Mercator Ocean newsletter 21

Mercator-Ocean Quarterly Newsletter #21 – April 2006 –

GIP Mercator Océan

Quarterly Newsletter

Editorial - April 2006

Salt is a key ingredient in the exchanges between the surface ocean and the subsurface/deep ocean as it controls the density of the surface water in tandem with its cousin: the temperature.

By this way, it is involved in the oceanic overturning circulation process. As examples of some contributions to the oceanic and climate variability, we can name the “Salt Barrier Layer” in the Equatorial Pacific area and its implications in the El Nino signal, as well as the “Great Salinity Anomaly” which occurred in the Northern Atlantic Ocean of the later 1960s and its climatic impact (convection strength, sea ice formation …).

credits : Nathalie Garo

Thanks to the development of in situ observations networks (ARGO project for instance), studies of the salinity impact on the oceanic paths is and will be facilitated.

For these reasons, we wanted to entirely dedicate this issue to the oceanic salt parameter.

Firstly, you will find in the following a News page which describes an E-P ECMWF forcing term corrections method used for Mercator-Ocean analyses purposes.

Secondly, Tranchant &al. will detail an OSSE experiment which has been performed with simulated SMOS/Aquarius SSS data and the new Mercator-Ocean assimilation scheme SAM2.

Thirdly, the Kalsal project will be described and some results detailed by Reverdin & al. and, in the next article, preliminary results of a PROVOR floats analyses in the Western Pacific warm pool will be presented by Delcroix & al : Two articles dedicated to the use of in situ observations data sets.

Finally, the last article highlights a propagation process of salt anomalies in the North Atlantic Ocean through simulation analyses (Laurian & al.).

Have a very good read!


GIP Mercator Océan

Contents

NEWS: SURFACE FRESHWATER BALANCE FOR GLOBAL MERCATOR-OCEAN ANALYSIS PURPOSES..........................................................................................................................................................3

OSSE PERFORMED WITH SIMULATED SMOS/AQUARIUS SSS DATA AND THE SAM2 SCHEME .............................................................................................................................................................7

SURFACE SALINITY (KALSAL, COSMOS) ..............................................................................................19

USING PROVOR FLOATS TO ASSESS THE LINK BETWEEN ENSO AND THE SALINITY VARIABILITY IN THE WESTERN PACIFIC WARM POOL..................................................................24

POLEWARD PROPAGATION OF SPICINESS ANOMALIES IN THE NORTH ATLANTIC OCEAN............................................................................................................................................................................30

NOTEBOOK -...................................................................................................................................................38


News : Surface freshwater balance for global Mercator-Ocean analysis purposes

News: Surface freshwater balance for global Mercator-Ocean analysis purposes. By Gilles Garric1 1CERFACS/Mercator-Ocean

In the actual eddy permitting ¼° global operational Mercator-Ocean Ocean forecasting system (i.e. PSY3v1), the salinity is prognosed from surface to the bottom, and relaxed at the surface towards the Levitus surface salinity climatology. This artificial constraint is supposed to damp biases coming from different sources: dynamical processes miss- or un-resolved by the model, errors induced by the assimilation scheme and errors in initial and boundary conditions of the system. Although the large majority of these errors are resolution-dependent and must be handled with system of adequate resolution, large scales errors are encountered and can be studied with “lighter” configuration. This is the case for the boundary conditions and, particularly for systematic biases found in the atmospheric forcing used to drive the system at the surface.

In order to study the surface atmospheric forcing applied to the next version of the global prototype, we performed an experiment with the ORCA2-LIM 2° configuration (the “light” version) with no assimilation, and driven at the surface by the interannual ECMWF (European Centre for Medium-Range Weather Forecasts) analysis dataset over the 1993-2004 period; the ECMWF Integrated Forecast System is the one identified to drive all the Mercator-Ocean operational system.

Figure 1 shows the modelled interannual global mean of both sea surface salinity (SSS) and the sea surface height (SSH). The two very large trends respectively negative for the simulated SSS (-0.07 PSU.year-1) and positive (+280 mm.year-1) for the simulated SSH reveal unrealistic freshening and unrealistic accumulation of the global surface water masses. According to [Huffman et al., 1997], these surface mass and freshening anomalies are undoubtedly issued from the large overestimation found in the ECMWF rainfalls (+1mm.day-1 in average) compared to the GPCPv2 interannual variability (Figure 2). The surface mass budget is consecutively permanently positive and leads to a net freshening and mass accumulation all the experiment long.

In addition to the large overestimation discussed above, the ECMWF rainfalls exhibit well marked trends during this decadal period. These are strongly linked to the changes implemented in the ECMWF forecast model. The impact of the 4DVAR assimilation system which has been implemented in 1998 is clearly depicted on the parameterised precipitations (Figure 2). In other studies as the GPCPv2 dataset and a review made on precipitation measurements (New et al., 2001) no significant trend has been found.

Figure 1

Interannual global mean of the simulated sea surface salinity (SSS) (PSU, solid line) and sea surface height (SSH) (M,

dotted line).

Figure 2

Interannual global mean of ERA40 (dashed line), ECMWF analysis (dotted line) and GPCPv2 (solid line) rainfall datasets

for the 1993-2004 period.



The surface salinity anomaly and rainfall overestimations are essentially found in the tropical bands (Figure 3 and Figure 4). Discussed by (Troccoli and Kallberg, 2004), these excessive amounts of precipitations over tropical oceans are due to a fundamental problem in the variational analysis of humidity over tropical oceans in areas of high density observations, such as satellite radiances. However, medium-and-large-scales rainfall biased features are also observed in others areas (Figure 3): the large underestimation found in Southern mid-latitudes results from a less marked circumpolar trough in the ECMWF model; systematic overestimation (resp. underestimation) are found upstream (resp. downstream) orography (the Andes cordillera, the Antarctic Peninsula, Greenland); underestimation is also found at the end of the Atlantic storm tracks in the Norwegian Sea.

In order to reduce this large-scale anomaly and deliver a more realistic mean SSH for the assimilation scheme and to reduce the surface mass damping, we describe in this paper how we attempted to correct this systematic bias from the ECMWF rainfall dataset.

Figure 3 Mean modelled sea surface salinity damping (10-3 mm.day-1) over the 1993-2004 period. Positive damping (resp. negative)

values represent fresher (saltier) surface waters compared to the climatology.

The method

We adopted the method described in (Troccoli and Kallberg, 2004). Briefly, they make four assumptions : 1) ‘The ECMWF precipitation field should conform to the observed precipitation field’ (the GPCPv2 dataset for our purpose); 2) ‘The water budget, precipitation minus evaporation (calculated by the ORCA-LIM model in our case, i.e. also including Runoff and SSS damping), has to be zero in a global sense for the concerned period’; 3) ‘The evaporation field is treated as error-free’ ; 4) ‘The temporal variability is not affected to first order.’

The minimisation procedure only involves a single coefficient, α, calculated as the solution of the following system:

)1(0

)(

=−+

−−=

ERPPPPP

c

obsc α this can be reduced to the simple equation: )2()(

)(

obsPPERP

−−+

=α

where P is the mean ECMWF precipitation, obsP is the mean GPCPv2 climatological precipitation. CP is the mean

corrected precipitation, E is the mean evaporation and R is the mean river runoff set to 0.31 mm.day-1 (elaborated from

different references not listed here) in our experiments. All these quantities have been evaluated for ocean areas only.

In the remaining part of our experiment, we have adopted a different method than the [Troccoli and Kallberg, 2004] one: the correction is applied on the global ocean and not only on the tropical band; the correction is applied locally and is not latitudinally dependent; we use a monthly GPCPv2 dataset sampling. We also wanted to distinguish the two time series (1998-2004) and (1993-1997) which respectively use and do not use the 4DVAR assimilation scheme in the ECMWF forecast system. Figure 4 shows the values of the α coefficients and the zonal mean of the corrected precipitations for both periods.



Results

We used the evaporation calculated from the first experiment (hereafter EXP1) to evaluate the α coefficient and the ECMWF analysis corrected precipitation. A second experiment (hereafter EXP2) is then performed with this corrected precipitation field

CP . With different mean values of the evaporation flux at different periods, Table 1 shows a nearly closed surface mass budget

during the two time periods. In a general view, the more α is close to 1.0, the more the corrected precipitation is close to observed value and the more the modelled evaporation flux is close to the real value. Although the equilibrated budget during the period 1993-1997 is due to the cancellation of gained and loss of mass (Figure 5), the nearly closed budget during 1998-2004 exhibits a realistic no-trend. Compared to EXP1, the SSS trend, nearly equilibrated (+0.003PSU.year-1) during 1998-2004, has been reduced drastically. Moreover, the SSH trend during 1998-2004 is in the realistic bounds of the estimated thermosteric sea level rise (1.6±0.3 mm.year-1 for1993-2003) [Willis et al., 2004] or the total observed sea-level rise (2.5±0.4 mm.year-1 for1993-2003) [Lombard et al., 2005]. The magnitude of the mass damping is also divided by more than two between EXP1 and EXP2 during the 1998-2004 period. The stabilisation of the damping term after the year 2000 at a value of 10-3*mm.day-1 is particularly encouraging (Figure 5). We can also note that the damping term is no longer concentrated in the tropical band but is spread all over the global ocean with peaks in the mid-(Northern Atlantic) and high-(Arctic Ocean) latitudes (not shown).

This work has allowed us: 1) to reach both nearly equilibrated globally averaged SSS trend and a realistic globally averaged surface mass trend over the last decade and 2) to close the surface mass budget in hindcast modes for the recent ECMWF analysis datasets. This method is planned to be applied to the spin up for the PSY3v2 prototype as long as the GPCPv2 datasets are available. These precipitation datasets, delayed to present time, are not available in real time. However, we used the monthly mean fields which are not designed to be used in forecast mode. A method to use the GPCPv2 datasets low frequency (monthly) variability and to correct these excessive amounts of precipitations in forecast mode is under investigations. Daily GPCPv2 datasets will be also used in the method to check the input of higher variability in the previous results.

Figure 4

Mean zonal precipitation for the ECMWF analysis (dotted line), the GPCPv2 datasets (solid line) and the corrected precipitations (dashed line) over the 1993-1997 (left panel) and the 1998-2004 (right panel) period. The values of the

corresponding α coefficients (see text) is shown.

