Ocean (surface) salinity an in situ perspective G. Reverdin LOCEAN, UMR CNRS/UPMC/IRD, Paris, France (indebted to numerous colleagues in France, US, Spain…)

Salinity Water with anions and cations in fixed’ ratios (Dittmar, 40 samples from the Challenger Expedition ) Thus, to 0-order S=fn(C,T), and density=fn(T,S,P) (small increase 0(0.003 psu) due to increase DIC; small overall decrease due to glacial melt/change of ocean mass) S evolves with E-P+R, ocean circulation, mixing

What do we measure Measured with respect to a standard water (more or less since 1900) -first by measuring Chlorinity; -more recently, conductivity Lost in Fathom: London

How do we measure it Discrete samples 110 y (0.1 psu) Thermosalinographs 40 y (ships, drifters) (0.02 psu) Argo floats (0.01 psu) (and other profilers)

Different networks SO SSS G. Alory, T. Delcroix SOCAT, GOSUD, SAMOS (all require careful validation) Surface drifters (SPURS) L. Centurioni, V. Hormann J. Font, G. Reverdin Requires also careful validation)

Mapping of climatology Sufficient data: Last 40 years Reverdin et al, 2007; Delcroix et al., 2005 Gordon and Giulivi, 2014

Examples of variability Subpolar gyre : TSGs (20 years) earlier sampling 100 years NAC LC Binning in boxes or along tracks…

Subpolar gyre Binning data 1°x1 month : scales resolved a few degrees and a few months Large spatial coherence : modulation of gyre Larger signal : circulation

West Greenland / Nfld shelves Larger seasonal modulation on shelves; harder to interpret (for example, expectation of huge melt in , and no low S) Possible phase opposition West Greenland/ Newfoundland shelf

Spatial mapping of individual fields (~10 years) April-June km scales can be retrieved (away from fronts), even with the spare Argo sampling (and surface TSG sampling) ISAS (Gaillard, 2009); Reverdin et al., 2002; Reverdin et al., 2007

Displacement of the fresh pool western equatorial Pacific ( binning Delcroix et al., 2011; scales 1000 km*100km*3 months) Singh et al., 2012, but also Delcroix and Picaut, 1998, Maes et al, 2006 Displacements advected by zoal currents + P influence

ENSO/El Niño Singh et al., 2011 Different flavors of ENSO (EP, CP) With SSS patterns that relate to currents/P

Warm pool (Cravatte et al, 2007) Trends in PDO and SSS

Trends– natural variability A climate change perspective Trend SSS/century in climate models (compared to obs ) Terray et al 2012; Delcroix et al., 2011 Pacific seems to be robust, but N. Atlantic within natural variability

Longer time series (feasible in NE subpolar gyre with some data adjustments over the last 120 years) Reverdin et al., 2010 T and S correlated, but present also differences Low-frequency S presents weak seasonal dependency related to modulation of westerlies (NAO) with 0-4 years lag (0.63) Studies on the 1990s transition indicates that in winter it is related mostly to changes in ocean circulation/ inputs to the gyre (and E-P) Multi-decadal filter Fresh Salty Fresh

Further south The worse sampling Better sampling further north But gaps remain In all cases: issues of corrections/qualification of data (particularly in the 1920s) (requires adjustment of data) There is often a need to bin in ‘big’ boxes (and interannual smoothing + averaging different seasons)

Trends No relation (in phase) between NATL And NASG (possible delayed phase) 45-50°N, already signal characteristic of NASG (2 years earlier) NA TA IG

Meso-scales need to be resolved to study higher frequency variability (seasonal or less) even on the large scales Examples from SPURS (1-year survey of NA subtropical gyre) Feature associated with transport of fresh/warm anomaly from south (Busecke et al., 2014) R. Schmitt A. Gordon

The largest scales have Rms =0.14 (cor=0.5 with SMOS) The smaller meso-scales can have 0.2 psu signals as Much as large scales over 1000 km… Thus possible strong fronts and vertical circulations… (but in some areas/seasons T and S Compensated, which is scale-dependent) Kolodziejczyk et al., 2014 Meso-scale SSS variability 100 km 20 km

SPURS SSS seasonal signal influence of the eddy and meso-scale structures to d(V’S’)/dy Gordon et al, 2014 (SODA reanalysis)

Local SPURS budget evaluated 100 km x 100 km box Estimates of dSSS/dt from drifters Near SPURS moring (red) Compared to Aquarius+Aviso estimate Centurioni et al., 2015 Farrar

Near surface stratification The low wind; high SW case (Asher et al., 2014) Gradients day-time, a few percents of the surface of the oceans

Rainfall-induced stratification S(15-cm) – S(45-cm) 17 events SVP-BS / Surplas (ITCZ/SPCZ) Individual rain events : 25% more at 15-cm, but for less than an hour (also, T decrease and stratification); wind estimated ‘SSMI’, not measured

Extreme rain event! No wind! (means no wave energy in 20 cm – 1m wave length domain) Little S gradient between 4cm and 50 cm (~1 unit) thus at least 12 cm Rain in 90 minutes, but should be more as decrease below 50 cm... Strong T-gradient (implies probably ~100 W/m2 cooling); secondary drops? 1 h

Conclusions & Perspectives -In situ data can be used to have long time series, but issues on some old data remain and cast doubts on some results - No identification of Atlantic basin-scale SSS trends over the last 118 years - When data density high enough, spatial patterns of low frequency modes of variability can be investigated (from 2-D to 3-D in the last 10y) - Dedicated surveys/instrumentation to test certain balances on smaller spatial scales: example of SPURS1 (but also COARE) V’S’ ~ 1 E-3 from SSS (smoothed 200 km/1 month) + Aviso current V’S’ ~ 3 E-3 from drifter SSS + velocities Thus cm of equivalent rainfall (compared to 130 cm for E-P) - Complementarity with satellite data to be developed both in SSS and improved surface current products