStream is a turbulent jet that migrates into deeper waters after reaching Cape Hatteras, forming large-amplitude meanders downstream of Cape Hatteras from baroclinic and barotropic instability processes (see sketch above). Large enough individual meanders can separate from the main current, loop back on to themselves and form independent rings (Saunders, 1971; Csanady,1979). Rings that form from Gulf Stream meander crests engulf parcels of warm, salty Sargasso Seawater in their core and begin to interact with their surrounding waters through either mixing or stirring processes(Parker,1971; Churchilletal.,1993)

The common occurrence of Gulf Streal warm-core rings (WCRs) in the Slope & shelf seas, and their role in initiating cross-frontal events like shelf water entrainment in the western North Atlantic, have been well documented through satellite altimetry, sst and chlorophyll imagery, theoretical models and field observations. However, little is known about the salt transported by these Warm core rings and the new satellite microwave sea surface salinity products might bring  very interesting insight in that context.

During the first two years of mission operations, SMOS data in this area were howevever heavily contaminated by Radio Frequency Interferences (RFI) from the military radar arrays installed over North America. Since end 2011, a dramatic improvment was observed due to on-going action to refurbish L-band radar stations in Canada (see dedicated news).  In that frame, we analyzed the complete year 2012 of SMOS data to investigate if we can now tackle realistic SSS variability in the Gulf stream area with SMOS data. First results were highlighted during the 3-years SMOS press event organized by ESA in Madrid (see here).  Results from a more in depth analysis are presented herebelow.


 Figure 1: Climatologies of (a) the Mean Dynamic Topography (color) and geostrophic currents (arrows) (Rio, 2009); (b) World Ocean Atlas SSS; (c) Pathfinder SST (Casey) and (d) SeaWIFS 1997-2007 Chlorophyll. The black curve extending across the North Atlantic is the separating streamline estimated from the steady atltimer-based streamfunction.

   Annual climatologies of Sea surface Height (SSH), SSS, Sea Surface Temperature (SST) and Chlorophyll-a shown in Figure 1 reveal that in average , very large  SSH, SSS, SST and Chl-a concentration gradients are encountered along South-North sections trough the gulf stream. In particular the SSS exhibits zonal gradients greater than 5 unit/10° of  latitude and the SST can range from 24°C in the central waters down to less than 5°C in the shelf and slope waters. It therefore a good test-area for evaluating the ability of microwave salinity missions to retrieve SSS in moderate to cold seas.

 Specific scientific questions we are investigating are the following:

1) What accuracy of SMOS products at moderate to low SSTs ?


2) How SMOS data complement SST & SSH informations to better track meso-scale features


3) Can SMOS data be used to better monitor biological productivity in the separated Gulf stream area ?


4) Is SMOS able to track Near-Surface

Transport Pathways of salt in the North Atlantic Ocean ?
ÞLooking for Throughput from the Subtropical to the Subpolar Gyre


To answer the first question above, we conducted an ensemble of co-localizations between SMOS and in situ data collected in that area over year 2012.





Figure 2: SMOS SSS data (a) co-localized with in situ SSS data (b). In situ observations include VOS TSG data and ARGO float upper depth measurements in the Gulf stream area. The data are collected from 03/15/2012 to 10/31/2012. TSG data were spatially averaged at the SMOS product 0.25° x 0.25° spatial resolution. SMOS SSS data were temporally averaged over a 11 days-period centered on each ship or ARGO observation date. c) Comparison between co-localized SMOS and TSG SSS data as function of sea surface temperature (color). d) Comparison co-localized SMOS and ARGO SSS data as function of sea surface temperature (color); (e) Average differences ΔSSS between co-localized in situ and SMOS SSS observations per 0.2°C-width bins of Sea Surface Temperature. The black curve indicate the median  ΔSSS within each bin and the blue vertical bars indicate ±1 standard deviation. The number of samples per bin is indicated by the green curve.




