Sensitivity of global ocean heat content from reanalyses to the atmospheric reanalysis forcing: A comparative study

The global ocean heat content evolution is a key component of the Earth's energy budget and can be consistently determined by ocean reanalyses that assimilate hydrographic profiles. This work investigates the impact of the atmospheric reanalysis forcing through a multiforcing ensemble ocean reanalysis, where the ensemble members are forced by five state-of-the-art atmospheric reanalyses during the meteorological satellite era (1979-2013). Data assimilation leads the ensemble to converge toward robust estimates of ocean warming rates and significantly reduces the spread (1.48 +/- 0.18 W/m(2), per unit area of the World Ocean); hence, the impact of the atmospheric forcing appears only marginal for the global heat content estimates in both upper and deeper oceans. A sensitivity assessment performed through realistic perturbation of the main sources of uncertainty in ocean reanalyses highlights that bias correction and preprocessing of in situ observations represent the most crucial component of the reanalysis, whose perturbation accounts for up to 60% of the ocean heat content anomaly variability in the pre-Argo period. Although these results may depend on the single reanalysis system used, they reveal useful information for the ocean observation community and for the optimal generation of perturbations in ocean ensemble systems.