Increased AMOC slowdown in high-resolution models
The AMOC — colloquially often referred to as the "Gulf Stream" — transports warm surface water from the tropics to the high latitudes (Labrador Sea, North Atlantic) and flows back south as cold deep water after cooling and the associated sinking. Fluctuations of the AMOC have a significant influence on the northward heat transport in the ocean and thus on the climate in Europe and the North Atlantic. The sea surface temperatures determine climate phenomena such as droughts in the Sahel or the frequency of hurricanes in the Atlantic.
How the AMOC and the associated climate phenomena can be influenced by global warming has so far only been investigated using climate models in which important small-scale processes, such as the ocean eddies, are not resolved and are only represented by parameterization. To find out whether these small-scale processes influence the climate system's response to increasing greenhouse gas concentrations, climate change simulations with higher spatial resolution must be carried out. Previously, such simulations were not possible because an increase in resolution led to an unrealistic decline of the AMOC. The paper by the MPI-M scientists examines the decline with the model configuration used in the EU project PRIMAVERA.
Most previous model studies assumed that the decline in the AMOC intensity is determined by a change in deep convection. This happens in particular as a result of local fresh water inflows and, to some extent, by melting glaciers and icebergs. Furthermore, it was previously assumed that wind effects were not relevant for a decline in the AMOC. However, the scientists find out in their study that the wind plays an indirect but decisive role in the decline. This is because the wind over the North Atlantic is weaker at a higher resolution of the atmosphere than at a lower resolution. This in turn leads to a weaker wind-driven ocean gyre and consequently to a reduced transport of salt water into the Labrador Sea, which promotes the decline of the AMOC.
Method
The authors of the paper use sensitivity studies to identify the various factors (fluxes) that contribute to the strong decline in AMOC. They analyze the freshwater flux, the buoyancy (heat and fresh water) flux and the influence of wind stress on the water surface. The paper investigates the sensitivity of the AMOC to changes in these surface fluxes, which change at a higher atmospheric resolution. The costs of running the high-resolution models are extremely high, making it difficult to conduct many high-resolution sensitivity studies. For this reason, the scientists used a special calculation concept for their study to carry out the sensitivity studies at considerably lower computational costs. Model configurations of the MPI-ESM — consisting of the global atmosphere model ECHAM6, the land vegetation model JSBACH and the ocean and sea-ice model MPIOM — are used for the sensitivity studies. These are also used in the framework of the EU project PRIMAVERA for the project HighResMIP (High Resolution Model Intercomparison Project) of the Coupled Model Intercomparison Project Phase 6 (CMIP6).
The high-resolution simulations are performed with the configurations HR (high resolution) and XR (extra high resolution). HR and XR only differ in the resolution of the atmosphere, which is 1° in HR and 0.5° in XR. The ocean is resolved at 0.4° in both configurations, i.e. the ocean model is the same in HR and XR. As a reference for the further sensitivity runs with changed variables, control runs over 100 and 150 years were carried out with HR and XR. In order to identify the role of individual fluxes, the scientists carried out sensitivity experiments with flux corrections. While such flux corrections were previously used to compensate for model errors and avoid model drift, in this study they are used to mimic the behavior of the higher-resolution XR model in a cost-effective and time-saving manner in the lower-resolution HR configuration. Therefore, the fluxes between atmosphere and ocean in the HR model are replaced by those of the respective XR model.
Results
The AMOC in the MPI-ESM is primarily determined by the deep convection and salinity differences in the Labrador Sea. In HR the volume flow is at 17 Sv (Sv = Sverdrup = 106 cubic meters per second). At higher resolution, XR, the value drops to 9 Sv, i.e. by almost 50 percent. This strong decline is due to the formation of sea ice and the complete stop of subpolar deep convection. At identical ocean conditions, this can only have been caused by different fluxes of fresh water, heat and momentum in HR and XR.
Therefore, the freshwater flux, buoyancy flux and the influence of wind above the water surface were specifically investigated in order to separate the buoyancy flux from the dynamic wind effects over the North Atlantic and in the AMOC area, and to find out which flux is the most important for the AMOC decline. The salinity at the sea surface, the temperature above the North Atlantic, the extent of sea ice and the deep convection are almost identical in the HR and XR runs. However, the dynamic wind effects determine freshwater input in the XR runs, leading to comparatively weak deep convection and a stronger AMOC decline. In the wind stress studies, the AMOC declines more in the high-resolution runs (XR) than in the buoyancy studies. A weaker wind stress means less advection of warm and saline water from the subtropical gyre, which is crucial for cooling and increasing freshwater content in the subpolar gyre.
The long-standing question about the strong AMOC decline in high-resolution MPI-ESM was answered with the sensitivity studies. The new calculation concept can in this form also be used for other sensitivity studies at higher resolutions. The lower computational costs are a technical benefit for the modelling community that runs high-resolution models.
Publication:
Putrasahan, D.A., K. Lohmann, J.-S. v. Storch, J.H. Jungclaus, O. Gutjahr and H. Haak (2019) Surface flux drivers for the slowdown of the Atlantic Meridional Overturning Circulation in a high-resolution global coupled climate model. Journal of Advances in Modeling Earth Systems (JAMES), doi: 10.1029/2018MS001447
Contact
Dr. Dian Putrasahan
Max Planck Institute for Meteorology
Phone: +49 40 41173 468
Email: dian.putrasahan@mpimet.mpg.de