 We considered Volountary Observing Ships Thermo-SalinoGraph data averaged at 0.25°, ARGO & SMOS data 10 days average at 0.25° (CEC-Ifremer v2 product).  In situ data include  ARGO (~2000 pts) and TSG (GOSUD (Matisse+Oleander)+ SAMOS data (Atlantis, Henri Biglow, Knorr, Nancy Foster, Pisces, Okeanos Explorer, RonBrown) ) with a spatial sampling distribution as shown in Figure 2. In general, we found an rms difference between SMOS and in situ observations of 0.7 and 0.5 for TSG and ARGO float data, respectively.  Given the large gradient and strong temporal variability encountered in this area, SMOS products data can therefore be used to perform some scientific analysis.  However, bining the in situ-minus-SMOS SSS differences as function of SST, it appears (Fig 2d above) that systematically, SMOS L3 CEC V2 products are too salty compared to in situ observations as SST decreases below about 13°C.  The colder the SST below that threshold, the more important the bias. The reasons for that are not yet fully clear but certainly involve a mixture of dielectric constant modelling issues (we rely on Klein adnd Swift's model), badly corrected roughness effects, rfi and land contaminations.



  How SMOS data complement SST & SSH informations to better track meso-scale features ?


To investigate that point, we combined SMOS observations with GHRSST SST products (ODYSSEA) and merged altimeter geostrophic+wind-induced ekman currents from OSCAR. The animation below is showing the co-varying SSS (color) and currents (arrows) in the Gulf stream area in 2012 (clik on the image to see a larger view) 


With an unprecedented space and time resolution, SMOS surface salinity data now bring additional information on the evolution of the meandering Gulf Stream. Off Cap Hatteras, strong lateral gradients are caused by the convergence of subtropical water carried northward by the Gulf Stream with sub-polar water carried southward along the east coast of North America by the Labrador current. Stirred by mesoscale (50-100 km) and larger scale processes, SMOS observations help delineate meanders pinching off from the current to form Gulf Stream rings. Saltier (fresher) water parcels from the southern (northern) water masses are then trapped in these eddies, to further propagate in distinct ocean salinity environment on both sides of the stream. 

Two examples of this capabilities of SMOs sensor to detect meso-scale features are given by the signal of warm-core and salty eddies detached from the Gulf stream in June and September around 65°W and 40°-42°N:




By chance the Matisse ship equipped with a TSG crossed the salty eddy dettached from the Gulf stream which was detected from mid-may to end July (top one in the figure above) with data acquired close in time to an ARGO float surfacing just in the eddy core:

Figure 3: Top: 10 days-averaged SMOS SSS field (color) superimposed with OSCAR currents (arrows) from 26/05 to 5/06. Colored squares indicate the location and SSS measured by ARGO profilers  from 24/05 to 8/06. Small colored circles indicate Matisse ship track and SSS from 24/05 to 3/06. Bottom: SSS measured by Matisse TSG at high resolution (thin black curve) and smoothed at 0.25° resolution along track (blue). The red curve shows the 10-days averaged SMOS SSS interpolated in space and time along the ship track. The small black square indicates the SSS measured by the ARGO float (#4901139) located close to the ship track at 64,42°W;41,24°N on 8 of June (see top panel).
As expected, the ±25 km and +/-5 days averaging of SMOS data is somehow resulting in smoother  SSS gradients than the very local in space and time in situ observations. The salty anomaly signal detected by SMOS in that eddy is however in general consistent with the SHIP TSG and ARGO float measurements. 

Figure 4: Latitudinal Sections at 64.4°W across the salty eddy centered at 41°N and detected in may-june of SSS (a) and Sea Level (b).  On (a), red & black curves show 10-days averaged SMOS SSS  from 24/05-29/05 and from 3/06 to 13/06, respectively; World Ocean Atlas SSS climatology  is illustrated by the black dashed-dotted curves. The SSS measured at 6 m depth by ARGO float #4901139 along that latitudinal section are indicated by colored squares: red on May 29  and black on 8 of June. In (b) red and black curves are showing the merged sea level height deduced from AVISO on the 29/05 and 8/06. The dashed black curve is indicating the mean dynamic topography (Rio, 2009).