α = 0.5 α = 0.95



OCEAN MEAN VALUES

DATASET P

(mm.day-1)

E

(mm.day-1)

PmE

(mm.day-1)

mass budget

(mm.day-1)

∂t(SSH)

mm.year-1

∂t(SSS)

PSU.year-1

SSS damping

10-3*mm.day-1

ECMWF(1993-2004)/EXP1 3.78 3.32 0.45 0.76 283 -0.07 -2.36

ECMWF(1993-2004)/EXP2 2.99 3.32 -0.3 5.10-4 -3.9 +0.0067 -1.55

ECMWF(1998-2004)/EXP2 2.93 3.23 -0.29 0.01 5.85 +0.003 -0.97

Table 1 Mean values for precipitation (P), Evaporation (E), Precipitation minus Evaporation (PmE), total mass budget (P+R-E-SSS damping), SSS trend (∂t(SSS)) in PSU.year-1, SSH trend (∂t(SSH)) in mm.year-1 and surface mass damping towards Levitus

seasonal salinity in 10-3*mm.day-1. Values issued from EXP1 and EXP2.

Figure 5 Interannual global mean of the modelled SSH (solid line), the (P-E) term (dotted line) and the surface mass damping towards Levitus

seasonal salinity in 10-3*mm.day-1 (dashed line) for the EXP2 experiment.

References

Huffman, G.J., P.A. Arkin, A. Chang, R. Ferraro, A. Gruber, J.E. Janowiak, R.J. Joyce, A. McNab, B. Rudolf, U. Schneider, and P. Xie, (1997), The Global Precipitation Climatology Project (GPCP) Combined Precipitation Data Set, Bull. Amer. Met. Soc., 78, 5-20.

Lombard, A., A. Cazenave, P.Y. Le-Traon, and M. Ishii, (2005), Contribution of thermal expansion to present-day sea-level change revisited, Global Planet. Change, 47, 1-16.

New, M., M. Todd, M. Hulme, and P. Jones, (2001), Precipitation measurements and trends in the twentieth century, Int. J. Climatol., 21, 1899-1922.

Troccoli, A., and P. Kallberg, Precipitation correction in the ERA-40 reanalysis, (2004), ERA-40 Project Report Seris.

Willis, J.K., D. Roemmich, and B. Cornuelle, (2004), Interannual variability in upper-ocean heat content, temperature and thermosteruc expansion on global scales, J. Geophys. Res. - Oceans, 109-C12036, doi:10.1029/2003JC002260.

.


OSSE performed with simulated SMOS/Aquarius SSS data and the SAM2 scheme

OSSE performed with simulated SMOS/Aquarius SSS data and the SAM2 scheme By Benoît Tranchant1, Lionel Renault2, Charles-Emmanuel Testut1, Nicolas Ferry1 and Pierre Brasseur3

1 Mercator-Ocean-océan, 8/10 rue Hermes, 31520 Ramonville-St-Agne, France 2LEGOS, 14 Av. Edouard Belin, 31400 Toulouse, France. 3LEGI, B.P.53, 38041 Grenoble Cedex 9, France

Introduction

Ocean salinity variability plays an active role in the global ocean circulation and is a key indicator of the underlying processes that link the ocean circulation and hydrologic cycle (Lagerloef 2002, see diagram below). Moreover, a better understanding of the interactions between hydrologic cycle, ocean circulation and global heat transport variations is essential for climate studies.

At the end of 2007, a new satellite called SMOS (Soil Moisture Salinity Ocean) will be launched by ESA (European Spatial Agency) to measure ocean surface salinity (SSS for Sea Surface Salinity) as well as moisture of continental surface. A US/Argentinean AQUARIUS mission measuring sea surface salinity is also in preparation. These two missions should provide measurements of SSS with a global ocean coverage to the scientific community and should counterbalance the lack of sea surface salinity data. So far, sea surface salinity data are very scarse. Most of them come from oceanographic cruise or from ships of opportunity, anchored buoys or TSG.

Although satellite measurements have a lower precision than in-situ ones, the satellite data space-time coverage will make it possible to study large scale processes and their evolution over several years. Several levels of products will be available corresponding to various levels of data treatment. The expected accuracy is approximately 0.1PSU, with a space resolution of 200km x 200km every 10 days, in accordance with GODAE recommendations.

The need for accurate SSS data has been pointed out in many studies to avoid spurious salinity drifts in ocean model. This drift is mainly due to important errors in the freshwater fluxes, but also to inaccurate initial conditions and errors in the model physics. To compensate such errors, the future satellite SSS observations can be used in two ways: (i) to correct directly the systematic biases of the freshwater fluxes used in the model forcing, (ii) and in data assimilation systems, see (Durand et al., 2002) and the Mercator Newsletter #17.

In this study, we performed Observing System Simulation Experiments (OSSEs) with simulated SMOS/Aquarius SSS data (level 2 and level 3 products) and the new Mercator-ocean multivariate assimilation system named SAM2v1. The first objective is to assess which level of SSS product is the most efficient for the assimilation system SAM2v1/MNATL. The second objective is twofold: (i) to assess the benefit of using various SSS satellite observation systems (ii) and to estimate the observation error threshold from which the associated SSS satellite data has a significant influence on the MERCATOR OCEAN assimilation system.

Model and Assimilation Scheme Description

Ocean model

The ocean model component chosen for this study is the MNATL configuration (eddy-permitting) (see Newsletter # 13).

The model domain covers the North Atlantic basin from 20°S to 70°N and from 98.5°W to 20°E, with a horizontal resolution of 1/3° x 1/3° cos(latitude). The vertical discretization has 42 geopotential levels, with a grid spacing that increases from 12 m at

http://www.mercator-ocean.fr/documents/lettre/lettre_17_en.pdf

http://www.mercator-ocean.fr/documents/lettre/lettre_13_art1_en.pdf



the surface to 200 m below 1500 m depth. The bathymetry is derived from Smith and Sandwell (1997). The model solution is relaxed toward climatology within buffer zones defined off Portugal, in the Norwegian Sea and along the Southern boundary to simulate respectively the supply of Mediterranean Water and the exchange with the Artic and South Atlantic basins.

Data assimilation scheme

The impact studies presented in this report have been performed using the next Mercator operational forecasting system based on the second generation SAM2v1 assimilation scheme. This scheme is being integrated in a pre-operational phase for the main Mercator operational prototypes and we present hereafter its main features.

The SAM2v1 tool is a multivariate assimilation algorithm consisting of a Singular Extended Evolutive Kalman (SEEK) filter analysis method (Brasseur et al., 2005). The SEEK filter is a reduced-order Kalman filter introduced by Pham et al. (1998) in the context of mesoscale ocean models.

The error statistics of this method is represented in a sub-space spanned by a small number of dominant 3D error directions. The formulation of the assimilation algorithm relies on a low-rank error covariance matrix, which makes the calculations tractable even with state vectors of very large dimension. The extrapolation of the data from observed to non-observed variables is performed along the directions represented by these error modes which connect all dynamical variables and grid points of the numerical domain. Unlike the original SEEK filter, SAM2v1 does not evolve the error statistics according to the model dynamics. This would require prohibitive costs given the size of the operational system. However, some form of evolutivity of the background error is taken into account by considering different error sub-spaces for the four seasons. Then, a method involving a computation of empirical 3D modes obtained from a hindcast simulation performed during the 1993-1999 years has been applied for each season; this approach leads to corrections of the model trajectory that are 3D multivariate and seasonally consistent.

SAM2v1 uses a diagonal modelling of the observation error matrix meaning that correlated data could be misrepresented. In order to prevent the data from exerting a spurious influence at remote distances through large-scale signatures in the 3D modes, a simplification of the analysis scheme has been adopted by enforcing to zero the error covariances between distant variables which are believed to be uncorrelated in the real ocean (e.g. Penduff et al. (2002), Testut et al. (2003), Birol et al. [2004]). This simplification is implemented in SAM2v1 by assuming that distant observations have negligible influence on the analysis.

In practice, the SAM2v1 system consists in a sequential method using a 7-day assimilation cycle. The innovation is calculated during the model integration or the forecast step using the First Guess at Appropriate Time (FGAT) approximation corresponding to a misfit between the model and the observation computed at appropriate time.

The Operational Data Set and the Reference Simulation All the assimilation experiments showed hereafter use the same observation data set which is also used in the current MERCATOR OCEAN multi-data and multivariate operational systems:

Altimeter data: The along track sea level anomalies from JASON-1, ERS-2/ENVISAT and GFO satellites are assimilated with the FGAT method. The observation errors are diagonals and equal to 2 cm for JASON and ERS-2/ENVISAT, and equal to 3,5 cm for GFO.

MSSH: A pseudo-data of Mean Sea Surface Height (MSSH) (coming from a previous work based both on a MERCATOR OCEAN re-analysis and the work of Rio and Hernandez (2003)) is used as a reference level for the Sea Surface Height (SSH). An estimate of the MSSH error has been added to the overall altimetric observation error.

In-situ data: Vertical T/S profiles (down to 2000 m depth in some cases), measured by ARGO floats, XBT/CTDs, moorings or buoys, and provided by the CORIOLIS operational data centre (Brest) are assimilated in SAM2v1 with the FGAT method. Observation errors of in-situ data are depth depending and have been inferred from previous hindcasts.

Climatological profiles: Monthly climatology for the temperature and salinity profiles on a 2°x2° grid are assimilated from ~2000m depth to bottom are assimilated at the date of analysis.

SST: The sea surface temperature (SST) is an ECMWF product (RTG_SST) which comes itself from daily analyses received from NCEP, Washington in a 0.5 x 0.5° grid. This SST is re-mapped on a 1°x1° grid and is only assimilated at the analysis date. The associated observation error is spatially constant and fixed to 1.2°C.

Using these data sets, a reference simulation (REF) has been conducted. Then, a hindcast simulation is performed using the SAM2v1/MNATL system previously described and starting in January 2003 from temperature and salinity climatology (Reynaud et al., 1998), and integrated from 2003 to 2004 using a 7-day assimilation cycle.



Simulated Data and protocol for OSSEs

OSSEs have been performed with simulated SSS SMOS and Aquarius data generated by MERCATOR OCEAN PSY2v1 which is univariate and only assimilate altimeter data. Note that the SAM2V1 scheme previously described is multivariate and assimilate multi-data.

Most of the simulated SSS products have been processed by CLS (see Boone et al., 2005). It concerns the Level 2 and 3 products for SMOS and the Level 2 products for AQUARIUS. In addition to these sets, we have processed two new sets of SMOS Level 2 products by modifying the initial specification of the observation error.

Table 1, middle column sums up the different SSS products described below.

Simulated SMOS data

Level 2 products: regular data

Generally, a Level 2 (or L2) product is equivalent to a product at the captor pixel resolution (or similar).

SMOS L2 product are generated from daily SSS fields over 2003 from Mercator PSY2v1 (9°N-70°N) re-sampled at a 1/3 degree (roughly equivalent to the 40 km2 SMOS mean pixel area) fields. It is then sub-sampled on the pixel daily location corresponding to the effective satellite tracks. A Gaussian noise associated to the prior specification of the future SMOS satellite is then added to this field (see Figure 1). This error varies in space and time. It defines the initial specifications of the observation errors. An example of the simulated daily SMOS SSS field for July 6, 2003 is given on Figure 1.

