Which Salinity Data are Best for You?

SSS Retrieval – It's Complicated

Satellite salinity instruments measure natural microwave emission from the top 1 cm (or less) of ocean surface in terms of brightness temperature (TB). These data are processed into SSS at various levels:
  • Level 1: Radiometer observations at full resolution (1A) or processed to sensor units such as brightness temperature (1B);
  • Level 2: Derived geophysical variables from Level 1 source data;
  • Level 3: Variables mapped on uniform space-time grid scales; and
  • Level 4: Model output or results from analyses of lower-level data (e.g., variables derived from multiple measurements).

An example of the Level 4 product is the Optimum Interpolation Sea Surface Salinity (OISSS) analysis which combines Level 2 observations from NASA's Aquarius/SAC-D and SMAP satellite missions into a continuous and consistent, gridded sea surface salinity dataset. Measurements from ESA's SMOS satellite are used to fill in gaps when SMAP is unable to deliver scientific data. The OISSS analysis also uses a dedicated bias-correction algorithm to correct the satellite retrievals for large-scale biases with respect to in-situ data. | OISSS maps | OISSS Climatology maps | OISSS Anomaly maps |

It can be useful to "dive deeper" and be aware of the types of documents describing how the data are acquired, validated, calibrated, processed and distributed. The complexity behind SSS data has been captured in the PO.DAAC's Aquarius Documentation Roadmap. In addition to User Guides, there are Validation Analyses, Algorithm Descriptions, Calibration and other Technical Reports. The PO.DAAC also offers similar documentation for SMAP SSS processing.

Aquarius documentation roadmap
Aquarius documentation roadmap.

The complexity behind SSS data has been captured in the PO.DAAC's Aquarius Documentation Roadmap. In addition to User Guides, there are Validation Analyses, Algorithm Descriptions, Calibration and other Technical Reports. The PO.DAAC also offers similar documentation for SMAP SSS processing.

There are differences among the retrieval algorithms used to generate satellite SSS data. Not only that, the algorithms used to process various missions' data evolve over time. So, the information below is meant to provide an overview of the types of variations among data products.

The variation among SSS data products falls into these general categories:

  • Ancillary Inputs
    • Typical inputs include sea surface temperature (SST), wind speed and direction, galactic map, sea ice mask, dielectric constant model, and a reference salinity field (e.g., HYCOM model). Differences among these inputs can affect the consistency among SSS products.
  • Corrections
    • A variety of corrections are applied to SSS data. Although the specific corrections differ from product-to-product, typical SSS corrections are summarized in the e-brochure, Ocean Salinity From Space.
  • Flags, Filters & Masks
    • Flags are employed when values exceed threshold values, thus indicating potentially degraded algorithm performance.
      • Each SSS data product employs its own type of flagging. For example, flags can affect data availability and/or quality near coasts (i.e., land contamination) and at high latitudes (i.e., sea ice contamination).
      • "Research Insights" show the affect of ice flagging and masking on retrievals of SSS data from the Arctic Ocean.
    • Flags can also be used to filter values according to your specific needs.
      • For example, RSS provides rain-filtered SMAP SSS products.
    • Masks are used to omit data that do not meet established quality criteria
  • Error Sources and Uncertainty Estimation
    • Approaches to estimating errors or uncertainties vary among SSS products.

The specific application of corrections, flags, filters, and masks – along with approaches to estimating errors or uncertainties – are described in each product's User Guide and other literature.

Another consideration: Some data are processed by multiple institutions. SMAP SSS data, for example, are processed by both RSS and the NASA Jet Propulsion Laboratory (JPL). In the next section, we focus on similarities and differences in how these SMAP data products are processed.

Did We Mention it's Complicated?

A potentially confusing aspect is that there can be overlap between items in these categories. For example, RSS uses an ancillary "land mask" as an input to its SSS processing. Land corrections are applied to ensure the quality of coastal SSS data. RSS products also include land-related quality flags ranging from "light contamination" (i.e., removed from averaging in smoothed products) to "strong contamination" (i.e., masked altogether). Finally, RSS includes systematic errors from land contamination in its empirical uncertainty estimates.

Argo floats drift at depth then rise to the surface while measuring temperature and salinity
Figure 1. Argo floats drift at depth then rise to the surface while measuring temperature and salinity.
Saildrones - autonomous sailing drones - are used as a tool to provide high quality oceanic and atmospheric observations
Figure 2. Saildrones - autonomous sailing drones - are used as a tool to provide high quality oceanic and atmospheric observations.
Timeline of major release dates for three examples of SSS data products
Timeline of major release dates for three examples of SSS data products. Click image to enlarge.