We compared SMOS and in situ observations to the World Ocean Atlas SSS climatology along a South-North section at 64.4°W acrross the eddy. As shown by  Figure 4 (a) , SMOS detected  a significant +2-+2.5 salty anomaly in the northern flank of the Gulf Stream around 41°N. The peak SSS at 41°N also correspond to a +30cm sea level height anomaly, indicating the signature of the warm and salty core eddy.

 Apparently, as revealed by S and T vertical distiburtions at the ARGO float located into that eddy, that meso-scale structure (~200 km diameter)  transported a very significant depth-integrated anomaly of both heat and salt:



According to historical CTD and profiler observations from NODC collected within [67°W-63°W; 40°N-42.5°N], this is a fairly rare situation as detected from the surface salinity perspective:


Less than 5% of historical observations in that box indeed exhibit SSS >34.5!
Coherency between SSS satellite observations and SST, SSH and chlorophyll-a data was further analysed by estimating the co-variation of these variables along the latitudinal section throughout year 2012.  This is illustrated here below:

Figure 5: Time-Latitude Hovmöller diagrams at 65°W of (a) SMOS SSS; (b) AVISO merged altimeter SSH; (c) GHRSST Level 4 ODYSSEA SST and (d) 8-days composite MODIS total Chlorophyll-a. The black curve is the iso-haline at 35.6 from SMOS data.

The plots in Figure 5 reveal that, SMOS SSS is an excellent proxy of the sea level height variability in this region, particularly during the summer season (may to end october). During that period, the enhanced signals associated with the 2 eddies previously detected in may-july & aug-oct clearly appear north of ~39°N in both SSS and SSH Hovmuellers. In addition, the Chlorophyll-a concentration is showing 3 distincts water masses: the rich region in the North corresponding to cold (and generally fresh waters), the poor subtropical gyre waters (warm and generally salty) and a transition zone, certainly corresponding to the mixing-zone induced by the stream turbulence. The chlorophyll-a concentration is certainly SST-driven at first order throughout the year and particularly during winter. However, in summer, the chrolorphyll-a poor region is clearly well delineated by the iso-haline at 35.6 (deduced from SMOS data). This suggests a significant dependency of chl-a concentration with SSS and SST during that season.

 As illustrated by Figure 6, by bin-averaging MODIS chl-a concentrations as function of co-localized SST and SMOS SSS values, considering all data acquired over the domain defined by [75°W-40°W;30°N-50°N] and during all the year 2012, we were in a position to highlight such dependencies:



 Figure 6: Seasonal variabilty of the Chlorophyll-a concentration average dependencies with SST and SSS in the North Atlantic domain [75°W-40°W;30°N-50°N] for year 2012 (clik on image for better definition).

Little SSS-dependencies are found in the Chlorophyll-a concentration during winter (left plot). However, during the warm boreal summer (right plotà, as SST exceeds over 17°C, Chlorophyll-a concentrations is observed to drop significantly at a given SST with increasing SSS  from 34 to 37. Salt transported by warm-core eddies detached from the Gulf stream and their summer incursions in the shelf and slope sea is therefore a potential moderator mechanism for biology in this region.  

Similar Gulf-stream meanders-induced warming events were recently highlighted near the Middle Atlantic Bight shelfbreak using sea surface temperature imagery and altimetry by Gawarkewicz et al   together with in situ salinity. Now that SMOS and Aquarius data are available, new and  complementary information on cross-frontal exchanges of salt across the gulf stream can be regularly gained. Better knowledge of such events which have significant implications for the shelf ecosystem (e.g. silver hake population) is certainly a progress.