Consistency of the current global Ocean Observing Systems from an Argo perspective (2014)

K. von Schuckmann(1), J.-B. Sallée (2), D. Chambers (3) , P.-Y. Le Traon (4), C. Cabanes (5), F. Gaillard (6), S. Speich (7) and M. Hamon (8)    

              

(1) : MOI UM 110, La Garde, France

(2) : LOCEAN, Paris, France

(3) : College of Marine Science, University of South Florida, USA

(4) : Mercator Ocean and Ifremer, Ramonville, France

(5) : CNRS, DT/INSU, Plouzané, France

(6) : Ifremer, Brest, France

(7) : UBO, Brest, France

(8) : Mercator Ocean, Ramonville, France

Full paper: von Schuckmann, K., J. B. Sallée, D. Chambers, P. Y. Le Traon, C. Cabanes, F. Gaillard, S. Speich and M. Hamon (2014). "Consistency of the current global ocean observing systems from an Argo perspective." Ocean Science 10(3): 547-557, doi:10.5194/os-10-547-2014.
Context

Global Ocean Indicators (GOIs, von Schuckmann and Le Traon, 2011), such as global Ocean Heat Content (GOHC), global Ocean Freshwater Content (GOFC) or global Ocean Steric Sea Level (GSSL), are a useful benchmark for ocean and climate models and an important diagnostic for changes in the Earth’s climate system (Hansen et al., 2011 ; Levitus et al., 2005). 

In particular, observations during the era of the international ARGO program (Roemmich and the A.S. Team, 2009) have a high potential to deliver accurate data to be used for such studies on the variations in the world’s ocean heat storage and its associated volume changes (von Schuckmann and Le Traon, 2009). 

The objective of this study is to quantify the consistency of near-global and regional integrals of OHC and SSL (from in situ temperature and salinity data), total sea level (from satellite altimeter data) and ocean mass (from satellite gravimetry data) during the period 2005-2012.

Data

This study is an inter-comparison of GOIs obtained from three different global ocean observing systems : 

1 - the Argo network 

The global OHC and global SSL time series are evaluated using a weighted box averaging scheme from Argo data as described in von Schuckmann and Le Traon (2011). Heat content and steric sea level are integrated between 10m depth to 1500m depth, as the number of Argo profiles with data in the range 1500-2000m dramatically drops before the year 2009 (Cabanes et al. 2009)

In order to minimize systematic biases in the global Argo dataset, profiles and platforms known to have problems (mainly with pressure sensors) are excluded from the Argo data set. Every profile « on alert » has been checked visually, which allows excluding spurious data. Monthly gridded fields of temperature and salinity properties of the upper 2000m over the period 2005-2012 (D2CA1S2 re-analysis - see http://wwz.ifremer.fr/lpo/SO-Argo/Products/Global-Ocean-T-S/Monthly-fields-2004-2010 for details) are also used in this study. 

2 - satellite gravimetry from the Gravity Recovery and Climate Experiment (GRACE) 

The variations in the mass component of sea level are computed using observation from GRACE, using the most recent Release_05 data processed by the University of Texas Center fro Space Research (Bettadpur, 2012), modified to correct deficiencies in the geocenter and C2.0 coefficient as described by Chambers and Schröter (2011)

3 - satellite altimetry 

The sea level anomalies are computed from the delayed-mode AVISO gridded merged data product (SSALTO/DUACS, www.aviso.oceanobs.com) based on multiple satellite altimeters.

 

Method

Global sea level change SLTOTAL is related to global steric height time series (SLSTERIC) and mass variability (SLMASS) through : 

 SLTOTAL = SLMASS+SLSTERIC+SLRES     (EQU 1) 

The residuals SLRES are estimated using those three major global observing systems : SLTOTAL is computed from altimetry products (AVISO), SLMASS from satellite gravimetry (GRACE) and SLSTERIC in the upper 1500 meters from temperature and salinity observations from Argo. The method relies on two assumptions : first, it assumes that systematic errors in either satellite altimetry or gravimetry are negligibly small ; second, steric changes in the deep ocean, below 1500 meters, are excluded.  The study  is limited to +/-60° latitude, as the quality of Argo coverage and the quality of altimeter product fall off rapidly  poleward of +/-60° latitude.

