Assessment of the Aquarius Space-borne Sea Surface Salinity Retrievals in Polar Ocean
[18-Dec-2014] Dinnat, E., Brucker, L. and Álvarez, I.O.C.
Presented at the 2014 AGU Fall Meeting
Ocean salinity and temperature drive the thermohaline circulation and play a key role in the ocean-atmosphere coupling. With the availability of passive L-band (1.4 GHz) space-borne observations, Sea Surface Salinity (SSS) can be monitored globally at weekly time scales. SSS in the polar regions may be used to better constrain deep water formation, and to monitor changes due to freshening by the melting cryosphere. However, SSS remote sensing in the polar oceans is challenging because L-band radiometric observations are less sensitive to salinity in cold waters, SSS retrieval is less accurate for very rough seas and the presence of sea ice, icebergs and land in the radiometer field of view adds complexity to the retrieval process.
Aquarius is a NASA space-borne instrument operating three L-band radiometers. While Aquarius SSS retrievals are performed with a good accuracy in tropical and mid-latitude oceans, a thorough assessment has not been performed in the colder waters of the polar oceans. To assess Aquarius data at high latitudes, we compare them to in-situ measurements from ship cruises. In the northern hemisphere, cruises between Denmark and Greenland are used. In the south, we use cruises in the Austral Ocean between Tasmania and Antarctica. These quality-controlled shipborne measurements (more extensive than the Argo profiling floats which are rare at high latitudes) allow us to assess Aquarius SSS over long transects, repeated weekly or monthly, over the three year period during which Aquarius has been operating. Our results show that significant contamination of SSS retrievals by ice and land are observed, despite the correction for land contamination applied in the Aquarius retrieval algorithm. Such long track comparisons with ship data will help refine the future versions of Aquarius space-borne products. Nonetheless, excluding the data contaminated by land or ice, the agreement with ship data is good. Specifically, the standard deviation is ~0.3-0.6 psu on SSS averaged over about 10 days, and a low bias of the order of ~0.25 psu is identified. Moreover, a good agreement of local SSS patterns can be identified between both ship- and space-borne measurements, such as an increase in salinity signal south of Tasmania, or the strong salinity gradient between southeast and southwest Greenland.