Figure 1 SMOS pixel location on July 6th 2003 (Level 2): salinity in PSU (left) and the associated RMS errors (right).

Level 2 products: modified data

In order to study the sensitivity to the observation errors, we have generated two new modified data sets from the initial Level 2 SSS data generated by Boone et al, (2005).

Data locations along simulated SMOS satellite tracks have been preserved, only values and associated errors have changed. The data values have been regenerated by multiplying the initial Gaussian noise and the associated RMS of observation errors (eo) by 2 and by 0.5.

Consequently, two new SSS Level 2 products have been built over 2003 and will be tested in the OSSE context. It should help us to determine a threshold of specification for the observation error.

Level 3 products

Generally, a Level 3 (or L3) product is equivalent to a gridded or mapped product, i.e. represented on a regular grid, and here at GODAE scales (200km, 10 days).



Figure 2 SMOS SSS map on July 6th 2003 (Level 3): salinity in PSU (left) and the associated RMS errors (right).

The SMOS Level 3 products has been built from the SMOS Level 2 Products by Boone et al. (2005) using an objective analysis. The SMOS Level 3 data are then available on 10-days maps using a spatial resolution of 200x200km (see Figure 2). An associated observation error varying in space and in time is also produced by the objective analysis. Note that this Level 3 error (see Figure 2) is smaller than the error of the SMOS Level 2 products. Consequently, the SMOS Level 3 data are more precise than the SMOS level 2 products. In order to have a time coherency with the 7-days assimilation cycle, the original 10-days SMOS Level 3 Products have been re-interpolated on a weekly frequency.

Simulated Aquarius data: L2 products

The same protocol as SMOS L2 data has been used for Aquarius L2 data. A noise estimated from Boone et al., (2005) is added to the daily Mercator PSY2v1 SSS over 2003. The spatial resolution is about 100kmx100km along simulated AQUARIUS satellite tracks (Figure 3)

Figure 3 Aquarius pixel location on July 6th 2003 (Level 2): Salinity in PSU (left) and the associated RMS errors (right).

OSSEs protocol

Experiments

We performed several experiments by assimilating SSS data over the year 2003 starting from the Reynaud climatology, and using a 7-day assimilation cycle, as used in the REF (or CONTROL RUN) simulation.



SSS data product characteristics Name of the experiment over year 2003

Level (spatial and time resolution) Estimated observation error range (RMS in PSU)

SMOS L3 Level 3 SMOS (map of 200kmx200km, 10 days) 0,02 - 0,5

SMOS L2 Level 2 SMOS (40kmx40km along tracks, 1 day) 0,2 - 2,5

SMOS L2_2 Level 2 SMOS (40kmx40km along tracks, 1 day) 0,4 - 5

SMOS L2_0.5 Level 2 SMOS (40kmx40km along tracks, 1 day) 0,1 - 1,25

Aquarius L2 Level 2 Aquarius (100kmx100km along tracks, 1 day) 0,1 - 1,5

SMOS L2+ Aquarius L2 Level 2 SMOS + Level 2 Aquarius (40kmx40km

+100kmx100km along tracks, 1 day) 0,2 - 2,5 and 0,1 - 1,5

Table 1 Experiments: names and main characteristics.

Diagnostics: Statistical results

In this study, we focus on the overall1 domain where mean and variance of misfit between experiments and “truth” have been calculated. The “truth” is the original2 SSS located on the SMOS L2 products. We only calculated statistics relative to the “truth” and not to independent data. There are two reasons for that: (i) the “truth” is too far from a realistic SSS field, (ii) and the “truth has not been generated from the same model used for analysis/forecast runs like in a twin experiment. It is thus difficult to compare with independent data or to compare with others variables of interest at various depths.

Results

Results are coming from the comparisons of various assimilation experiments using different data types (SMOS L3, SMOS L2, Aquarius L2) and combinations of data type (SMOS L2 + Aquarius L2). This study can be split into three different sensitivity studies: A sensitivity to level product, a sensitivity to observations error and a sensitivity to observing systems.

Sensitivity to level product

The annual mean misfit and the variance of misfit between three experiments (REF, SMOS L3 and SMOS L2) and the “truth” averaged over 2003 is showed in Figure 4. At first, we can see that the misfit between REF and the “truth” is not spatially homogeneous. In particular, important patterns appear in three types of regions: (1) turbulent regions characterised by meso-scale activity (Gulf Stream); (2) coastal regions where river runoff play a significant role (Mississippi delta, Saint-Laurent delta, Labrador Sea, …); (3) and more specific regions (North Sea) where the observation errors for all data type have been voluntary increased in order to avoid spurious effects of the assimilation.

The SSS constraint from SMOS L3 has a positive impact. Indeed, in comparison to the REF experiment, this simulation has both slightly reduced the bias and the variance of the misfit. It means that this simulation is much closer to the “truth”. Nevertheless, this reduction is very weak, which indicates that large scales constraint coming from SMOS L3 SSS is not pertinent enough to have a major impact. Note also that important misfit values are not necessarily inconsistent with the observation error magnitude associated to the SMOS Level 3 Products (see Figure 2). Indeed, these observation errors used in the analysis step are directly related to the time and spatial scales of the Level 3 Products. Consequently, we have to consider (in these diagnostics and not in the assimilation) an error of representativity due to the difference of resolution in space (between roughly 40x40km and 200x200km) and in time (between 1 day and 10 days).

1 The overall domain is referred as the North Atlantic PSY2v1 domain, i.e., from 9°N to 70°N corresponding to geographic locations of simulated SSS data.

2 The original SSS comes from the North Atlantic and Mediterranean high resolution MERCATOR-OCEAN prototype named PSY2v1 re-sampled at a 1/3 degree (Boone et al., 2005).



Mean Variance

REF - ‘truth’

SMOS L3 -‘truth’

SMOS L2 - ‘truth’

Figure 4 Mean (left) and variance (right) of misfit between three different estimates: Reference (top), SMOS L3 (middle), SMOS L2

(bottom) and “truth” averaged over one year (2003).

In comparison to SMOS L3, the SMOS L2 simulation presents some significant and very important decreasing of the misfit to the “truth” in regions of interest. Thus, the effects are strong in regions of river runoff and in the Gulf Stream current. However, as expected, the important pattern in the North Sea is not sufficiently reduced due to the parameterization of the observation error which tends to make the data assimilation vanish in this region.



Sensitivity to observation errors

SMOS L2_2 – ‘truth’

SMOS L2 – ‘truth’

SMOS L2_05 –‘truth’

Figure 5 Mean (left) and variance (right) of misfit between three different estimates: SMOS L2_2 (top), SMOS L2 (middle), SMOS L2_05

(bottom) and “truth” averaged over one year (2003).



Figure 6 Spatial average of the mean (left) and variance (right) of misfit between three different estimates control run or REF (red dashed

line), SMOS L2 (blue solid line), SMOS L2_0.5 (green solid line), SMOS L2_2 (purple solid line) and “truth” every ten days during one year (2003) for the overall domain.

In this sensitivity study, various observation errors levels are tested and inter-compared. Data locations along simulated SMOS satellite tracks have been preserved, only values and associated errors have changed. The data values have been regenerated by multiplying the initial Gaussian noise and the associated RMS of observation errors by 2 and/or by 0.5.

We want to identify the observation error threshold corresponding to a significant impact in a system close to the current multivariate and multi-data Mercator data assimilation system. This method provides a first guess of the error threshold. It is nevertheless important to keep in mind that observation errors associated to each dataset are related by their ratio. This limitation is very important in a system assimilating combined data. Indeed, the threshold found in this study is only valid for these sets of data used with these specific sets of observation errors. Some future improvements both in term of additional data sets and/or in term of better estimation of observation error could modify this threshold. In addition, to date, only random errors have been included in the simulated observations. There is a recognized need to include systematic large-scale errors. This weakness of simulated error may also mask or reduce the impact of simulated observations from other observing systems.

The annual mean misfit and the variance of misfit between three experiments (SMOS L2, SMOS L2_2 and SMOS L2_0.5) and the “truth” averaged over 2003 are showed in Figure 5. We can see that misfits (mean and variance) have been reduced in all regions when observation errors decreased. For SMOS L2_0.5 and SMOS L2, important biases have been removed. On the one hand, even if this impact is weaker for SMOS L2_2, Figure 6 shows that the time evolution of mean misfits is quite equivalent to the others. On the other hand, the time evolution of the variance for SMOS L2_2 is similar to REF during the first three months. It shows that it takes three months until the SSS assimilation exerts a significant influence (after three months, the time evolution of variance is equivalent to SMOS L2_0.5 and SMOS L2).

As in the previous section, small scales are certainly more constrained for SMOS L2 and SMOS L2_0.5 which use lower observation errors. The level of SSS observation error in the SMOS L2_2 appears too weak in comparison to the other data constraints. The concept of threshold can be introduced here. Indeed, it seems necessary to specify a minimum level of observation errors (as specified into the SMOS L2 experiment) to obtain a significant impact. It allows us to reduce the misfit from about 0.5 to 0.3 PSU (RMS).

Sensitivity to observing systems

We consider two SSS assimilation experiments using satellite data from two simulated SMOS and Aquarius satellites. In addition, we have also performed an assimilation experiment combining these two observing systems. The objectives are to appreciate both the skill and the potential complementarity of every one of them. The time evolution of mean and variance of misfits are showed in Figure 7.



Figure 7 Spatial average of the mean (left) and variance (right) of misfit between three different estimates control run or REF (red dashed

line), SMOS L2 (blue solid line), Aquarius L2 (green solid line), SMOS L2+Aquarius L2 (purple solid line) and “truth” every ten days during one year 2003 for the overall domain.

Aquarius L2

We can see that the Aquarius L2 simulation is relatively better than the REF experiment which means that the SSS constraint coming from the simulated Aquarius observation system has a positive impact in our system. However, this impact seems to be weak in comparison to the SMOS L2 simulation. Several explanations can be proposed for the relative inefficiency of the Aquarius L2 simulation in comparison to the SMOS L2 simulation. Each of them contributes to explain results obtained from the Aquarius L2 simulation.

- The daily data coverage is very different between these two products. The SMOS L2 space and time coverage is approximately twice as large as the one from Aquarius L2. The SSS SMOS L2 Products are thus associated to a stronger constraint in the assimilation step than the AQUARIUS L2 Products. Moreover, it is interesting to note that the decorrelation scales (in days) in a 1/3° re-analysis over 11 years (Greiner et al., 2004) (Figure 8) is generally less than 4 days. It is representative of the ocean-atmosphere exchange processes that take place in the North Atlantic. But, since each ocean grid point is observed by Aquarius measurements every 7 days and by SMOS measurements every 3 days, only SMOS L2 data are able to constraint an eddy-permitting model (spatial resolution less than or equal to 1/3°).