Results

Global OHC and SSL from Argo

The Argo-based time series of global OHC and global SSLincrease from 2005 to 2012 with mean rates of 0.5 ± 0.1 W m-2 and 0.5 ± 0.1 mm year-1 respectively (area of the world ocean between 60°S and 60°N, and the 10–1500 m depth layer, Fig. 1). The trend of Argo global OHC is unchanged from the 6-year period (2005–2010) estimated by von Schuckmann and Le Traon (2011) (0.5 ± 0.1 W m-2 ). 

For global SSL, the 8-year trend is 30 % smaller than the previously computed 6-year trend (0.75 ± 0.2 mm year-1 ) of von Schuckmann and Le Traon (2011). However the two trends are consistent within their uncertainty. The difference between the 6 and 8-year trends in GSSL is likely due to the strong interannual signature of El Niño–Southern Oscillation (ENSO) variability in the tropical Pacific during the end of 2010 and beginning of 2011,which is known to strongly affect trend estimates for periods less than about 15 years (e.g., Cazenave et al., 2014). 

Figure 1. Global ocean (60°S–60°N) heat content (upper, global OHC) and steric sea level (lower, global SSL) during the period 2005–2012 from Argo after the method of von Schuckmann and Le Traon (2011). The 8-year trends (red line) of global OHC and global GSSL account for 0.5 ± 0.1 W m-2, and 0.5 ± 0.1 mm year-1 for the 10–1500 m depth layer, respectively. Error bars include data processing and climatology uncertainties, but exclude systematic errors.

 

Observed biases in SLRES

In order to assess the coherence of the three observing systems, the sea level budget (EQU.1) is computed for the global ocean [60°S-60°N] as well as three sectors of the world oceans : the Northern Ocean (NO) between [30°N-60°N], the Tropical Ocean (TO) between [30°S-30°N] and the Southern Ocean (SO) between [30°S-60°S]. 

Figure 2. Residual of the sea level budget at different latitude bands using Argo steric sea level (Fig. 3a, red), AVISO delayed- mode gridded fields and GRACE data. Residual trends amount to 0.3 ± 0.6 mm years-1 for the global ocean, 1.6 ± 0.7 mm years-1 for the Tropical Ocean, 3 ± 0.9 mm years-1 for the Northern Ocean, and 0.7±0.7 mm years-1 for the Southern Ocean.

The global residual SLRES shows a slight and non-significant positive grand of 0.3+/-0.6 mm year-1 (Fig. 2a). In the TO sector, the regional sea level budget has a significant positive trend of 1.6 +/-0.7 mm year-1 (Fig.2b). In the NO sector, the residual of the regional sea level budget has a significant negative trend of -3+/-0.9 mm year-1 (Fig.2c), which makes it the region of the largest redial trend. The SO shows a residual trend of -0.7 +/-0.7 mm year-1 (Fig2d).

Although the global sea level budget over 60°S-60°N can be closed within the error bars for the years 2005-2010, there must be systematic errors in one or all of the three observing systems in the TO or NO sectors that cancel out in global coverage.

Argo sampling issues in the TO sector

Argo has significantly coarser resolution in both, time and space than the satellite systems (especially multi-satellite altimetry) and can therefore alias high-frequency regional signals and be more affected by mesoscale eddies. To test the sensitivity of GOIs to this sampling issue, we subsample altimeter data on position and time of Argo profiles. Global mean SLTOTAL is then recomputed following the procedure of von Schuckmann and Le Traon (2011). 

For the global ocean, the residual trend derived using the subsampled altimeter data [SLRES]SUB changes sign, but the magnitude is still within the calculated standard error (0.6 ± 0.6 mm year-1, Fig. 3a). We can therefore observe an effect of Argo sampling on global SLRES, but it is small enough to remain within error bars. 

Sampling errors in extra tropical areas (NO and SO) are large, but do not fully explain the biases observed in Fig.2. The largest sensitivity of SLRES to Argo sampling is observed in the TO area. When using consistent sampling for altimetry and Argo, the significant positive bias observed for SLRES (Fig. 2b) is strongly reduced for [SLRES]SUB (0.2 ± 0.7 mm year-1 , Fig. 3b). In particular, we show that the area around the Tropical Asian Archipelago (TAA) is important to closing the global sea level budget on interannual to decadal timescales.

