Understanding Interdecadal Changes in the Ocean Carbon Sink (UNICORNS)

The oceans have absorbed more than a quarter of the carbon dioxide (CO2) that humans release to the atmosphere, a process that substantially slows the pace of climate change. However, this ocean "sink" for CO2 is not constant: observational estimates based on surface ocean measurements suggest it changed little through the 1990s, but after 2000 it began to increase rapidly. The reasons for this variability are not understood. Earth system models for the climate and carbon cycle do not capture the variation, indicating fundamental deficiencies in the models. This is a major problem because the models are our tools to project how the climate will change in the future. Changes in the physical circulation of the ocean, or its biogeochemistry, may be responsible for the observed variability - or perhaps sparse observations in earlier decades have biased observations and exaggerated the variability.

In UNICORNS we will bring several newly developed techniques to resolve these questions. (1) We will apply several machine learning methods to observations of the ocean interior, to reconstruct the ocean carbon content in greater detail than previously possible. This will enable us to test whether the time history of the carbon inventory as revealed by the interior observations, is consistent with that deduced from surface measurements. This will provide an independent test of the apparent sink variability. (2) To examine possible ocean circulation mechanisms that could cause the variability, we will adapt and apply a novel "inverse" technique that constructs budgets for temperature and salinity within water masses. This can explicitly derive regional mixing and transport, changes that could lead to the variable carbon sink. (3) Much of the sink variability may be attributable to changes in circulation redistributing "natural" carbon that has been in the ocean since pre-industrial times. We will apply a framework that explicitly identifies added and redistributed carbon to model output to examine this hypothesis and to evaluate the results of our inverse method. Different techniques used in the literature define the split between "anthropogenic" and natural, pre-existing carbon, in different ways and with this part of our project we will aim to bring more clarity to this distinction.

Our results will enable a more assured interpretation of the global carbon budget over recent decades, improvements to carbon-climate models, and more confident projections of future climate.