- The error associated to Aquarius L2 Products is lesser than that of SMOS L2 Products. Nevertheless, re-mapped onto the Aquarius grid (100x100km), and using a non-correlated errors assumption, the equivalent SMOS L2 observation error becomes smaller.

- At last, the spatial resolution of Aquarius is also a potential explanation of the difference. Indeed, the Aquarius Level 2 data (100kmx100 km) are no able to constraint smaller scales.

However, it is relatively difficult to really distinguish the impact of each of these reasons in the difference between the SMOS L2 and Aquarius L2 simulations.



Figure 8 Temporal decorrelation scales from a re-analysis at 1/3°(10°N-70°N) over 11 years calculated from annual time series. Limit of

time correlation is fixed at 0.4.

SMOS L2 + Aquarius L2

The impact of the assimilation experiment combining the two simulated SMOS and Aquarius L2 Products is weak compared to the SMOS L2 simulation, and it is only marked at the end of the simulation. The complementarity of the two product coverage is perceptible by slightly improving statistical results.

Note also that, as it has been written in the previous sub-section, the impact of one dataset is related to the observation error ratio between each dataset.

Summary and Discussion

Summary

Some regions can be distinguished by their skill in reducing the misfit when simulated SSS products (L2 and L3) are assimilated (e.g., Runoff regions). Misfits are generally highest in: (i) High latitudes where observation errors are highest, (ii) Gulf Stream where variability is high, (iii) Near the coast where the assimilation system presents some limitations to assimilate observations with too important errors.

We can summarize the OSSEs by Table 2, Table 3. RMS of misfit (PSU) between all OSSE experiments and “truth” averaged overall the domain are reported in these two Tables.

Experiments

REFERENCE

SMOS L3

SMOS L2

AQUARIUS L2

SMOS L2

+

AQUARIUS L2

RMS (PSU) of the misfit between experiments and “truth” overall the domain (year 2003)

0,4859

0,3945

0,3104

0,4353

0,3077

Table 2 RMS of misfit between experiments and “truth” overall the domain (year 2003). The best experiment is underlined and the misfit

between the reference and “truth” is also mentioned.



Experiments REFERENCE SMOS L2_2 SMOS L2 SMOS L2_0.5

RMS (PSU) of the misfit between experiments and “truth” overall the domain (year 2003)

0,4859

0,4114

0,3104

0,2847

Table 3 RMS of misfit between SMOS L2 experiments for different observation errors and “truth” overall the domain (year 2003). The

best experiment is underlined and the misfit between the reference and “truth” is also mentioned.

The assimilation of SMOS L3 Product has a weaker impact than the SMOS L2 Product. It is certainly due to the difference of scales (space and time scales). The misfit resulting from the SMOS L3 experiment is higher than that found from the SMOS L2 experiment since the misfit is calculated with respect to “the truth” which is at a different scale. Nonetheless the misfit is likely due to the inability of the Level 3 product to resolve the model smaller scales. The variance of misfit has been divided by 2 when SSS SMOS L2 products (data and associated errors) prescribed by Boone et al. (2005) are assimilated, thus reducing the misfit from around 0.5 to 0.3 PSU rms.

The initial observation errors associated with the SMOS L2 products given by Boone et al. (2005) are satisfactory. Consequently, we can consider that the specification of the initial observation errors into the SMOS L2 products is sufficiently well prescribed for the overall domain. We can consider this level of specification as a threshold. It is the kind of information expected by ESA.

The Aquarius L2 misfits are not as satisfying as those found with SMOS L2/L3 experiments. It is probably due: (i) to the spatial coverage which is lesser than the SMOS L2 Products, (ii) to the magnitude of the observation error, (iii) and/or to the difference of resolution between model (roughly 40x40km) and data (100x100km). As a direct consequence, the combination of SMOS and Aquarius L2 product has only a weak impact on results here.

Discussion

In the context of the SMOS and Aquarius missions, final products will be strongly constrained by the complexity of SSS retrieval. An important question is then: “what is the best strategy to optimally use the future SMOS and Aquarius data in the context of ocean prediction systems, from the perspective of monitoring the mesoscale ocean circulation?”

For this reason, OSSEs performed with the MERCATOR-OCEAN assimilation system SAM2v1 tried to estimate the best level and accuracy of SSS product which will have a sufficient positive impact on the MERCATOR-OCEAN forecasting system. This OSSE study has been performed with two different satellites (SMOS and Aquarius), two different Levels of product (L2 and L3) and different observation errors. Our results highlight several conclusions:

- The assimilation of level-2 products is a better approach than the assimilation of level-3 products, at least in the context of high-resolution models (1/3° and/or higher resolution) of the ocean circulation. In the case of the assimilation of Level 3 products, we need “daily Level 3 products” for our assimilation system. It would be better to assimilate SSS products with spatial resolution close to the ocean model resolution. Nevertheless, a non negligible part of the results difference could also come from various deficiencies in the way of producing SMOS L3 products (estimation of the error in the objective analysis) and/or in the way of taking into account the SMOS L3 Products in the assimilation step (observation operator not sufficiently appropriated for largest scales).

- The use of the synthetic SMOS L2 product gives satisfactory improvement in the model, since it provides a measurable impact of the quality of ocean estimates from operational systems. The observation error variance as specified by Boone et al., (2005) is a minimum requirement to extract the best possible information from SSS measurements (in the context of the MERCATOR OCEAN forecasting system available today).

- The impact of the Aquarius L2 Products is weak compared to the SMOS L2 Products: it is quite equivalent to the SMOS L3 Products. The combination of the two L2 Products had thus a small effect on final results. Further studies are necessary to better understand this feature of the Aquarius L2 Products. In particular, it is interesting to distinguish the main reasons of this weakness between several explanations: (i) the spatial coverage which is less than the SMOS L2 Products (ii) the underestimation of the specified SMOS L2 observation errors, (iii) the overestimation of the Aquarius L2 observation errors, (iv) the magnitude of the observation error, (v) and/or the difference of resolution between model (roughly 40x40km) and data (100x100km). Furthermore, a combined multi-data (SMOS/Aquarius) Product on the same time and space scales as those of the initial SMOS L2 Products could be a better compromise for our oceanic solution.



Conclusions have to be counterbalanced by the fact that operational and realistic data (sea level anomaly, SST and in-situ) have been assimilated into the assimilation system. Indeed, the assimilated simulated SSS data comes from SSS field relatively far from the other assimilated data. It is possible, thus, that this data incoherency leads to spurious effects. Generally, it is difficult to troubleshoot and to discriminate effects in favour of one parameter due to interactions between the parameters (observation errors, observation operator, error covariances…). For example, due to the fact that the assimilation system does not correct any fluxes, in particular the E-P (Evaporation – Precipitation) fluxes, it leads to under-estimate information coming from SSS.

The next steps would be:

- To introduce the multi-mission concept, i.e., assimilating SMOS/Aquarius combined products on the same time and space scales as the SMOS L2 Products.

- To extend the SSS assimilation in a global model (low and high resolution) in order to apprehend important physical processes in Pacific region (see article of Delcroix and Maes in this newsletter).

- The main expected improvements in the MERCATOR OCEAN assimilation scheme are:

- To control air-sea fluxes in the assimilation scheme which tends to reduce the background errors and then enhance the interest to have more precise observations

- To introduce an adaptative scheme of the error variance

- To compute more relevant background error covariances

References

Birol, F, WP3100: Review of the use of SSS products in data assimilation systems, septembre 2004, ESA study 18176/04/NL/CB, MOO-NT-431-289-MER

Boone, C., Larnicol, G. and Obligis, E., WP2000: Characteristics of SMOS/AQUARIUS level-3 product, July 2005, CLS-DOS-NT-05-116.

Brasseur, P., Bahurel, P., Bertino, L., Birol, F., Brankart, J.M., Ferry, N., LOSA, S., REMY, E., Schröter, J., Skachko, S., Testut, C.E., Tranchant, B., Van Leeuwen, P.J., and Verron, J.,(2005), Data assimilation in operational ocean forecasting systems: The MERCATOR and MERSEA developments, submitted to Q. J. R. Met. Soc., (June 2005).

Durand, F., L., Gourdeau, T., Delcroix and J. Verron, Assimilation of Sea surface Salinity in a Tropical OGCM :a Twin Experiment Approach, Journal of Geophysical Research, 107 (C12),8010,doi:10.1029/2001JC000849,2002.

Greiner, E., M. Benkiran, E. Blayo, G. Dibarboure, 2004: MERA-11 Scientific Plan, 1992-2002 PSY1v2 reanalysis, MERCATOR OCEAN, Toulouse, France.

Lagerloef, G.S.E., 2002: Introduction to the special section: The role of surface salinity on upper ocean dynamics, air-sea interaction and climate., J. Geophys. Res., Vol. 107, N0.C12, 8000,doi:10,1029/2002JC001669

Penduff, Th., Brasseur, P., Testut, C.-E., Barnier, B. and Verron, J., 2002: Assimilation of sea-surface temperature and altimetric data in the South Atlantic Ocean : impact on basin-scale properties. J. Mar. Res., 60, 805-833.

Pham,D.T., Verron, J., Roubaud, M.C., 1998. Singular evolutive extended Kalman Filter with EOF initialization for data assimilation in oceanography. J. Mar. Syst. 16 (3-4), 323-340.

Reynaud, T., Legrand, P., Mercier, H., Barnier, B., (1998) A new analysis of hydrographic data in the Atlantic and its application to an inverse modelling study, International WOCE Newsletter, 32.

Rio, M.-H., and F. Hernandez, A Mean Dynamic Topography computed over the world ocean from altimetry, in-situ measurements and a geoid model, in Journal of Geophysical Research, 2003b

Smith, W.H.F., Sandwell, D.T., 1997. Global sea floor topography from satellite altimetry and ship depth soundings. Science 277, 1956-1962

Testut, C.E., Brasseur P., Brankart J.M. and Verron J., 2003 : Assimilation of sea surface temperature and altimetric observations during 1992-1993 into an eddy-permitting primitive equation model of the North Atlantic Ocean, J. Mar. Syst., 40-41 (april 2003), pp 291-316.