Figure 3. Same as Fig. 2, but using subsampled Altimeter data to quantify biases owing to Argo sampling. Residual trends amount to 0.6±0.6 mm years-1 for the global ocean, 0.2 ± 0.7 mm years-1 for the Tropical Ocean, 2.1 ± 0.9 for the Northern Ocean, and 1.5 ± 0.7 for the Southern Ocean.

Perspectives

The comparison of Argo GOIs to other global ocean observing systems such as total sea level from altimetry, and ocean mass observations from satellite gravimetry via the global sea level budget, is not only a potential quality control method to identify systematic biases in the Argo observing system, but also to test the effect of Argo sampling issues on GOI estimations.

We showed that the three observing systems are consistent at global scales within error bars. At regional scale, however, we have identified a systematic bias in some parts of the Tropical Ocean due to sampling issues, in particular the Tropical Asian Archipelago (TAA) region. 

We also found that uncertainties in the observing systems are still too large to indirectly derive deep-ocean steric changes below 1500 m depth via the global sea level budget. This emphasizes the importance of continuing sustained effort in measuring the deep ocean from ship platforms and by beginning a much needed automated deep-Argo network.

References
  • Abraham, J. P., Baringer, M., Bindoff, N. L., Boyer, T., Cheng, L. J., Church, J. A., Conroy, J. L., Domingues, C. M., Fasullo, J. T., Gilson, J., Goni, G., Good, S. A., Gorman, J. M., Gouretski, V., Ishii, M., Johnson, G. C., Kizu, S., Lyman, J. M., Macdonald, A. M., Minkowycz,W. J., Moffitt, S. E., Palmer, M. D., Piola, A. R., Reseghetti, F., Schuckmann, K., Trenberth, K. E., Velicogna, I., and Willis, J. K.: « A review of global ocean temperature observations: Implications for ocean heat content estimates and climate change », Rev. Geophys., 51, 450–483, doi:10.1002/rog.20022, 2013.
  • Bettadpur, S.: « UTCSR Level-2 Processing Standards Document for Level-2 Product Release 0005 », GRACE 327-742, CSR Publ. GR- 12-xx, Rev 4.0, University of Texas at Austin, 16pp., 2012.
  •  Cabanes, C., Grouazel, A., von Schuckmann, K., Hamon, M., Turpin, V., Coatanoan, C., Paris, F., Guinehut, S., Boone, C., Ferry, N., de Boyer Montégut, C., Carval, T., Reverdin, G., Pouliquen, S., and Le Traon, P.-Y.: « The CORA dataset: validation and diagnostics of in-situ ocean temperature and salinity measurements », Ocean Sci., 9, 1–18, doi:10.5194/os-9-1-2013, 2013.
  • Chambers, D. and Schröter, J.: « Measuring ocean mass variability from satellite gravimetry », J. Geodynam. 52, 333–343, doi:10.1016/j.jog.2011.04.004, 2011.
  • Hansen, J., Sato, M., Kharecha, P., and von Schuckmann, K.: Earth’s energy imbalance and implications, Atmos. Chem. Phys., 11, 13421–13449, doi:10.5194/acp-11-13421-2011, 2011. 
  • Levitus, S., J. Antonov and T. Boyer : « Warming of the World Ocean, 1955-2003 », Geophysical Research Letters, 32, doi:10.1029/2004GL021892
  •  Nerem, R. S., Chambers, D. P., Leuliette, E. W., Mitchum, G. T., and Giese, B. S.: Variations in global mean sea level associated with the 1997–1998 ENSO event: Implications for measuring long term sea level change, Geophys. Res. Lett., 26, 3005–3008, 1999
  • Roemmich, D. and the A. S. Team: Argo: « The Challenge of Continuing 10 Years of Prog. Oceanogr. », 20, 26–35, 2009
  • von Schuckmann, K. and Le Traon, P.-Y.: « How well can we derive Global Ocean Indicators from Argo data? », Ocean Sci., 7, 783– 791, doi:10.5194/os-7-783-2011, 2011