CoAuthors

Peter Gleckler (Program for Climate Model Diagnosis and Intercomparison, Lawrence Livermore National Laboratory, USA); Felix Landerer (Jet Propulsion Laboratory, California Institute of Technology, USA); Karl Taylor (Program for Climate Model Diagnosis and Intercomparison, Lawrence Livermore National Laboratory, USA)

The global ocean is responsible for storing more than 90% of the heat associated with observed greenhouse-gas-attributed warming. Here, we use multiple observational estimates and a large suite of climate models to quantify how estimates of global upper-ocean warming since 1970 are likely biased low. This underestimation is attributed to poor observational coverage of the Southern Hemisphere, and also to analysis method limitations that estimate temperature changes in data-sparse regions too conservatively.

Existing in-situ observations of upper ocean warming place only 35-50% of the global ocean warming since 1970 in the Southern Hemisphere, although the Southern Hemisphere accounts for more than 60% of the upper ocean volume. Ocean warming drives thermosteric sea level rise. However, we find that the inter-hemispheric warming differences cannot be reconciled with simulations and observations of large-scale sea surface height, even when accounting for non-thermosteric sea level components in both hemispheres. This partitioning of Northern and Southern Hemispheric sea surface height changes simulated by models is consistent with the highly accurate, near global altimeter observations, whereas the simulated hemispheric partitioning of upper-ocean warming is inconsistent with the poorly constrained temperature observations.

Using the correspondence between hemispheric-scale ocean heat content and steric changes, we therefore adjust the poorly constrained observed Southern Hemisphere warming to yield a hemispheric ratio consistent with the modeled range. The method assumes that the comparatively well-sampled Northern Hemisphere upper-ocean warming is accurately estimated, which we infer from better observational agreement. Our ratio-based scaling adjustments imply that global ocean heat content change estimates are biased low by 10-24%, equivalent to additional global heat content increases of 0.9-2.8 x 10e22 J since 1970. These results represent approximately double the net heat gain for all non-ocean heat reservoirs (terrestrial, cryospheric and atmospheric) combined, and have important implications for sea level, energy budget and climate sensitivity assessments.