Abstract
<jats:p>Southern Ocean (SO) sea surface temperatures (SSTs) have exhibited large-scale cooling over the satellite era (1979-present). Several potential drivers of this cooling have been proposed -- including changes in winds, freshwater fluxes, and natural variability -- but their relative contributions have been challenging to identify because climate models largely fail to simulate the observed trend. Moreover, observational estimates of SST exhibit substantial uncertainty. In this work we present novel ocean-only simulations individually forced by observationally-informed surface flux trends in heat, momentum, and freshwater from precipitation minus evaporation, ice sheet melt, and sea ice hydrology. We focus not only on SST changes, but also on changes in sea surface salinity (SSS) and subsurface temperature and salinity, because any proposed driver of SST change must be dynamically consistent with available, spatially-varying observations of all fields at once (within observational uncertainties). We show that the response to each surface flux change has a unique temperature and salinity fingerprint, and a linear combination of these responses captures key features of observed SO temperature and salinity trends. We conclude by performing a Monte Carlo analysis to refine this linear estimate, constraining with observed SST, SSS, and ocean heat content to create an estimate where ~80% of the SO area is within observational uncertainty for SST and ~70% for SSS. By considering salinity, regional changes, and observational uncertainty, we provide a holistic approach to understanding observed SO trends and suggest that known changes in winds and freshwater are sufficient to explain much of recent SO change.</jats:p>