Climate Variability and Response in High-Resolution Earth System Models

Abstract

In times of anthropogenic climate change, refining our understanding of the climate system is crucial. Earth System Models can simulate all climate subsystems and their interactions under a variety of scenarios; they are among the most important tools of climate science. The ocean is a vital component of the climate system as it covers the majority of the Earth's surface and constitutes the largest heat, water, and carbon reservoirs. Equivalent to atmospheric high and low pressure systems, the ocean exhibits turbulence which consists of dynamic current filaments and ring-like structures, so-called mesoscale eddies, with typical length scales of 10--100~km. These mesoscale features influence the large-scale ocean mean state, its variability, and its response to forcing. A fine enough ocean model grid is needed to resolve the ocean mesoscale; this is computationally expensive and only with the increased computing power of the last years has it become feasible to perform century-long climate model simulations with strongly-eddying oceans. In this thesis, we investigate the effects of mesoscale turbulence on the large spatial and long timescale ocean and climate state.

One key question of climate science is how the climate varies internally allowing us to distinguish human-caused changes in the climate. One of the many mechanisms that lead to variability is chaotic mesoscale turbulence which is associated with timescales of days to months but can affect much slower variability on decadal to multidecadal timescales. By investigating sea surface temperature patterns of multidecadal variability, we find enhanced multidecadal variability when mesoscale turbulence is simulated. As modern climate models do not generally employ eddying ocean components, multidecadal variability may be systematically underestimated in these simulations which are widely used for climate change projections. Further, we investigate the mechanisms of one particular mode of multidecadal variability, the Southern Ocean Mode, by looking at the mechanical energy balances involved. This mode does not appear in the non-eddying model and the energetics suggest a crucial role for mesoscale eddies and their interaction with the time-average flow field, suggesting that this form of variability is only possible in strongly-eddying ocean models.

Another key question is how the climate will change under a greenhouse gas emission scenario because it allows us to set meaningful climate change mitigation targets. We compare multiple aspects of the climate response of the strongly-eddying and non-eddying climate models. While we find numerous local differences, globally-integrated measures are similar between the simulations and do not immediately suggest systematic differences. However, some climate-subsystems, such as the Atlantic Meridional Overturning Circulation (AMOC), are susceptible to "tipping point" behavior. One indicator of whether the AMOC could quickly and irreversibly weaken is the salinity distribution in the Southern Atlantic which is why we investigate changes to the Atlantic's freshwater budget. Our results suggest that, in contrast to the non-eddying model simulation, the AMOC in the high-resolution simulation may become unstable under a high-emission scenario. This is primarily due to reduced salinity biases suggesting that non-eddying models may give a false impression of the stability of the AMOC.