Quantifying uncertainty in spatio-temporal changes of upper-ocean heat content estimates: an internationally coordinated comparison

TL;DR

This study compares different methods of estimating upper-ocean heat content changes to understand uncertainties and guide future improvements in global ocean heat content assessments.

Contribution

It provides an international comparison of mapping and bias correction methods, highlighting key sources of uncertainty in upper-ocean heat content estimates.

Findings

Mapping method uncertainty dominates over bias correction at global scale.

Uncertainty is largest in Indian Ocean and eddy-rich regions.

Spread in estimates increased since the 1990s.

Abstract

The Earth system is accumulating energy due to human-induced activities. More than 90 percent of this energy has been stored in the ocean as heat since 1970, with about 64 percent of that in the upper 700 m. Differences in upper ocean heat content anomaly (OHCA) estimates, however, exist. Here, we evaluate spread in upper OHCA estimates arising from choices in instrumental bias corrections and mapping methods, in addition to the effect of using a common ocean mask. The same dataset was mapped by six research groups for 1970 to 2008, with six instrumental bias corrections applied to expendable bathythermograph (XBT) data. We find that use of a common ocean mask may impact estimation of global OHCA by +- 13 percent. Uncertainty due to mapping method dominates over XBT bias correction at a global scale and is largest in the Indian Ocean and in the eddy-rich and frontal regions of all basins. Uncertainty due to XBT bias correction is largest in the Pacific Ocean within 30N to 30S. In both mapping and XBT cases, spread is higher since the 1990s. Important differences in spatial trends among mapping methods are found in the well-observed Northwest Atlantic and the poorly-observed Southern Ocean. Although our results cannot identify the best mapping or bias correction schemes, they identify where and when greater uncertainties exist, and so where further refinements may yield the largest improvements. Our results highlight the need for a future international coordination to evaluate performance of existing mapping methods.