Surface salinity (KALSAL, COSMOS)

Surface salinity (KALSAL, COSMOS) By Gilles Reverdin1, Elodie Kestenare2, Jacqueline Boutin1,Nadine Chouaib1, Fabienne Gaillard3, Delphine Mathias4 1LOCEAN/IPSL, CNRS/UPMC Paris, 2 LEGOS Toulouse, 3 LPO Brest, 4 Coriolis Brest

Surface salinity

Sea surface salinity (SSS) together with sea surface temperature (SST) controls the density of the surface water and therefore is a key ingredient in the exchanges between the surface ocean and the ocean interior. This contribution to near-surface stratification, to ventilation and therefore to the overturning circulation of the oceans has been motivating a large part of the efforts to monitor and understand its variability (cf http://www.legos.obs-mip.fr/en/observations/sss/). Sea surface salinity is related to the concentration of dissolved ions in sea water, and thus also responds to the freshwater budget at the ocean surface. For these various reasons, SSS is one of the fundamental variables for which global sustained observations are needed in CLIVAR, the international program on CLImate VARiability and predictability, and GOOS1, the Global Ocean Observing System sponsored by the IOC2, WMO3, UNEP4, and ICSU5. A poor control on surface salinity in Ocean General circulation model (OGCM) simulations can also degrade the water masses produced and the general circulation of the model. In the following, sea surface will refer to a layer between 10 cm and a few meters of the sea surface, which is the layer that was sampled in the past.

SSS fields 1977-2002

SSS fields in the Atlantic Ocean present an alternance of high salinity regions in the evaporation-dominated subtropical regions and low salinity regions in the subpolar region or where rainfall or fresh water input from rivers dominate in the tropics. Transition between regions is often rather sharp with frontal structures (unresolved in the climatological average of Figure 1), and in the tropics SSS fields present often large seasonal contrasts. Our work is dedicated in improving the low resolution SSS fields in the Atlantic Ocean.

Figure 1 Upper panels: average SSS for March and September in 1977-2002; lower panels: deviation of the 1977-2002 average fields in

March and September with respect to the guess fields we used based on earlier climatologies.

1 GOOS: Global Ocean Observing System (http://www.ioc-goos.org/) 2 IOC: Intergovernmental Oceanographic Commission (http://ioc.unesco.org/iocweb/index.php) 3 WMO: World Meteorological Organization (http://www.wmo.ch/index-en.html) 4 UNEP: United Nations Environment Programme (http://www.unep.org/) 5 ICSU: International Council for Science (http://www.icsu.org/index.php)



Efforts to monitor SSS have been ongoing for 110 years in the Atlantic Ocean, but the quality of the observations, their spatial distributions have hindered efforts to map it, except for selected periods of time or along well-sampled tracks (Reverdin et al., 2002; Delcroix et al., 2005). Since the late 1960s in the Pacific Ocean, and since 1977 in the Atlantic south of 50°N, the situation is slightly better with a large number of SSS data being collected, in particular from merchant vessels (ORSTOM/IRD being a key player in this monitoring). We have first attempted to map monthly these observations with an objective mapping scheme. We are helped in that the variability seems often to have a fairly high persistence, so that one can take into account data at previous or later times in order to improve the analysis (Reverdin et al., 2006). Nonetheless, the fields created still portray large areas where data density is insufficient and the estimates are not reliable. Nonetheless, this helps identify some systematic deviations in 1977-2002 from the climatology, as illustrated in the lower panels of Figure 1, in particular slightly fresher waters from 30°N to 10°N, and more saline waters further south. This mapping also contributed to identify fairly consistent interannual patterns of variability at a resolution not achieved up-to-now (figure 2) and can be used to test how SSS relates to various key variables. We identified SSS signals related to ENSO and NAO. For ENSO, this is particularly well pronounced in the western subtropical North Atlantic (low SSS there following an ENSO) as well as in the western equatorial Atlantic (high SSS during ENSO). As mentioned earlier, we expect a relation between SSS and components of the surface freshwater budget. A statistic test of that relationship showed a local lagged relation of SSS with freshwater fluxes (E-P) (for Evaporation-Precipitation), as well as to the Ekman advection term. In the northern subtropics and at mid-latitudes, this relation is more pronounced with winter season freshwater variability than in other seasons. A note of caution is however needed, as we found the freshwater fluxes to be rather uncertain (the P used is mostly based on GPCP6, whereas E is from ERA40 analyses).

Figure 2 Interannual rms variability in the monthly SSS analyses averaged for 4 seasons (period 1977-2002, but only individual months

with relative error less than 0.8 are considered).

This suggests that part of the variability can be reproduced by the use of simple advective/diffusive models with prescribed external forcing. One interest of this approach is to provide a guess on what the SSS anomalies would be to improve on the persistence hypothesis of our earlier objective mapping. The model can also be thought to be used within a Kalman-assimilation approach framework, and the structure of the fields it produces, used to analyse the patterns of variability (or extract a reduced set of functions able to reproduce a large share of the variability). This seasonally-dependent model can be written as:

SSSrEkmanPEhSSSdtdSSS *)(*// −−−=

6 GPCP: Global Precipitation Climatology Project (http://precip.gsfc.nasa.gov/)



where d/dt represents evolution along near surface geostrophic currents (“quasi-Lagrangian”), E-P-Ekman are forcing terms, h is an equivalent depth (which we will refer to as mixing depth), r is a damping time, and an average seasonal cycle has been removed to all the terms. The model parameters are fitted to the data, and because of this formulation, we expect the model performances to be best for the fast (seasonal time scale) variations of SSS. We find a strong seasonal and spatial dependence in h and r. Both n areas insufficiently sampled (southern hemisphere) and near the equator where other processes are as important (if not more), the results do not make much sense (h very different from mixed layer depth estimates, for example). In the northern hemisphere we find an equivalent depth h larger in February and shallowest in August (Figure 3), as expected from mixed layer depth estimates. More surprisingly, we find a large seasonal modulation of damping rate r, which is largest in November (shorter damping time scale) and smallest in May in the northern hemisphere. This might be coherent with the strong vertical mixing through the autumn which should alter earlier SSS anomalies.

Figure 3 Equivalent depth h estimated by fitting the model to the SSS analysed data (retaining only the gridded data which relative error

is less than 0.8). The results are shown for two contrasted seasons (February and November).

The model is then integrated in time on a low resolution (2*5° lat*long) grid in order to produce fields that can be compared to the observed variability. These fields have usually lower variability levels than the observed ones. This is particularly pronounced in the north-western tropical Atlantic (which is expected, as we have not taken into account geostrophic advection of the fresh water from the Amazon or the Orinoco), or near major currents (Gulf Stream). Tendencies and low frequencies are also clearly not reproduced by these model simulations. Nonetheless, we feel that this simple model or an improved model incorporating also horizontal geostrophic advection, could be the basis for implementing a Kalman filter assimilation of the data. We plan now to extract a reduced space from the simulated fields (through EOF decomposition) and apply the Kalman filter in this reduced space. We will also have to eliminate from this approach areas where the model performance is too poor to be able to reproduce the variability.



Extension of the analysis with CORIOLIS SSS analyses

Thanks to the advent of the ARGO profiling fleet in the last 5 years providing low resolution surveys of T and S, Coriolis now performs routinely 3-D analyses (www.coriolis.eu.org) of temperature and salinity from a level at 5m below the surface to 2000 meters depth. We carry within the ORE SSS7 a regular sampling of SSS since 1997 by the M.V. Nuka Arctica (ship of opportunity) along 60°N, which data are carefully validated against surface samples. These data were used to validate the 5m Coriolis analyses (we used the 2000-2005 re-analysis: CORA_ATL_02 produced by E. Autret). The comparisons suggest that the Coriolis analyses present little biases in those regions with small spatial variability. The two analyses present a rms difference of 0.035 psu and similar trends over 4 years (2002-2005). These comparisons are very encouraging and suggest that one can merge SSS data with the Coriolis analyses in order to produce reliable fields of SSS. With the data on hand, this effort can help produce better fields for the period 2002 to 2005, and therefore extend the previous SSS analysis in time to the present.

Various attempts were carried to test the merging of the two data sets, which suggest the benefit of incorporating the SSS data into the Coriolis analysis. In particular, we tested incorporating data from the 2003 Picasso cruise on the Marion Dufresne which covers areas not adequately sampled by the ARGO floats. We also tested incorporating data from a set of SSS collected by 15 salinity drifters in the Bay of Biscaye in 2005 (Cosmos project), and which resolve finer scales of the variability than the other data set. In both cases, the residuals from the analyses are coherent with errors and meso-scale variability, and there is an impact of the additional SSS data on the field, in particular in the regions of marked fronts (near the western boundary). Near surface data from thermosalinometer and drifting buoys can thus be merged with the ARGO data set increasing thus the resolution and the space coverage (Figure 4).

Figure 4 Salinity at 5m on June 1 2003. An

analysis produced by combining profile data (black dots) with TSG

measurements (green) from selected cruises (PICASSO on R.V. Marion

Dufresne, R.V. Thalassa and Suroit cruises).

Conclusions

We have good hope to be able to map since 2002 rather accurate large scale fields of SSS that can be used for studies of climate variability or water cycle, and that can be integrated as a boundary condition to numerical simulations. However, these fields (as well as the subsurface Coriolis fields) will not resolve the meso-scales, and a clever approach has to be put in place to incorporate this missing information in numerical simulations. For the prior period (1977-2002), the current product obtained by objective mapping techniques does present large errors in undersampled areas, and great care has to be applied in its use. The attempts to use more sophisticated techniques for its analysis might improve it, but will nonetheless be insufficient in some poorly sampled areas, for example in the southern Atlantic or in the region straddled by freshwater originating from the Orinoco or the Amazon. In this last region, there might be hope to improve the product using also information on dissolved coloured

7 http://www.legos.obs-mip.fr/en/observations/sss/

www.coriolis.eu.org



organic matter from Seawifs satellite data (since 1997), which is apparently correlated to SSS. We also need to test the improvement to the model obtained by incorporating currents mapped from altimetric sea level fields (since 1992).

Acknowledgments In addition to support by GMMC, we have to thank the hard work done at Coriolis to validate the float data and produce valuable near-reliable analyses of temperature and salinity. The dedication of the crew on ships of opportunity and of personnel of IRD and the ORE SSS has also been extremely instrumental in keeping on-going surface data collected for now three decades.

References

Delcroix, T., A. Dessier, Y. Gouriou, and M. McPhaden, 2005: Time and space scales for sea surface salinity in the tropical oceans. Deep Sea. Res., 52/5, 787-813, doi:10.1016/j.dsr.2004.11.012.

Reverdin, G., F. Durand, J. Mortensen, F. Schott, H. Valdimarsson, W. Zenk, 2002. Recent changes in the surface salinity of the North Atlantic subpolar gyre. J. Geophys. Res., 107, C12, 8010, doi:10.1029/2001JC001010.

Reverdin, G., E. Kestenare, C. Frankignoul, and T. Delcroix, 2006: Surface salinity in the Atlantic Ocean (30°S-50°N). Progress in oceanogr., Accepted.


