"Early warning of climate tipping points" by Timothy M. Lenton, Nature Climate Change, 1 (2011) 201-209

Nature Climate Change, 1 (June 19, 2011) 210-209; doi: 10.1038/nclimate1143

Early warning of climate tipping points

Timothy M. Lenton


Abstract

A climate 'tipping point' occurs when a small change in forcing triggers a strongly nonlinear response in the internal dynamics of part of the climate system, qualitatively changing its future state. Human-induced climate change could push several large-scale 'tipping elements' past a tipping point. Candidates include irreversible melt of the Greenland ice sheet, dieback of the Amazon rainforest and shift of the West African monsoon. Recent assessments give an increased probability of future tipping events, and the corresponding impacts are estimated to be large, making them significant risks. Recent work shows that early warning of an approaching climate tipping point is possible in principle, and could have considerable value in reducing the risk that they pose.

Introduction

Elements of the climate system known as tipping elements¹ — which could pass a tipping point this century and undergo a qualitative change in state within this millennium — include the Atlantic thermohaline circulation (THC), West Antarctic ice sheet, Greenland ice sheet, Amazon rainforest, boreal forests, West African monsoon, Indian summer monsoon, and El Niño/Southern Oscillation (ENSO). Passing a tipping point (defined in Box 1) is typically viewed as a 'high-impact low-probability' event. However, recent (re)assessments give an increased likelihood of 'large-scale discontinuities' in the climate system under a given level of global warming², such that unmitigated climate change could result in some becoming 'high-impact high-probability' events^1, 3, demanding early warning capability⁴.

Early warning can take several forms, ranging from the knowledge that an event could occur, through qualitative assessment that it is becoming more likely, to a forecast of its timing. Recently, there has been growing interest in generic early warning signals⁵ for critical transitions in complex systems, especially slowing down⁶ as a bifurcation is approached. Furthermore, slowing down has been found in climate-model output^{7, 8, 9, 10, 11, 12} and palaeoclimate data^9, 10, 12 approaching abrupt transitions. This suggests probabilistic forecasting of some conceivable future climate tipping points may be feasible¹, especially if such statistical early warning indicators can be combined with dynamical modelling. However, critics have questioned the statistical robustness of proposed early warning signals¹³, and have noted that some types of abrupt transition carry no early warning signals^13, 14. These potential problems are not unique to climate tipping points.

For several rapid-onset natural hazards, for example, hurricanes¹⁵ and tsunamis¹⁶, quite sophisticated early warning systems are already in place¹⁷, whereas for some slower-onset hazards, for example, drought¹⁸ and malaria outbreaks¹⁹, seasonal climate-forecasting skill is beginning to be used in early warning. The United Nations has called for the development of a globally integrated early warning system for all natural hazards^20,21. This should include climate tipping points, because they present significant risks in themselves, and they will affect shorter-term hazards. A tipping point can be seen as a nonlinear shift in the shape or location of the frequency distribution of events that represent the climate, in the tails of which are extreme events. For example, tipping of the Greenland or West Antarctic ice sheets would accelerate sea-level rise, in turn increasing the impact of hurricane-driven storm surges or tsunamis. Dieback of the Amazon or boreal forests would cause increased wildfires. Disruption of the West African monsoon would affect drought in the Sahel.

If early warning can be achieved for climate tipping points, it could have considerable value for societies, as hinted at by the value of shorter-term, seasonal, climate forecasting to agriculture^22, 23, 24. For example, if El Niño increases in amplitude in a warming world, as some forecasts suggest²⁵, the resultant annual damages to the agricultural sector could exceed $1 billion (ref. 26), but these damages could be greatly reduced by effective response to seasonal early warning of El Niño²⁶. Even earlier warning of, for example, potential future El Niño 'regime shifts' would add further value to adaptation efforts.

Here, recent scientific progress on the early warning of climate tipping points is reviewed, noting that successful early warning systems would rely on social and technological factors as well as on scientific capability^17, 20, 21. In the first section, recent estimates of the proximity of climate tipping points are summarized, including a discussion of the structural weaknesses of present models. This leads to a focus on statistical methods of forecasting, as a complement to model-based approaches. In the second section, generic early warning indicators of approaching bifurcations are introduced and contrasted with the lack of forewarning of purely noise-induced transitions. The third section reviews recent tests of bifurcation early warning methods on climate model output and palaeoclimate data. The fourth section discusses the limitations of early warning methods, including false alarms and missed alarms. Finally, the research needed to improve scientific early warning capability and translate it into effective risk reduction is identified.

Full paper:  http://www.nature.com/nclimate/journal/v1/n4/full/nclimate1143.html