… Les séminaires ITAP …
Mardi 23 octobre 2012 à 16h, Salle Occident (Irstea, bâtiment Minéa)
Bernard De Saedeleer (Centre de recherche sur la Terre et le Climat, Institut Georges Lemaître, Université Catholique de Louvain, Belgique)
Titre : Roving of an astronomical synchronisation: do we live on a single stable paleoclimate history ?
Abstract : The mystery of ice ages induced by a varying incoming solar radiation has drawn ceaseless attention for several decades. We investigate here in depth the synchronisation properties of a simple paleoclimatic toy model (a van der Pol-like relaxation oscillator) for which reasonable qualitative agreement with ice volume proxies is easily found.
The model is first used in a deterministic framework. We discovered that the astronomical forcing may synchronise the climate to several coexisting climatic attracting trajectories, contrary to the uniqueness hypothesis of [Tziperman et al, “Consequences of pacing the Pleistocene 100 kyr ice ages by nonlinear phase locking to Milankovitch forcing”, Paleoceanography, 21:PA4206, 2006]. We also found that temporary desynchronisations may occur due to loss of local stability (positive largest local Lyapunov exponent), leading to divergence of climatic orbits for some period of time. As the attracting trajectories can sometimes lie quite close to the boundary of their basins of attraction, additional disturbances may then cause a jump to another climate history over the last millions years of the Pleistocene, reducing the predictability of the timing of the glacial inceptions and terminations. Such a conclusion is of major importance, because of its potential impact on the overall theory of ice ages.
A stochastic version of the model is then used in order to assess the conditions for such a jump to occur. Extensive Monte Carlo experiments are performed and reveal that the jumps occur preferentially at specific times or locations in the phase space, for a given level of noise. These statistical experiments successfully confirm previous theoretical findings.
Our approach relies on several concepts and tools borrowed mostly from dynamical system theory (like attractors, non autonomous systems, basins of attraction, global and local stabilities, largest Lyapunov exponents, quasiperiodicity, multistability) but also from modern mathematical theories like the pullback attractor, and from statistical analysis.
Finally, the scope of this study is extended by mentioning several ongoing research collaborations.