Using PROVOR floats to assess the link between ENSO and the salinity variability in the Western Pacific warm pool

Using PROVOR floats to assess the link between ENSO and the salinity variability in the Western Pacific warm pool By Thierry Delcroix 1 and Christophe Maes1 1 IRD LEGOS

Introduction

The distribution of salt in the global ocean and its variability on different time scales are of great importance in understanding the ocean’s role in the Earth’s climate. Notable in this regard are salinity changes in the western tropical Pacific, which influence air-sea interactions involved in the El Niño Southern Oscillation (ENSO) phenomenon. This region, usually referred to as the warm pool, is characterized in permanence by the warmest water in the World Ocean, with sea surface temperature (SST) warmer than 28°-29°C over an area larger than the western Europe, and with relatively low sea surface salinity (SSS) fresher than 35 (Figure 1). Because of the high mean SSTs in this region, model results indicate that small SST anomalies, O(0.5 to 1°C), result in significant changes in the ocean-atmosphere coupling of relevance to ENSO and extra-tropical weather anomalies (Palmer and Mansfield, 1984; Hoerling and Kumar, 2003).

Figure 1 Mean sea surface (left) temperature and (right) salinity in the tropical Pacific

The eastern edge of the warm pool is most often characterized by a zonal sea SSS front which results from the convergence of low- and high-salinity water masses advected from the western and central Pacific, respectively (Picaut et al., 1996). The position of this front is subject to large eastward and westward displacements of as much as 8000 km in synchrony with El Niño and La Niña event (Delcroix and Picaut, 1998). Aside from the existence of a well-marked SSS front in longitude, the warm pool is also characterized by a peculiar thermohaline structure in the vertical. There, the temperature mixed layer (typically extending to about 80 m depth) is not mixed in salinity most of the time, so that the density mixed layer (typically extending to about 50 m depth) is controlled by the salt stratification. The difference between the bottom of the density mixed layer and the bottom of the temperature mixed layer has been called the barrier layer (Lukas and Lindstrom, 1991) and its formation results from different mechanisms that implied surface forcing and complex dynamical responses of the oceanic upper layers (Figure 2).



Figure 2 Schematic of the longitude-depth thermohaline structure along the equator and at the eastern edge of the western Pacific warm

pool.

In the presence of a relatively thin salinity-controlled mixed layer and barrier layer, the ocean can develop a very energetic response to wind and heat flux forcing (Vialard and Delecluse, 1998). Sensitivity experiments with coupled models indicate that the existence of a barrier layer is crucial for the El Niño development (Maes et al., 2002). Moreover, regional comparisons between dynamic height anomalies computed from temperature and salinity profiles (or altimeter-derived SLA) versus dynamic height anomalies computed from temperature profiles and mean TS curves highlighted differences of the order of 5-10 cm (Delcroix et al., 1987). These differences reflect the contribution of the salinity variability in sea level (Maes et al., 2000). Models assimilating altimeter data and correcting only the temperature field do not account for this contribution. Such an approach may yield to inaccurate initial conditions resulting from the assimilation process (Ji et al., 2000) and potentially affects ENSO prediction (Ballabrera et al., 2002). It is thus crucial to improve our understanding of the actual role of salinity changes in the western Pacific warm pool and to ensure that vertical projections of the sea surface height variability are correctly ensured in terms of both salinity and temperature anomalies.

First results

Due to the relative lack of in situ measurements in the western Pacific warm pool, ten PROVOR floats were deployed along the equator in April 2005 during the FRONTALIS 3 cruise onboard the R/V Alis (Figure 3). For the first time, these drifters provide almost-continuous time series of the thermohaline changes on each side of the zonal salinity front located at the eastern edge of the warm pool. (Note that 10 PROVOR floats were also deployed between 10°S and 20°S along 165°E during the cruise, for scientific objectives which are not discussed here).

Figure 3 (left) Location on May, 23, 2005, of the 20 PROVOR floats deployed during the FRONTALIS 3 cruise on board the R/V l’Alis.

(right) Deployment of one float in the western Pacific warm pool with calm sea conditions (credits IRD).



The returned measurements are presently being analyzed, as one year of data is at least required to derive robust conclusions. Early results were however obtained regarding, at least: a) the quality of the PROVOR sensors and, b) the analysis of temperature and salinity profiles collected prior to the FRONTALIS 3 cruise.

Figure 4 shows the excellent agreement between the temperature and salinity measurements derived from the descent profile of the PROVOR (WMO-ID 5900904) and nearby CTD profile, obtained at 0°-166°E on May, 5, 2005. Similar encouraging results were obtained from the other descent profiles of the floats, as well as when comparing the deepest PROVOR measurements with the 165°E climatology established by Gouriou and Toole (1993). Moreover, Time series of 10-day apart temperature and salinity data on the 27.4 kg.m-3 isopycne (located within 950 and 1100 m depth) show a remarkable stability of the sensors for most floats, at least for the first year of data. This is illustrated in Figure 5 for a given PROVOR float (WMO-ID 5900911), with a salinity increase of about 0.05 during the May 2005 – April 2006 period. Such an increase likely reflects a real physical feature which is presently being analyzed.

Figure 4 Vertical profiles of (red) potential temperature, (blue) salinity, and (black) potential density derived from the descent profile of the

WMO-ID 5900904 PROVOR (full lines) and almost concomitant CTD profile (dashed lines) collected at 0°-166°E on May, 5, 2005.



Figure 5 Times series of (top) the depth of 27.4 kg.m-3 isopycne, (middle) the temperature and (bottom) the salinity on this isopycne for a

PROVOR float (WMO-ID 5900911) deployed by the end of May 2005 along the equator in the western Pacific warm pool.

Figure 6 shows the longitude-time section of SSS and SST changes averaged within 3°N-3°S for the 3-year period preceding the deployment of the PROVOR floats. It reveals that the warmest SST are mainly found near and just west of the SSS salinity front, with values higher than 29.75°C well above the 28°-29°C threshold required for organized atmospheric convection featuring the ENSO ocean-atmosphere coupling. The ongoing analysis of the ten PROVOR-derived measurements, now deployed for one year in the western Pacific warm pool, will provide us the unique opportunity to assess the key role of salinity in this coupling.



Figure 6 <3°N-3°S> averaged SST (left) and SSS (right) variability in the western Pacific, as obtained from the combined use of

TAO/TRITON, TSG and ARGO data (Adapted from Maes et al., 2006).

Acknowledgements

The 20 PROVOR floats deployed in May 2005 in the western tropical Pacific were founded by the CORIOLIS / MERCATOR project. Contribution of the IRD US025 scientists prior to and during the FRONTALIS 3 cruise, as well as the usual assistance at sea of the captain and crews, was much appreciated.

References

Ballabrera-Poy, J., R. Murtugudde, A. Busalacchi, 2002. On the potential impact of sera-surface salinity observations on ENSO predictions. J. Geophys. Res., 107, 8007, doi:10.1029/2001JC000834.

Delcroix T., G. Eldin et C. Hénin, 1987. Upper ocean water masses and transport in the western tropical Pacific (165°E). J. Phys. Oceanogr., 17, 2248-2262

Delcroix, T., and J. Picaut, 1998. Zonal displacement of the western equatorial Pacific fresh pool, J. Geophys. Res., 103, 1087-1098.

Gouriou, Y., and J. Toole, 1993. Mean circulation of the upper layers of the western equatorial Pacific Ocean. J. Geophys. Res., 98, 22 495-22.520.

Hoerling, M., and A. Kumar, 2003. The perfect ocean for drought, Science, 299, 691-694.

Ji, M., R. W. Reynolds, and D. W. Behringer, 2000. Use of TOPEX/Poseidon sea level data for ocean analyses and ENSO prediction: some early results. J. Climate, 13, 216-231.

Lukas, R., and E. Lindstrom, 1991. The mixed layer of the western equatorial Pacific Ocean, J. Geophys. Res., 96 (suppl), 3343-3357.

Maes C., D. Behringer, R. W. Reynolds, and M. Ji, 2000. Retrospective analysis of the salinity variability in the western tropical Pacific Ocean using an indirect minimization approach. J. Atmos. Oceano. Tech., 17, 512-524.



Maes C., J. Picaut, and S. Belamari, 2002. Barrier layer and onset of El Nino in a Pacific coupled model, Geophys. Res. Let., 24, doi:10.1029/2002GL016029. Maes, C., K. Ando, T. Delcroix, W. Kessler, M. McPhaden, and D. Roemmich, 2006. Observed correlation of surface salinity, temperature and barrier layer at the eastern edge of the western Pacific warm pool. Geophys. Res. Let., 33, doi:10.1029/2005GL024772.

Palmer, T., and D. Mansfield, 1984. Response of two atmospheric general circulation models to sea surface temperature anomalies in the tropical east and west Pacific, Nature, 310, 483-485.

Picaut, J., M. Ioualalen, C. Menkes, T. Delcroix and M. McPhaden, 1996. Mechanism of the zonal displacements of the Pacific Warm Pool, implications for ENSO, Science, 274, 1486-1489.

Vialard, J. and P. Delecluse, 1998. An OGCM study for the TOGA decade. Part I: Role of salinity in the physics of the western Pacific fresh pool, J. Phys. Oceanogr., 28, 1071-1088.


Poleward propagation of spiciness anomalies in the North Atlantic Ocean

Poleward propagation of spiciness anomalies in the North Atlantic Ocean By Audine Laurian1,Alban Lazar1and Gilles Reverdin1 1 LOCEAN/IPSL, CNRS/UPMC, Paris

Introduction

The upper branch of the Atlantic thermohaline circulation (THC) plays an important role in global climate variability by transporting large amounts of heat and salt poleward. While most coupled models [Manabe et al., 1993; Häkkinen, 1999] suggest a weakening of the THC in response to global warming and freshening at high latitudes, Latif et al. [2000] among others suggest that poleward advection of tropical waters with anomalously high salinity could compensate global warming effects by stabilizing the THC. In the North Atlantic Ocean (NATL) the salinity maximum water (SMW) region is a main region for the renewal of the subtropical gyre thermocline waters and large volumes of anomalously salty waters can be produced there by ventilation [Blanke et al., 2002; Lazar et al., 2002]. The advection of salinity anomalies from low to high latitudes appears from Thorpe et al. [2001] to be probably achieved by a subsurface mechanism over a time scale of 5-6 decades [Vellinga and Wu, 2004]. A possible candidate to account for this mechanism is the subsurface interannual compensated salinity anomalies (CSAs). They have been shown to propagate along isopycnal surfaces at current speed in the Pacific Ocean [Schneider, 2000; Nonaka and Xie, 2000; Yeager and Large, 2004; Luo et al., 2005] and in the South Atlantic Ocean [Lazar et al., 2001]. The research project sponsored by the GMMC focuses on the analysis of the propagation of these CSAs in the NATL and their relation to the surface signals with a 55-years flux corrected simulation based on the OPA8.1 and with the MERA-11 product. Results based on the flux corrected simulation have been submitted to Geophysical Research Letters [Laurian et al., 2006].

