Representation of Large-Scale Ocean Circulation in the Atlantic and Southern Ocean in Climate Model Simulations and Projected Changes under Increased Warming

Abstract

The global ocean acts as a mediator of Earth’s climate due to its role in the storage of heat and carbon. Presently, the ocean accounts for the storage of approximately 93% of the anthropogenic heat on our planet and ~27% of the anthropogenic CO2. Two regions in particular, the Southern Ocean and North Atlantic Ocean (SO, NA), act as gateways for the exchange of CO2 and heat between the atmosphere and the interior ocean. This is due to the unique water mass transformation processes that occur in these regions. Despite their disproportionate role in the climate system, large uncertainty exists with respect to understanding how the ocean circulation patterns and properties are projected to change in these regions throughout the 21st century. One pathway toward reducing projection uncertainty in these regions is to use modern observations and observational products to comprehensively diagnose, quantify, and improve upon mean state biases that exist in the climate simulations used to produce future climate projections. The work presented in this dissertation is a comprehensive analysis of the large-scale ocean circulation and properties in historical and 21st century simulations of large-ensembles of fully-coupled climate and Earth System Models contributed to multiple generations of the Coupled Model Intercomparison Project (CMIP). In the subtropical NA, a key region through which properties from the tropics are advected to the subpolar latitudes, the volume transports of the major flow regimes are reasonably represented in many CMIP5 models relative that observed by the Rapid Climate Change (RAPID) instrumental array at 26.5ºN. As the climate warms, all components of the total flow through the subtropical NA, with the exception of the wind-driven surface Ekman transport, are projected to weaken. Particularly, by applying the dynamical theory of Sverdrup balance, this work highlights the fact that the wind-driven NA subtropical gyre itself is projected to spin-down in response to a reduced wind stress curl over the subtropical latitudes. This spin-down, in conjunction with the reduced overturning at high-latitudes, acts as a source of significant additional weakening to the northward western boundary current flow in the upper ocean. In the SO, despite its dominant role in the oceanic uptake of anthropogenic carbon and heat relative to other basins, the large-scale circulation and properties have been poorly represented in climate models, resulting in low confidence ascribed to 21st century projections of the state of the SO. A comprehensive analysis of the simulation of the large-scale circulation and properties is presented for the Southern Ocean (SO) across thirty-one CMIP5 models. The main focus lies in building a framework to understand the major contributors to a model’s ability to represent the Antarctic Circumpolar Current (ACC) transport. Across the CMIP5 ensemble, the models fall into five different categories: 1) models that produce a reasonable ACC transport for approximately the right reasons, 2) models that accurately simulate key metrics, yet produce a too weak ACC, 3) models that simulate the wind stress forcing at the ocean surface accurately, but have errors in the density gradient, 4) models that simulate an accurate density gradient, but exhibit errors in the wind stress forcing, and 5) models that produce errors in all the metrics. Building on the framework presented in the CMIP5 study, a comprehensive assessment of the large-scale circulation and properties as simulated in the SO is performed across ensembles of models contributed to the past three CMIP generations (CMIP3-CMIP6). The CMIP6 models show improved representation of key observable-metrics in the SO including surface wind stress and wind stress curl, strength of the ACC, and meridional density gradients in the region of the ACC. However, some persistent biases have carried over into CMIP6 including an upper ocean that remains too fresh and too warm, significant warm biases at depth in several simulations, and a poor representation of Antarctic sea ice extent (SIE). These biases in observable metrics need to be considered when interpreting projected trends or biogeochemical properties in this region.