News-Blog‎ > ‎

Horizontal advection of the Amazon Freshwater Plume at the surface of the North Western Tropical Atlantic as detected in 2010 by SMOS and Altimetry

posted Jun 21, 2011, 7:59 AM by Salinity CERSAT   [ updated Jul 1, 2011, 3:29 AM ]

The Amazon River, whose average discharge at Obidos, Brazil is (1.55 x 105 ± 0.13) m3 sec-1 is by far the largest single source of terrestrial freshwater to the ocean and contributes about 30% of total river discharge to the Atlantic Ocean (Wisser et al., 2010). Its flow is distinctly seasonal with minimum discharge occurring in November (~0.8x105 m3 sec-1) and maximum discharge occurring in late May (2.4x105 m3 sec-1) (Lentz, 1995). This flux impacts surface ocean density in the Caribbean more than 3500 km from the Amazon source (Hellweger and Gordon, 2002), and causes a surface discoloration readily detectable in satellite ocean color data that can extend over areas as large as 500,000 km2 (Muller-Karger, 1988; 1995, Longhurst, 1995, Hu et al., 2004; Fratanoni and Glickson, 2002; DelVeccio and Subramaniam, 2004). 

 Figure 1:  Area of interest. The northwestern Tropical Atlantic.  On land, the Amazon River and its major tributaries within our region of interest are shown, as are the Orinoco, Xingu and Tocatins (Toc.) Rivers.  The location of the Obidos, Brazil discharge gauge is indicated as a solid circle.  In the ocean, major currents are indicated: North Brazil Current (NBC), Northern Equatorial Current (NEC), North Equatorial Counter Current (NECC), Guyana Current and Caribbean Current.  The small flow vectors show the mean surface current direction and relative velocity during September. These were created from climatological drifter float trajectory data


Amazon water enters the Western Tropical Atlantic near the equator (Figure 1) and is carried northwestward along the Brazilian Shelf by the North Brazilian Current (NBC), (Muller-Karger et al., 1988; 1995).  Although some of the Amazon’s water continues northward into the Guyana Current (Hellweger and Gordon, 2002), a sizable fraction reaching 3-10ºN is carried eastward by the North Equatorial Counter Current (NECC) (Muller-Karger et al., 1998; 1995; Fratoni and Glickson, 2002), which reaches maximal velocities during the boreal summer-to-fall period (Fonseca et al., 2004; Richardson and Reverdin, 1987).   Along the Brazilian shelf, up to 200km offshore, the plume formed by the Amazon’s discharge lies at the surface with a depth ranging from 3-10 meters (Lentz and Limeburner, 1995).  Beyond the shelf, freshwater within the plume gradually attenuates with depth as it travels away from the source, with a penetration depth of 40m to 45m as far as 2600km offshore (Hellweger and Gordon, 2002; Hu et al., 2004).  Along with the freshwater, the Amazon provides the largest riverine flux of suspended (1200 Mt y-1) and dissolved matter (287 Mt y-1), which includes a dissolved organic matter (DOM) flux of 139 Mt y-1 (Meybeck and Ragu 1997).   These fluxes can have a dramatic effect on regional ecology as they represent potential subsidies of organic carbon, nutrients, and light attenuation into an otherwise oligotrophic environment (Muller-Karger et al., 1995).  

