At a “tipping point”, a small additional amount of greenhouse gases can trigger a sudden and large alteration in the climate. Such abrupt climate alterations are a special risk of climate change, because neither society nor ecosystems can adapt to such too-rapid changes. And our climate models are also not precise enough to enable us to estimate the probability of tipping points. Therefore studies are being undertaken as to whether certain early-warning signals, that might warn us of abrupt climate changes, could be detected from observations of the climate. The theory of dynamic systems indicates that this is possible in principle. The more a stable equilibrium is destabilized, the more susceptible the system becomes to perturbations of external origin. This leads to longer and larger anomalies – the variance and autocorrelation increase.

But in practice, this simple theory is not easy to apply. For example, climate change is already so rapid that it would be difficult to obtain a sufficiently long time series before the tipping point is reached. And early-warning signals are at best an extrapolation of current changes, and not a genuine prediction. Furthermore, there are numerous mechanisms that permit abrupt alterations in climate without the existence of early-warning signals. And on the other hand, early-warning signals can occur without any abrupt alteration in climate following, since the variance and autocorrelation are influenced by numerous processes that are independent of tipping points. So early-warning signals are not specific enough to permit definite conclusions. Without specific background knowledge of the underlying processes, they are not truly informative. For this reason, I am studying in detail whether and how the theory of early-warning signals is applicable to climate changes in particular regions, such as the origins of the Sahara, or the expected loss of sea ice in the Arctic in the future.

Early-warning signals of an abrupt decrease in sea-ice?

Together with colleagues, I have investigated whether an abrupt loss of sea ice can be predicted on the basis of observed changes in the variance and the time correlation between successive years (autocorrelation). This would offer the possibility of predicting a possibly abrupt loss of sea ice without the use of models, and would be an independent source of informative about which future scenario of sea-ice loss is the most plausible.

In contrast to this promising prospect from very idealized dynamic systems, we found no increase, but rather a robust decrease of variance and autocorrelation of the volume of sea ice before the loss of summer sea ice, regardless of how abruptly the loss of sea ice occurred in any sea-ice models. This is because thinner ice can adapt to disturbances more rapidly. Afterwards, the ice’s reaction time – and thus also the autocorrelation – increases, because the ice’s reaction time is determined by the large thermal capacity of the sea water during the longer and longer ice-free season. These changes are not affected by the characteristics and origin of the climate variability in climate models, and do not depend on whether the loss of sea ice in the Arctic occurs abruptly or irreversibly in a model. We also show that the current climate change in the Arctic is too rapid for significant changes in variability to be detected in time. Based on these results, the prospect of recognizing early-warning signals before an abrupt loss of sea ice at a tipping point appears to be very limited.

On the other hand, the results enable us to determine the variability of sea-ice properties in various climates better. Measurements are not sufficient for this, because only short series of observations are available, and the fluctuations of ice coverage in past climates are little known. Since all climate models indicate a reliable connection between the #mean and the variability of the properties of the sea ice, and this can be understood with the help of fundamental principles, one can conclude the variability of the sea ice in the past and future indirectly by means of the theoretical results – from a globe entirely covered by ice to a completely ice-free Arctic.

Publications

  • Bathiany, S., van der Bolt, B., Williamson, M. S., Lenton, T. M., Scheffer, M., van Nes, E. H. & Notz, D., 2016: Statistical indicators of Arctic sea ice stability – prospects and limitations, Cryosphere, 10, 1631-1645.

http://www.the-cryosphere.net/10/1631/2016/tc-10-1631-2016.pdf

  • Bathiany, S., Dijkstra, H., Crucifix, M., Dakos, V., Brovkin, V., Williamson, M. S., Lenton, T. M. & Scheffer, M.: Beyond bifurcation – using complex models to understand and predict abrupt climate change, Dyn. Stat. Clim. Sys., in press

http://climatesystem.oxfordjournals.org/content/early/2016/11/22/climsys.dzw004.

  • Williamson, M. S., Bathiany, S. & Lenton, T. M., 2016: Early warning signals of tipping points in periodically forced systems, Earth Syst. Dynam., 7, 313-326.

http://www.earth-syst-dynam.net/7/313/2016/

  • Bathiany, S., Claussen, M. & Fraedrich, K., 2013: Detecting hotspots of atmosphere–vegetation interaction via slowing down – Part 2: Application to a global climate model, Earth Syst. Dynam., 4, 79-93.

http://www.earth-syst-dynam.net/4/79/2013/

  • Bathiany, S., Claussen, M. & Fraedrich, K., 2013: Detecting hotspots of atmosphere–vegetation interaction via slowing down – Part 1: A stochastic approach. Earth Syst. Dynam., 4, 63-78.

http://www.earth-syst-dynam.net/4/63/2013/esd-4-63-2013.html