Meetings: Documents

Evaluation of Sea Surface Salinity Observed by Aquarius and SMOS
[17-Apr-13] Abe, H. and Ebuchi, N.
Presented at the 2013 SMOS-Aquarius Science Workshop
Sea surface salinity (SSS) derived from the Aquarius and SMOS missions were validated using in-situ salinity data focusing on physical process around the sea surface. The Aquarius and SMOS SSSs were collocated with in-situ observations from Argo floats and offshore moored buoys and outputs from ocean optimal interpolation (OI) system and operational ocean assimilation system. In this abstract, results from the comparisons with observations by Argo floats and outputs from the OI system are reported. The Aquarius SSS Levels 2 and 3 data product version 1.3.7 released by the NASA PO.DAAC at JPL and the SMOS SSS Level 3 data product version 300 released by the SMOS Barcelona Expert Centre were analyzed. For the Level 2 SSS, salinity profiles of Argo floats released by the Global Assembly Data Center (shallower than 12.5 dbar) were used for a period from August 25 2011 to November 30 2012. Spatial and temporal separations were limited less than 200 km and 24 hours, respectively. For the Level 3 SSS, monthly gridded salinity field released by Japan Agency for Marine-Earth Science and Technology (JAMSTEC) were used for the same period. This gridded SSS data are constructed by an optimal interpolation method using salinity profiles of Argo floats and hydrographic reports with resolutions of 1° x 1° and 1 month. The salinity data in the shallowest level (10 dbar) were used to compare with the Level 3 SSS. The Aquarius Level 2 SSS generally agreed well with the Argo salinity in the low and mid latitude. The standard deviation of SSS residual was 0.54 psu at 45°S - 45°N. To the contrary, the Aquarius Level 2 SSS was highly deviated in the high latitude with a value of the standard deviation of 0.96 psu at 60°S - 45°S. Since the SSS estimation is based on the intensity of radiation emitted from the sea surface, it is expected that the SSS estimation depends on oceanic conditions on the sea surface such as sea surface temperature (SST), wind speed and wind direction. Large deviations of the SSS residual under low SST condition were detected, as expected from the low sensitivity of the brightness temperature to salinity at low SST. In addition, the SSS residual showed large values under high wind speed condition because the roughness correction did not work well under severe weather conditions. Dependence of the SSS residual on the azimuth angle between the wind direction and the sensor looking direction was not significant. The standard deviation of the SSS was expressed as a function of the SST and wind speed. The deviation was large (>1.0 psu) under low SST and high wind speed conditions (<5 °C and >15 m/s), while it was considerably smaller (<0.5 psu) under high SST and low wind speed conditions (>20 °C and <5 m/s). These results correspond to the large deviation in the high latitude.
In the analysis of Level 3 data, spatial patterns of Aquarius and SMOS SSSs agreed well with those of the outputs from JAMSTEC OI system. Spatial patterns of the SSS residual (satellite SSS minus OI SSS) for Aquarius were also qualitatively consistent with those for SMOS. Bias and standard deviation of the SSS residual at 45°S - 45°N for Aquarius V1.3.7 were -0.04 psu and 0.35 psu, respectively, which were smaller than those of the SMOS SSS (-0.23 psu in bias and 0.64 psu in standard deviation). Temporal variations in the bias with a period of several months and amplitude more than 0.1 psu were found in the SMOS SSS and Aquarius V1.3, while those were significantly reduced in the Aquarius V1.3.7. Negative bias (Aquarius SSS < OI SSS) was found in the tropical oceans through a year and along western boundary currents in the North Hemisphere in wintertime. The former correlated with distribution of precipitation in the Intertropical Convergence Zones and is considered to be due to the salinity stratification beneath the sea surface. However, the latter did not show any correlation with distributions of precipitation or evaporation.

View Document (PDF, 5.5 MB)