The distribution of ocean salinity is influenced by factors such as ocean currents, freshwater exchanges with the atmosphere, continental inputs, and sea ice formation and melting. Salinity is a fundamental variable in the seawater’s equation of state, influencing density, ocean dynamics, and stratification. To understand its distribution and impact, salinity variability must be accurately estimated from in situ observations. In the 20 \(^\textrm{th}\) century, temperature observations were scarce, but salinity data were even more limited, as many instruments measured only temperature. In 2002, the implementation of the Argo observing system improved sampling and reduced the disparity between both variables. Salinity variability can now be better assessed from the 20-year Argo period. In this study, information from the Argo period and estimates of salinity covariability with temperature are used to reconstruct monthly subsurface salinity fields. The reconstruction uses in situ measurements and employs the data-driven RedAnDA scheme, which combines Data Assimilation, Analog Prediction, and Reduced-space Interpolation, a method previously validated for temperature reconstruction. The coupling of temperature and salinity is achieved using density-weighted three-dimensional empirical orthogonal functions. The method is applied to the tropical Pacific to evaluate its ability to reconstruct variability at monthly, interannual and decadal timescales, with emphasis on ENSO-related variability, which is the primary mode of variability in the region. Compared to Optimal Interpolation results, salinity analysis by RedAnDA shows significant improvements. The results yield a salinity product that provides valuable insights into historical salinity variability, including fresh pool displacements and salinity-driven stratification changes. Changes in subsurface salinity associated with El Niño and La Niña events are estimated from 1930 to 2001.
