Table Of Contents

This Page

Foreseeing tipping pointsΒΆ

The title is taken from a Nature News and Views that describes briefly a recent method to predict tipping points in a complex system. The idea is that the frequency spectrum of perturbations differ when the system is far or close to a tipping point. For instance, a system takes more time to respond to external perturbations when it is close to a tipping point than far. In conclusion, the measured perturbations in the system have a larger lag auto-correlation –in other words, the system has more “memory”. Another consequence is the perturbations have a larger amplitude. These characteristics can a priori be observed way before the sytem actually shifts to another state and can be used as an early warning system.

See the Nature’s article in question (Drake and Griffen 2010) as well as the references given in this note (Held and Kleinen 2004, Dakos et al. 2008, Lenton et al. 2009, and the review paper by Scheffer et al. 2009).

From the review paper, I see van Nes and Scheffer (2007) and Lenton et al. (2008) as potentially interesting pieces.

However, recently, Ditlevsen and Johnsen (GRL, 2010) have cautioned the interpretation of climate jump in terms of a bifurcation. They first point out that both the variance and the auto-correlation need to increase; one is not enough to signal a bifurcation. Second of all, using a new high-resolution time series of these climate shifts, they find no significant increase in either of these two quantities suggesting that the shifts may be cause by large noise and may be unpredictable. These results contradict in particular the work of Dakos et al. (2008).