PICE Talk by Ruth Champan

Title: Stochastic data-adapted Atlantic Meridional Overturning Circulation box models

Abstract:
In this seminar I will present the most interesting results of my PhD thesis, the abstract of which is below.

The Atlantic Meridional Overturning Circulation (AMOC) forms an important part of the global ocean circulation which transports heat and salt around the globe  (thermohaline circulation). Many models across the model hierarchy suggest that the AMOC can display bi-stability and could undergo an abrupt transition from the current strong circulation (‘on’), to an alternative collapsed state (‘off’). A collapsed AMOC would have regional and global impacts, such as changes to precipitation patterns and European temperature distribution. Observational data shows that the AMOC has weakened this century. In this thesis we utilise a process-based, data-adapted ocean box model which has been calibrated to a global circulation model (GCM), HadGEM3. We demonstrate that this model can undergo tipping under different freshwater forcing profiles via three different mechanisms: bifurcation-, rate-, and noise-induced tipping.
Noise-induced tipping occurs when a random perturbation causes the model to unexpectedly transition to the alternative stable state, without any other critical threshold being passed. We estimate noise amplitudes from a selection of CMIP6 unforced simulation timeseries. We find that the internal variability estimated from the GCMs is small and does not lead to substantial probability of AMOC collapse unless combined with some freshwater forcing. We suggest that GCMs underestimate Atlantic Ocean variability, and do not reflect a ‘real-world’ scenario. Therefore, noise-induced effects should not be ruled out. Finally, we estimate the quasipotential for a reduced AMOC box model and study transition paths. We find that not all paths go via the saddle, as would be expected in the no-noise case.
We highlight the need for improved modelling of decadal ocean variability in GCMs so that tipping risks can be more thoroughly quantified.