Mean Sea Surface decription

The mean sea surface MSS_CNES_CLS_15 has been computed using a 20-year period of altimetry data.

Particular attention was paid to the homogeneity of the ocean content of this MSS, and specific processings were also carried out on data from geodetic missions. For instance, CryoSat-2, Jason-1 GM, and ERS-1 GM data were corrected from oceanic variability using results of optimal analysis of sea level anomalies (SLA).

The MSS CNES_CLS 2015 has been determined using a local least square collocation technique which provides an estimation of calibrated error. Our technique takes into account of altimetric noise, ocean variability noise and also along-track biases which are determined independently for each observation. Moreover, variable cross-covariance models are fitted locally for a more precise determination of the shortest wavelengths lower than 50 km. A specific data processings were used near the coast, the aim has been to preserve better continuity of the MSS accuracy from the open ocean to the coasts.

All these data have been preprocessed in order to be:

  • more homogeneous, and referenced to 1993-2012 period (i.e. GDR standard in Duacs products), 
  • less contaminated by the ocean topography variable signal (the mean ocean topography signal contained in the surface is thus corresponding the mean sea level during the period 1993-2012).

The mean sea surface is corrected from seasonal variations and corresponds to the sum of geoid + mean ocean circulation.
The relief of this surface reveals:

  • on a scale of thousands meters, undulations of hundred of meters due to differences in densities into the mantle;
  • on medium wavelenghts, currents and water masses of varying densities, leading to sea level variations of several meters;
  • on short wavelenghts, the heterogeneous distribution of matter concerning earth's surface layer and ocean's bottom topography, that implies geoid undulations of several meters.

The surface have been estimated on a 1/60 (1 minute) of a degree grid using a local inverse method, which also provides an estimation error field that is calibrated on the results of crossover point analysis of altimetric data.

This mean sea surface contains:

  • Over ocean, the mean geoid plus the mean ocean dynamic topography (1993-2012),
  • In continental areas (up to100 km inland): MSS is not defined

MSS_CNES_CLS_15 mean sea surface,
and relative errors. Credits CNES/CLS


Reference ellipsoidT/P
Reference time period1993-2012 (20 years)
Geographic coverageGlobal (80°S to 84°N) - Oceanwide where altimetric data are available. 
Spatial resolutionRegular grid with a 1/60° (1 minute) spacing (i.e. ~1.8 km)
Grid21600 points in longitude / 9811 points in latitude (-79.5°S to 84° N)
MethodLocal least square collocation method on a 6' grid where altimetric data in a 200-km radius are selected. Estimation on a 1' grid based on (SSH minus First Guess) values (remove/restore technique to recover the full signal). The inverse method uses local anisotropic covariance functions that witness the MSS wavelength content.
Estimation errorYES (in cm) - The Optimal Interpolation method provides a calibrated formal error.

Altimetric dataset


  • ~100 000 000 observations
  • Mean profiles:
    • T/P , Jason-1 & 2,
    • T/P tandem,
    • ERS-2, Envisat,
    • GFO
  • Geodetic phase:
    • ERS-1,
    • Jason-1,
    • CryoSat-2
  • See figure