There is a relatively high probability of encountering early warning signals when approaching a tipping point. However, this should not be confused with the probability that a tipping point is actually approaching if statistical early warning signals are observed. This is because there could be several alternative explanations for the presence of such early warning signals. In a Perspective article in the current issue of the scientific journal Nature Climate Change, a team of researchers from the Alfred Wegener Institute suggests taking a critical look at the evaluation of the statistical significance of these findings.
Tipping points are critical threshold values at which a small change in a parameter can lead to a system shifting from one stable state to another. When a system approaches a tipping point, statistical early warning signals can often be observed. Frequently cited climate tipping points are, for example, the collapse of ice sheets in the West Antarctic or the "tipping" of the Atlantic Meridional Overturning Circulation (AMOC). The AMOC is one of the most important components of the global thermohaline circulation, a huge conveyor belt in the ocean that circulates warm and cold water at different depths in the Atlantic. Changes in the fluctuations of the AMOC strength have been used in earlier studies as statistical early warning signals for the approaching tipping of the Atlantic overturning circulation.
Researchers from the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research (AWI) and Utrecht University have investigated the connection between statistical early warning signals and the prediciton of approaching climate tipping points in the scientific journal Nature Climate Change. "Studies that attempt to statistically predict climate tipping points using statistical early warning signals often confuse the probability that early warning signals occur in the observational data with the probability that a climate tipping point is actually approaching," reports Dr Vanessa Skiba from the AWI Potsdam. The confusion of these probabilities leads to a distorted assessment of statistical significance. Vanessa Skiba explains further: “Future studies must take this into account when determining the probability of an approaching climate tipping point. For this, we recommend using a Bayesian approach, in which the calculated probability of statistical early warning signals in the observational data is combined with the previously assumed probability of the climate tipping point under investigation. In addition, the Bayesian approach allows for the consideration of alternative mechanisms in the climate system that may lead to similar changes in the characteristics of climate tipping elements and spurious detection of early warning signals.”
Original publication:
Max Rietkerk, Vanessa Skiba, Els Weinans, Raphaël Hébert und Thomas Laepple: Ambiguity of early warning signals for climate tipping points, Nat. Clim. Change (2025). DOI: https://doi.org/10.1038/s41558-025-02328-8