Empirical reconstruction of salinity from temperature profiles with phenomenological constraints

Abstract. The problem of estimating the salinity when only temperature profiles are available is having an increasing interest mainly because of multi-parametric data assimilation in ocean forecasting models. In this paper, a new method based on the introduction of a correction factor for salinity deduced from recent measurements is proposed to calculate salinity from temperature profiles and climatological datasets. It is supposed that the seawater potential density in a specific area, as deduced from climatological monthly averaged temperature and salinity values, does not change. A certain but small variability on its values is admitted and estimated combining the uncertainty of temperature and salinity in-situ measurements, and the diurnal variation, as obtained from a set of recent CTD and MedArgo measurements in Tyrrhenian Sea. Then, the deduced range of variability for salinity and potential density is imposed to synthetic values, which are compared with CTD and XCTD data. Finally, this technique is used to calculate salinity profiles from XBT temperature profiles from Ligurian and Tyrrhenian Sea (XBT probes monthly dropped along the transect Genova-Palermo, within the Mediterranean Forecasting System-Toward Experimental Prediction project). Results are analysed and discussed.