Models and methodology

We use monthly mean outputs from the ORCA2 configuration of the OPA 8.1 model (ORCA2 hereafter). The zonal resolution is 2°, the meridional resolution is 2°×cos(latitude), increasing to 1/3° at the equator and there are 31 fixed vertical levels with variable thickness. The model uses the Gent and McWilliams [1990] eddy parameterization scheme poleward of 20°, and the turbulent kinetic energy scheme of Blanke and Delecluse [1993] is used for vertical mixing. Surface heat fluxes are calculated using bulk formulas based on daily mean NCEP reanalyses over 1948-2002, and the surface boundary condition for freshwater fluxes is a relaxation of salinity to annual mean values from the Boyer et al. [1998] climatology with restoring time of 12 days. We also use monthly mean outputs fromthe MERA-11 product in the MNATL configuration of the OPA 8.1 model (MERA-11 hereafter). The horizontal resolution is 1/3° at the equator, 1/4° north of 45°N, 1/6° north of 60°N and there are 43 fixed vertical levels with variable thickness. Altimetric data (sea surface height) and in situ observations (T and S profiles) are assimilated. Daily forcings over the period 1992-2001 are derived from corrected (evaporation, humidity) ERA40 atmospheric fields with improved CLIO bulk formulae. Neither the SST nor the sea surface salinity (SSS) are restored to the climatology. Observations of SST [Reynolds and Smith, 1994] and climatology of SSS are assimilated with a non gaussian error. Thanks to this method, the mesocale and the interanual signal are not damped at the surface.

Supposing that the CSAs are forced by the surface, we examine the mixed layer depth (MLD) cycle and the SMW distribution. The maximum MLD in the subtropical gyre in MERA-11 occurs in late February-early April, about a month later than in the observations (Figure 1) possibly because of a different criteria defining the MLD (0.05 kg.m-3 in MERA-11, 0.03 kg.m-3 in de Boyer Montégut et al. [2004]). The annual mean position of the SMW region in MERA-11 is in good agreement with observations (Figure 2). Isopycnal surfaces ranging from 25 σ to 26σ (σ = ρ -1000, ρ is the potential density in kg.m -3)

outcrop in the SMW region and we shall focus on 26 σ along which water masses are the most able to reach high latitudes at the interannual timescale possibly affecting the salinity field there.

To analyse and follow CSAs along their pathway, we project the salinity field onto time-varying potential density surfaces,

referenced to the surface pressure. A CSA is defined as the difference between the salinity on the isopycnal, σS and its

monthly climatology σS over 1992-2001 when comparing MERA-11 and ORCA2 and over 1948-2002 when analysing ORCA2 only.

In order to track these subsurface anomalies, we need a time integrated view of their circulation. Thermocline pathways can be described assuming that turbulent mixing is sufficiently weak (true away from the surface, upwelling regions and boundaries) so that waters conserve their density and Bernoulli function (B hereafter). Under this hypothesis, fluid trajectories can be approximated by the isopycnal projection of monthly B, defined as:



∫ −+++=0

)(00

2200

0

])([)(2

)(σ

σσηρρ

σz

dzzggvuB

where σ, u, v and η are monthly values of potential density, horizontal velocity components, and sea surface elevation. As shown later, the downstream tracking of CSAs can be achieved by plotting x or y averages of CSAs within a given tube of monthly B following Lazar et al. [2001].

Figure 1 Month of deepest monthly mean mixed layer in MERA11 (left) and in de Boyer Montégut et al. [2004] climatology (right). The

rectangle indicates the annual mean position of the SMW region. (colorbar in month)

Figure 2 Annual mean sea surface salinity in MERA-11 (left) and in the WOA01 database (right).

Results

The distribution of averaged annual mean subsurface salinity anomalies and surface salinity anomalies in the subtropical gyre (70°W-10°E, 0°-35°N) over the period 1992-2001 is shown on Figure 3. The amplitude of subsurface CSAs in MERA-11 is about 0.3 psu in absolute value. It is about three times weaker in ORCA2 where it ranges between -0.1 psu and 0.15 psu. Subsurface salinity anomalies in MERA-11 are particularly strong (about -0.6 psu) along 22σ and 26.5σ while in ORCA2

they are stronger between 23σ and 24σ and along 26σ (about 0.3 psu). The amplitude of subsurface salinity anomalies in MERA-11 is about three times the amplitude of surface anomalies. This relation does not hold in ORCA2.



Figure 3 Time averaged annual mean salinity anomalies as a function of density in 70°W-10°E, 0°-35°N over the period 1992-2001 at the

subsurface (black) and at the surface (green). Left: MERA-11, right: ORCA2.

To better understand these discrepancies, the spatial distribution and amplitude of the subsurface CSAs generated in the SMW region in MERA-11 and in ORCA2 is highlighted by their root mean square (rms) over 1992-2001 on Figure 4. Along 25.5 σ

and 26 σ the subsurface CSAs in MERA-11 are about 0.1 psu in the interior of the subtropical gyre but they are about 0.3 psu between 5°N and 15°N where the ITCZ is located while in ORCA2 the subsurface CSAs are about 0.03 psu to 0.1 psu in the subtropical gyre. The regions of strong CSAs rms in MERA-11 are not distributed along the mean currents as opposed to regions of strong CSAs rms in ORCA2, specially along 26σ (compare the spatial distribution of the CSAs with the isolines of

annual mean B on 26σ in MERA-11 and with the isolines of mean winter B on 26σ in ORCA2). This suggests that the propagation of the subsurface CSAs in MERA-11 is hard to identify compared to ORCA2. This may be explained by the strong precipitation occuring in the ITCZ region which yields to a strong variability in the daily forcing used. Note that the low rms of the subsurface CSAs in MERA-11 between 10°S and 20°S (between 0 and 0.06 psu) is due to the restoration of the salinity field to climatological values.

To illustrate a typical spatial distribution of the subsurface CSAs in MERA-11, Figure 5 shows a snapshot of the CSAs along 26σ on March, 1998. The amplitude of CSAs between 5°N and 15°N is stronger than 0.3 psu as indicated by their rms. In the eastern part of the basin, a negative CSAs of about -0.3 psu is generated along the outcrop line and enters the seasonal thermocline. However, it is not seen to propagate the following months (not shown) maybe because of the time discontinuities introduced by the assimilation scheme.

Figure 4 Root mean square of subsurface CSAs for the period 1992-2001 in MERA-11 on 25.5σ (upper left) and 26σ (upper right)

and in ORCA2 on 25.5σ (lower left) and 26σ (lower right). Winter mean outcrop line of density (red) and maximum of sea surface salinity (bold) are contoured. The annual mean Bernoulli isolines are shown for MERA-11 (upper right), the winter mean

Bernoulli isolines are shown for ORCA2. The CSAs in MERA-11 have been smoothed with a polynomial function.



Figure 5 Subsurface salinity anomalies along 26σ and sea surface salinity anomalies north of the outcrop line (black line) on March,

1998 in MERA-11. The arrow indicates the position of the generation of a negative anomaly. In the east of the basin, subsurface salinity anomalies are not defined south of the interannual outcrop line indicating that in this region, surface water masses are

lighter than in the climatology.

The spatial distribution and amplitude of subsurface CSAs in ORCA2 between 1948 and 2002 is represented by their rms along 26σ on Figure 6. A dipole south of the winter mean 26σ outcrop line between 41°W-23°N and 20°W-16°N is composed of two regions of high CSA rms (.2 psu) situated northwest and southeast of the SMW region. The amplitude of CSAs is of the same order as what was found in previous studies [Schneider, 2000; Yeager and Large, 2004]. Note the strong CSA rms (.18 psu) along the equator corresponding to the signals described by Lazar et al. [2001]. The late winter sea surface salinity anomalies (SSSAs) rms north of the winter mean 26σ outcrop line highlights a dipole as well. The regions of strong SSSAs rms are collocated with regions of strong SSS gradients (not shown) where current anomalies will have the strongest impacts thus favouring advective processes rather than atmospheric processes to account for the generation of CSAs. Note the maximum of variance of SSSAs in the Gulf Stream explained by this mechanism. The collocation between the dipoles of CSA rms and SSSA rms, which are persistent structures, strongly supports our hypothesis of surface generation of the CSAs.

0.2

0.18

0.16

0.14

0.12

0.1

0.08

0.06

0.04

0.02

0

LONGITUDE100°W 80°W 60°W 40°W 20°W 0° 20°W

LATITUDE

40°N

20°N

0°

Figure 6 Rms CSAs on 26σ and rms SSSAs north of the winter mean outcrop line (bold black contour) from 1948-2002. Winter mean

isolines of Bernoulli function (considering monthly means of ρ , u and v over 1948-2002) are contoured in white and the annual

mean isolines of SSS are also shown

We estimate the timescale of the circulation of the subsurface CSAs in ORCA2 simulation with lagged correlations (Figure 7). The reference time series corresponds to monthly CSAs along 26σ over 1948-2002 averaged in 55-60°W, 13-18°N. This box is selected to retain only the CSAs entering the Caribbean Sea since they have the greatest potential to modulate the salinity



field at high latitudes in a 5 to 10 years timescale (CSAs flowing north of the Caribbean tend to recirculate and reach high latitudes on a longer time scale; Blanke et al. [2002]). We correlate the monthly CSAs along 26σ with this time series. The amplitude of these CSAs is in the range ±15 psu and they propagate as a coherent signal over 6 years following the currents, from the SMW region to 30°N in the Gulf Stream. This time series and the time series of late winter SSSAs averaged in the SMW region (20°W-30°W, 22°N-28°N) lagged by 3 years are highly correlated (r=0.8), a result which supports further a generation mechanism by surface processes. The correlation drops by 50% when considering the time series of annual SSSAs in the same region indicating that late winter subduction mechanism is at play. These results hold for isopycnals ranging from 25 to 26.5 (not shown). Although correlations are not significant downstream of this region after a 3-year lag and despite a decrease of about 66% of the amplitude of the CSAs (compare time series in Figure 7), they determine to a large extent (nearly 70%) the subsurface spiciness of the Gulf Stream water up to 30°N, upstream of the outcrop region.

100°W 80°W 60°W 40°W 20°WLONGITUDE

LATITUDE

10°N

20°N

30°N

40°N0.120.060.040

-0.04

-0.06-0.12-0.06

1950 1960 19801970 1990 2000

r=.8r=.8r=.8r=.8

3

0

-2

-4-3

Figure 7 Correlations higher than 0.8 between monthly CSAs on 26σ and monthly CSAs on 26σ averaged in the index zone (black

rectangle and black time series) are shown for lags -4, -3, -2, 0 and 3 year. The dark grey time series is annual CSAs averaged in 80°W-78°W, 30°N with a 3-year negative lag and the light grey time series is late winter SSSAs averaged in 20°W-30°W,

22°N-28°N with a 3-year positive lag. The winter mean outcrop line of 26σ is shown.

