Noise-induced phenomena in the large-scale ocean circulation

Abstract

Each research question in climate science requires an appropriate climate model to be formulated, where specific approximations are made and a certain number of physical processes is considered. Conceptual climate models represent only fundamental processes and they are particularly useful when trying to capture the essence of a certain system. Moreover, when two well-separated time scales can be identified in the system, the processes characterised by the smaller time scale are parametrised, often using a stochastic approach. This is the case, for example, when the ocean circulation is forced by high-frequency atmospheric fluctuations. This thesis focuses on the effect of atmospheric noise on the large-scale ocean variability, in particular concerning sea surface height (SSH) variability and the stability of the Atlantic Meridional Overturning Circulation (AMOC). Both problems are addressed by using low-dimensional stochastic dynamical systems. The effect of high-frequency wind-stress variations can be represented as a correlated additive and multiplicative noise (CAM) stochastic model of sea-level variations. The one-dimensional model developed allows to formulate an appropriate null hypothesis for SSH variability: given a time series of sea-level anomalies, specific phenomena in the ocean circulation can be detected by analysing peaks in its power spectrum, and the significance of such peaks can be tested against the null hypothesis. In other words, the model can be used to attribute specific sea-level variability to other effects than wind-stress noise. From the power spectrum analysis of several time series of SSH anomalies in the ocean, we can conclude that the CAM noise process under investigation can explain most of the variability of the sea level. Moreover, we found that some peaks are significant under the traditional red-noise test, but not significant under the CAM noise test. This indicates that using the incorrect test may lead to erroneous attribution of phenomena in the sea-level variability. Atmospheric noise is not only responsible for the existence of a background signal in the ocean time series. Noise, indeed, can lead to the occurrence of tipping points. One of these systems is constituted by the AMOC, which is the zonally integrated volume transport generated by a complex system of currents in the Atlantic. It represents a crucial component of the climate system, as it redistributes heat northward in the Atlantic. Thanks to the presence of the AMOC, north-western Europe experiences a relatively mild climate, compared to maritime regions at similar latitudes on the Pacific (up to 6°C warmer). Whether the AMOC is in a multistable regime, i.e. if the system can be in two different states under the same external forcing, is under debate in the climate community. However, if the present-day AMOC were in a multiple equilibria regime, a potential collapse in the near future would cause large and rapid changes in the global climate system. The possibility of occurrence of an abrupt change induced by high-frequency atmospheric variability (i.e. noise) is especially alarming, as noise-induced transitions are inherently unpredictable. In this thesis, we calculate the probability of a noise-induced collapse of the AMOC, given different climate scenarios, by means of a rare-event numerical algorithm. We also suggest some improvements to this algorithm, which make it a promising tool to investigate noise-induced transitions in multistable systems. Lastly, by studying transition probabilities of noise-induced partial collapses of the AMOC in an ensemble of CMIP5 climate models, we revisit one of the stability indicators of the AMOC, i.e. the freshwater transport carried by the overturning circulation at the southern boundary of the Atlantic basin. A correction to this indicator, based on the transition probabilities, is suggested to measure whether an AMOC state is in a multiple equilibrium regime or not.