Estimating salinity to complement observed temperature: 2.Northwestern Atlantic

This paper addresses the problem of estimating salinity for a large region in the Atlantic Ocean containing the Gulf Stream and its recirculation. Together with Part 1 [Thacker, W.C., 2007-this issue. Estimating salinity to complement observed temperature: 1. Gulf of Mexico. Journal of Marine Systems. http://dx.doi.org/10.1016/j.jmarsys.2005.06.008.] dealing with the Gulf of Mexico, this reports on the first efforts of a project for developing world-wide capability for estimating salinity to complement expendable-bathythermograph (XBT) data. Such estimates are particularly important for this region, where the strong frontal contrasts render the task of assimilating XBT data into numerical models more sensitive to the treatment of salinity.Differences in salinity's co-variability with temperature and with longitude, latitude, and day-of-year from the northwestern part of the region with the Gulf Stream to the southeastern part more characteristic of the Sargasso sea suggested that the region be partitioned to achieve more accurate salinity estimates. In general, accuracies were better in the southeastern sub-region than in the more highly variable northwestern sub-region with root-mean-square estimation errors of 0.15 psu at 25 dbar and 0.02 psu at 300 dbar as compared with 0.35 psu and 0.50 psu, respectively, but in the southeast there was an unexpected error maximum around 1000 dbar where estimates were slightly less accurate than in the northwest. For pressures greater than 1400 dbar root-mean-square errors in both sub-regions were less than 0.02 psu.