To display the horizontal and vertical patterns of 26σ CSAs, we show them in February 1980 when strong signals emanating from the dipole described above prevail (Figure 8). Note the wide horizontal extension of the CSAs compared to MERA-11 (Figure 5).The horizontal structure (Figure 8a) displays two anomalies extending in the direction of the currents. A negative one (about -0.3 psu) has been created in the western SMW region, part of it extends toward the north of the Caribbean and part of it toward the Gulf of Mexico. A positive one (about 0.2) has been created in the east of the SMW region, part of it extends toward the Caribbean and part of it toward the North Equatorial Countercurrent. This structure is quite common from 1948-2002. The mixing between positive and negative CSAs of the dipole favours the damping of the signal and may explain the low rms of CSAs entering the Caribbean Sea (Figure 6). The vertical structure of the CSAs described above from the permanent thermocline in the SMW region to the Yucatan Strait and averaged in latitude between the two isolines of B noted by the arrow on Figure 8a is presented on Figure 8b. The core of the negative CSA (-0.18 psu) centered on 26σ extends from 25.5σ to

26.8σ (vertical extension of about 90m). A large positive CSA (0.3 psu) extends from 23.5σ to 25σ (vertical extension of about 15m). Again, the two opposite signals favour and increase the damping.



a)

b)

80°W 60°W 40°W 20°W 0° 20°ELONGITUDE

LATITUDE

100°W

40°N

20°N

0°

24

25

26

27

28

sigma

80°W 70°W 60°W 50°W 40°W 25°N 30°NLONGITUDE LATITUDE

76

86

185

425

meters

0.3

0.24

0.18

0.12

0.06

0

Figure 8 a) ORCA2 CSAs along the 26σ and SSSAs in February 1980. The winter mean isolines of B (white contours), the outcrop

region of 26σ in February 1980 and the index zone are also shown. The tracking of CSAs is achieved by plotting x-y averages of CSAs within the two isolines of B delimited by the arrow. b) Salinity anomalies averaged in longitude north of 20°N and in

latitude south of 20°N between the two isolines of B noted by the arrow on Figure 8a from the SMW region to the Yucatan Strait in February 1980 (potential density as vertical left axis, mean depth in meter as right axis) are contoured.

To document the variability of the creation and propagation of CSAs, a Hovmöller diagram of monthly CSAs along 26σ and along their pathway (averaged between the isolines of B noted by the arrow on Figure 8a is presented on Figure 9 as well as the time series of late winter SSSAs averaged in 20°W-30°W, 22°N-28°N. As was said before, the time series (a) strongly supports a subduction mechanism for the generation of CSAs. Moreover, sign-liked CSAs are generated during successive years. For instance, positive CSAs were created from 1948 to 1955 and propagated via the Gulf of Mexico (b-d) toward Cape Hatteras (e) in about 6 years with an amplitude decreasing from 0.15 to 0.05 psu along their way. From 1955 to 1966, negative CSAs were injected in the thermocline and propagated in 6-8 years with an amplitude decreasing from -0.2 to -0.06 psu along the same pathway. The decrease of amplitude of the signals is due to mixing along the pathway and amplified by the horizontal and vertical dipoles described above. Another mechanism alters the amplitude of the CSAs: for instance the positive anomalies reaching the vicinity of Cape Hatteras (e) between 1990 and 2000 have stronger amplitude than they did when they were generated (b). A possible explanation is the mixing along 26σ of these CSAs with stronger positive anomalies that drifted north of the Caribbean Islands reaching the same region or the mixing of these CSAs with stronger signals on lighter isopycnals. This alteration of amplitude remains an open question. Despite these reductions in amplitude, about 34% of the signal remains identifiable until Cape Hatteras (e). The propagation speed of these anomalies between the SMW region and the Yucatan Strait (b-c) is about 2cm.s-1, it increases to about 1m.s-1 in the Gulf of Mexico (d) and in the Gulf Stream (e), in agreement with measurements of current speed in these regions.



a) b) c) d) e)-.08 .0 .08 30°N 22°N 42°W 50°W 58°W 66°W 74°W 20°N 26°N 34°N

0.420.320.220.120.10.080.060.040.020-0.02-0.04-0.06-0.08-0.1-0.12-0.2-0.3-0.4-0.519

501960

1955

19651970

197519801985

1990

19952000

Figure 9 a) Time series of late winter SSSAs in the formation region (20°W-30°W, 22°N-28°N). b) to e) Hovmöller diagrams of the

monthly CSAs on 26σ along the pathway. b) Outcrop region in the SMW, c) from the SMW to the Gulf of Mexico, d) Gulf of Mexico, and e) from Florida to Cape Hatteras.

Discussion

We have analysed the generation and circulation of subsurface compensated salinity anomalies from the SMW region of the North Atlantic Ocean to 30°N in the Gulf Stream in a 55-years flux corrected simulation. The largest CSAs (0.1 up to .35 psu) are seen to propagate via the Gulf of Mexico along 26σ at current speed on a typical 6-year transit. The low-frequency of the signal appears to be forced by late winter SSSAs. Although mixing along the pathway reduces by about 66% the amplitude of the CSAs, they determine to a large extent (nearly 70%) the subsurface spiciness of the Gulf Stream water up to 30°N, upstream of the outcrop region. We have examined the main features of this propagative mechanism in MERA-11. The timing of the generation of the subsurface CSAs, the region where it happens and the mean circulation of water masses in MERA-11 are in good agreement with observations. However, the spatial structure of the compensated signals and their advection seems to be strongly dependent on the high frequency forcing and on the assimilation scheme. In particular, the subsurface CSAs have a strong variability in the ITCZ region where the precipitation rate is very high. Their propagation is therefore not obvious.

To further analyse and quantify the influence of the assimilation scheme on the propagative mechanism of interannual compensated salinity anomalies, we will compare MERA-11 to a similar product computed without assimilation (constructed by Eric Greiner).

References

Blanke, B., and P. Delecluse (1993), Variability of the tropical Atlantic ocean simulated by a general circulation model with two different mixed layer physics, J. Phys. Oceanogr., 27, 1038-1053.

Blanke, B., M. Arhan, A. Lazar, and G. Prévost (2002), A Lagrangian numerical investigation of the origins of the salinity maximum water in the Atlantic, J. Geophys. Res., 107, 3163, doi: 10.1029/2002JC001318.

Boyer, T.P., S. Levitus, J. Antonov, M. Conkright, T. O’Brien, and C. Stephens (1998), World Ocean Atlas 1998 Vol.5: Salinity of the Pacific Ocean. NOAA Atlas NESDIS 30. U.S. Gov. Printing Office, Wash., D.C., 166pp.

de Boyer Montégut, C., G. Madec, A. S. Fischer, A. Lazar, and D. Iudicone (2004), Mixed layer depth over the global ocean: an examination of profile data and a profile-based climatology, J. Geophys. Res., 109, C12003, doi:10.1029/2004JC002378.

Gent, P.R., and J.C. McWilliams (1990), Isopycnal mixing in ocean circulation models, J. Phys. Oceanogr., 20, 150-156.



Häkkinen S. (1999), A Simulation of Thermohaline Effects of a Great Salinity Anomaly. J. Clim., 12, 1781-1795.

Laurian, A., A. Lazar, G. Reverdin, K.B. Rodgers, and P. Terray (2006), Poleward Propagation of Spiciness Anomalies in the North Atlantic Ocean, resubmitted to Geophys. Res. Lett.

Latif, M., E. Roeckner, U. Mikolajewicz, and R. Voss (2000), Tropical Stabilization of the Thermohaline Circulation in a Greenhouse Warming Simulation, J. Climate, 13, 1809-1813.

Lazar, A., R. Murtugudde, and A.J. Busalacchi (2001), A model study of temperature anomaly propagation from the subtropics to tropics within the South Atlantic thermocline, Geoph. Res. Lett., 28, 1271-1274.

Lazar, A., T. Inui, P. Malanotte-Rizzoli, A.J. Busalacchi, L. Wang, and R. Murtugudde (2002), Seasonality of the ventilation of the tropical Atlantic thermocline in an ocean general circulation model, J. Geophys. Res., 107, 3104, doi:10.1029/2000JC000667.

Luo, Y., L.M. Rothstein, R.H. Zhang, and A.J. Busalacchi (2005), On the connection between South Pacific subtropical spiciness anomalies and decadal equatorial variability in an ocean general circulation model, J. Geophys, Res., 110, doi:10.1029/2004JC002655.

Manabe, S., and R.J. Stouffer (1993), Century-scale effects of increased atmospheric CO2 on the ocean-atmosphere system, Nature, 364, 215-218.

Nonaka, M., and S.P. Xie (2000), Propagation of North Pacific interdecadal subsurface temperature anomalies in an ocean GCM, Geophys. Res. Lett., 27, 3747-3750.

Reynolds, R.W. and T. Smith (1994), Improved global sea surface temperature analysis using optimum interpolation, J. Clim., 7, 929-948.

Schneider, N. (2000), A decadal spiciness mode in the tropics, Geophys. Res. Lett., 27, 257-260.

Thorpe, R.B., J.M. Gregory, T.C. Johns, R.A. Wood, and J.F.B. Mitchell (2001), Mechanisms Determinig the Atlantic Thermohaline Circulation Response to Greenhouse Gas Forcing in a Non-Flux-Adjusted Coupled Climate Model, J. Clim., 14, 3102, doi:10.1175/1520-0442.

Vellinga, M., and P. Wu (2004), Low-Latitude Freshwater Influence on Centennial Variability of the Atlantic Thermohaline Circulation, J. Clim., 17, 4498-4511.

Yeager, S.G., and W.G. Large (2004), Late-Winter Generation of Spiciness on Subducted Isopycnals, J. Phys. Oceanogr., 34, 1528, doi: 10.1175/1520-0485.


Notebook

Notebook -

Editorial Board

Nathalie Verbrugge

Secretary

Monique Gasc

Articles


By Gilles Garric


By Benoît Tranchant, Lionel Renault, Charles-Emmanuel Testut, Nicolas Ferry and Pierre Brasseur


By Gilles Reverdin, Elodie Kestenare, Jacqueline Boutin, Nadine Chouaib, Fabienne Gaillard and Delphine Mathias

Using PORVOR floats to assess the link between ENSO and the salinity variability in the Western Pacific warm pool

By Thierry Delcroix and Christophe Maes

Contact

Please send us your comments to the following e-mail address : [email protected]

Next issue : July 2006

Mercator Ocean newsletter 21

Environment

Transcript of Mercator Ocean newsletter 21