Satellite remote sensing data is known to provide several means to visualize the wide surface dispersal the Amazon plume, with ocean color data being the first to illustrate its reach to well beyond 1000 km (Muller Karger et al., 1988).  Since these first observations, the application of ocean color, altimetry, and SST satellite mapping in this region has increased in its sophistication, showing the ability to track surface plume area (e.g. Hu et al., 2004; Molleri et al., 2010), fronts along the shelf to the NW (Baklouti et al., 2007), and northward propagating eddies or waves shed near the North Brazil Current (NBC) retroreflection region, the so-called NBC rings (Ffield, 2005; Goni and Johns, 2001; Garzoli et al., 2004).   In each case, the satellite data are able to provide time-resolved information on advective processes up to certain limits that include cloud cover, limited/weak SST and ocean color gradients, non-conservative dilution processes for the ocean color to salinity conversions (Salisbury et al., 2011), and baroclinicity and subgrid variability for the altimetry SSHA tracking of the NBC rings. The new sea surface salinity products from satellite plateforms such as SMOS allow to gain a significant insight into the advection paths of the freshwater Amazon plume by surface currents. The SSS product derived from SMOS data for year 2010 is described in detail in a previous news here
 In the present post we tentatively illustrate the strength of these new satellite products to bring added value information on the freshwater transport in this important ocean area. The SMOS estimate for SSS evolution along with geostrophic surface currents estimate from altimeters from April to December 2010 is shown in the animation here below:
Legend: animation showing the daily SSS in the North western Tropical Atlantic computed by averaging SMOS retrievals over a +/- 5-days running window. Superimposed are the AVISO surface currents (black arrows) deduced from merged altimeter data.The solid thick gray contours represent the vorticity for anticyclonic eddies (clockwise rotation or negative vorticity)  and the dashed one represent the vorticity for cyclonic eddies.
As illustrated, a very good visual consistency is found between the altimetric surface current patterns and the SMOS SSS spatio-temporal distribution along the year. Mignot et al. (2007) show a long-term seasonal to monthly climatology that highlights two freshwater offshore pathways - the north passage to the warm pool and eastward entrainment into the North Equatorial Counter Current (NECC) – but they cannot clearly confirm or track this laterally with time in a given year.  Rather, advection is estimated using long-term SSS and current climatologies. In the following, we study the quality of SMOS SSS data and their spatio-temporal evolution along these  two freshwater offshore pathways .
NorthWestward Plume Transport
At the begining of june, the freshwater input at the mouth of the Amazon river is seen to be transported more than 2000 km north westward at the ocean surface and seems to travel as a passive tracer meandering in between  a succession of antycylonic/cyclonic eddy pairs, acting as a gear cluster for this large scale oceanic freshwater pump :
Legend: schematic view of the northwestward transport of freshwater from the Amazone river mouth to the subtropical gyre advected in surface current convergence zones  through a succession of pairs of cyclonic (C) and anti-cyclonic (A) eddies. 
 In the frame of the US/NFS project ANACONDAS (Amazon iNfluence on the Atlantic: CarbOn export from Nitrogen fixation by DiAtom Symbioses)  (see, a cruise have been conducted in this area from end of May to end of June 2010 with collection of UnderWay measurements of Sea surface Salinity and series of salinity profiles conducted at several CTD stations. Comparison between SMOS SSS and in situ data collected during this campaign are shown here below:
 Legend: Top :SSS measured by a TSG along the track of the R/V Knorr in May-June 2010. Bottom: co-localized SMOS SSS at 25 km resolution and averaged over a +/- 5 days running mean window.
Legend: Top: time series of the R/V Knorr SSS sampled at 3 m depth (blue) and of the SMOS SSS interpolated along the R/V track. Bottom: Dates along the R/V Knorr cruise track and locations of the CTD sections (black dots).
As illustrated above, the validation with TSG data show that SMOS well detected the freshwater lenses observed by the ship measurements at this period of the year. 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 ship observations. Largest disagreement between satellite and in situ observations are found when the sampled plume is the closest from the river mouth between the 2 and 6 of June (cyan to yellow color in the above bottom plot). In this case, SMOS detects a significantly fresher plume at the surface than the one seen by the ship at 3 m depth. As revealed by the  CTD vertical profiles in the stations 10 to 16 (see plots below), very strong vertical SSS gradients are found here (sometime reaching  20 pss in the upper 15 m) and a high temporal variability is found in the vertical salinity structure from one profile to the other at the same location on the same day (e.g. station 11), probably indicating tidal impacts on the vertical plume structure. The combination of the plume horizontal, vertical and temporal structure variability and of the SMOS versus in situ sampling specificities may explain these local differences.
Legend: Temperature (red) and Salinity (blue) profiles in the upper 30 m of the ocean at the different CTD stations shown in the figure above. SMOS SSS interpolated at the station location and time are shown by the blue squares.

