Pre-processing of SSS data

Level 2 Ocean Salinity User Data Product (UDP) data for 2013, are used in this project. Only ascending orbits are retained, and the Roughness Model 1 (SSS_1) is used. The full swath is retained.

Spatial resolution is ~7.5km, and temporal resolution is daily.

In order to use SMOS SSS data with DINEOF, the following steps have been performed:):

  1. Quality flags used  (provided with the L2 files):
    1.  Fg_ctrl_poor_geophysical
    2.  Fg_ctrl_poor_retrieval
    3.  Dg_quality_SSS1
  2.  A minimum and maximum check is done pixel by pixel, using as limits the monthly background fields described in this page.
  3. An outlier detection approach is applied, using the technique developed in Alvera-Azcárate et al, 2012. Below is an image showing the application of outlier detection to SSS data.


Three tests are realised:

– Departure from a DINEOF truncated EOF basis
– Departure from a local median
– Proximity to clouds, land or missing values

A weight of 1/3 is given to each of the three sub-tests, and those pixels that have a value higher than 1 are classified as outliers.