In general, the surface satellite measurements are found to pretty well correlate with the reported in situ vertical structure.  
Futher comparisons with other underway SSS data from the Colibri and Toucan ships for this period of the year are given here below:

Here again, SMOS detected a fresher plume at the surface than at the 5m depth intake of the ship in the proximity of the coasts. It is not yet clear wether this effect is of geophysical nature (e.g. skin effect in the vertical SSS gradient sensing) or associated with a SMOS brightness temperature that remains slightly contamined when the sensor approach land masses. Nevertheless, this is not a systematic observation, as seen in the below comparison with Toucan underway data, for which SMOS is here slightly saltier than the in situ obs in the plume, suggesting the prevalance of the geophysical variability effect (here propably the temporal variability of the horizontal extent of the plume) in explaining the satellite differences with in situ:

Eastward Freshwater Amazon Plume Transport along the NECC

Eastward entrainment of low salinity water from the mouth of the Amazon river into the North Equatorial Counter Current (NECC) is evident in the top figure animation in the second half of the year. To analyse this transport, virtual drifters were dropped  around the mouth of the river at the beginning of  june and temporally advected with the geostrophic currents deduced from merged AVISO altimeter products. The evolution of sea surface salinity from SMOS and AMSR-E sensors (see Reul et al., 2009 for details on the AMSR-E SSS product) , the Sea surface Temperature analysis products from OSTIA and the merged Meris-Modis CDOM coefficient along the path of such drifters was evaluated. An exemple is shown in the animation here below:

 Legend: Top :spatio-temporal evolution of the location of a virtual drfiter (white dots) dropped at 52°W6°N at the beginning of june 2010 and advected with surface currents estimated from altimetry. Superimposed are the +/-5 days averaged daily SSS fields from SMOS and the surface currents from altimetry (black arrows). The fixed white dots with subscripts numbers 16 and 18 are Pirata mooring at (8°N,38°W) and NTAS moorings (15°N,51°W), respectively. Bottom: time series of the co-localized SSS from SMOS (blue) and AMSR-E sensor (cyan), the analyzed SST (red), and the merged daily CDOM (black) along the drifter path.
As illustrated by this example , it can take  a duration of approximately 6 months and a cummulated path distance of 3700 km for a freshwater particle with a salinity of ~26-28 pss in the proximity of the Amazon mouth at about 52°W to relax to the open ocean surface salinity at ~36 pss after an eastward transport along the NECC  reaching a longitude of ~31°W. At the beginning of the period, the SSS is modulated by the mixing with the saltier waters transported westward by the NBC rings shed at the retroflection. The SSS modulation here is clearly consitent with the ocean color signal, fresher water being  systematically associated with colored waters with high cdom values, typical of the brackish plume waters. The drifter is then advected eastward along the NECC, remixed with 'younger' advected plume waters in august and reached a position slightly north of the Pirata mooring  (8°N,38°W) with an SSS of about 32 pss at the beginning of october. The SSS time series from SMOS at the Pirata mooring and at 1 m depth from the buoy is given here below:
Clearly, this analysis demonstrates that the low SSS seen at the Pirata buoy from end-august to end november is induced by the eastward transport of the freshwater plume along the NECC. From october to November, the drifter then continues its propagation eastward along the NECC associated with a continous mixing with salitier open ocean waters: a quasi-linear growth of SSS is then observed with a concomittant quasi-linear decay of SST and cdom.
In addition, the conservative mixing-law between SSS and cdom along the lagrangian trajectory of the virtual drifters can now be analyzed with a much higher spatio-temporal resolution than in Salisubry et al, 2011.