Runaway tipping points of no return

I wonder if any else has noticed that we appear to have crossed a threshold in the usage of the phrase ‘tipping point’ in discussions of climate? We went from a time when it was never used, to a point (of no return?) where it is used in almost 100% of articles on the subject. Someone should come up with a name for this phenomenon….

Regardless of the recent linguistic trends, the concept has been around for a long time. The idea is that in many non-linear systems (of which the climate is certainly one), a small push away from one state only has small effects at first but at some ‘tipping point’ the system can flip and go rapidly into another state. This is fundamentally tied to the existence of positive feedbacks and is sometimes related to the concept of multiple ‘attractors’ (i.e. at any time two different ‘states’ could be possible and near a transition the system can flip very quickly from one to another). Another ‘tipping point’ in non-linear systems occurs when as some parameter varies, the current attractor changes character or disappears. However it is currently being used interchangeably a number of potentially confusing ways and so I thought I’d try and make it a little clearer.

Positive feedback

A positive feedback occurs when a change in one component of the climate occurs, leading to other changes that eventually “feeds back” on the original change to amplify it. The classic ones in climate are the ice-albedo feedback (melting ice reduces the reflectivity of the surface, leading to more solar absorption, more warming and hence more melting) and the water vapour feedback (as air temperatures rise, water vapour amounts increase, and due to the greenhouse effect of the vapour, this leads to more warming), but there are lots of other examples. Of course, there are plenty of negative feedbacks as well (the increase in long wave radiation as temperatures rise or the reduction in atmospheric poleward heat flux as the equator-to-pole gradient decreases) and these (in the end) are dominant (having kept Earth’s climate somewhere between boiling and freezing for about 4.5 billion years and counting). But it is the postive feedbacks that make weather chaotic and climate interesting.

People often conclude that the existence of positive feedbacks must imply ‘runaway’ effects i.e. the system spiralling out of control. However, while positive feedbacks are obviously necessary for such an effect, they do not by any means force that to happen. Even in simple systems, small positive feedbacks can lead to stable situations as long as the ‘gain’ factor is less than one (i.e. for every initial change in the quantity, the feedback change is less than the original one). A simple example leads to a geometric series for instance; i.e. if an initial change to a parameter is D, and the feedback results in an additional rD then the final change will be the sum of D+rD+r²D…etc. ). This series converges if |r|<1, and diverges (‘runs away’) otherwise. You can think of the Earth’s climate (unlike Venus’) as having an ‘r‘ less than one, i.e. no ‘runaway’ effects, but plenty of positive feedbacks.

Tipping points

So are there ‘tipping points’ in climate? One way to assess that is by looking for elements of the physical system where we think that there is a threshold behaviour. Two frequently discussed examples are the overturning circulation in the North Atlantic and the summer sea ice in the Arctic. In both of these cases, the existence of these phenomena can be disrupted in models (and there is evidence of similar behaviour in the real world) by small changes in freshwater and increasing polar amplification, respectively. At some point, both could simply cease to be viable. But we are not very confident of where these points are or how sensitive the threshold is. These are examples of ‘known unknowns’.

There is also the existence of ‘unknown unknowns’ – tipping points that we are as yet unaware of. An example of this kind of surprise happened in relation to the Antarctic ozone hole, where unexpected chemistry on surfaces of ice particles lead to much more efficient destruction of ozone in the polar vortex than had been expected, making an existing concern into a serious problem. By their nature, we are not able to assess how important any such surprises might be, but it is impossible to rule them out entirely.

By far the most common examples of tipping points though are in relation to ecosystems. The extremely complex web of interdependencies that keep ecosystems dynamic and healthy give rise to plenty of potential thresholds and it is extremely difficult to predict consequences of external changes. The myriad influences on the health of ecosystems (habitat loss, logging, urbanization, species introduction etc. as well as climate change) means that it is most likely here that the tipping point concept will be most applicable. Examples such as a rise in minimum winter temperatures that allow a new insect species to gain a foothold in a new ecosystem (pine bark beetles in Alaska), or warming that leads to movement upward in altitude of ecosystem zones that end up reducing the area of existing alpine biomes. As the planet warms, it is easy to imagine an increasing number of ‘tipping points’ being passed, each related to some different sub-system of the climate or biosphere.

Points of no return

Are ‘tipping points’ the same as the ‘points of no return’ oft used in the media? For a species that becomes extinct as a result of crossing a threshold, the answer is obviously yes. But in the physical climate system, are there genii that can’t be put back in the bottle? This is really a question of time scale. Changes to aerosol concentrations can be reversed in a few weeks after an emission change. CO₂ levels however are much slower to change and are already very unlikely to revert to pre-industrial values in any scenario over the next few hundred years. In this minimal sense the climate is already past the point of no return compared to pre-industrial climate.

The ‘known’ physical tipping points described above have natural timescales that determine whether ‘returns’ are possible. The Arctic sea ice, for instance, has timescales of around 5 years to a decade, and so a collapse of summer ice cover could conceivably be reversed in a ‘cooling world’ after only a decade or so (interactions with the Arctic ocean stratification may make that take a little longer though). Model simulations of the thermohaline circulation indicate that for small perturbations, recovery can occur in a few decades. For larger perturbations (i.e. complete collapses) intermediate-complexity models suggest that in some regimes these changes can be quasi-permanent, although this behaviour has not yet been fully explored in current state-of-the-art GCMs. The clues from the paleo-record indicate that there is likely a bi-modal spectrum of overturning states in glacial climates, but there is no evidence of such multiple steady states in the Holocene. Thus there is no strong reason to think either of these ‘tipping points’ are really irreversible – though that is not to imply that the process of loss and recovery wouldn’t have significant impacts.

The big ‘point of no return’ though is usually associated with the melting of the ice sheets – in particular, Greenland and the West Antarctic Ice Sheet (WAIS). Currently the ice sheets exist in part because they already exist i.e. the reason it snows on Greenland is in some large part because there is a large ice sheet there. Should the ice sheet start to melt in a serious way (i.e. much more significantly than current indications suggest), then lowering of the elevation of the ice sheet will induce more melting simply because of the effect of the lapse rate (air being warmer closer to sea level due to pressure effects). Thus if Greenland disappeared, it is unlikely that it would grow back even under current climate, let alone in a warmer world. So loss of either of these ice sheets would indeed be an effect with ‘no return’, at least on any reasonable human timescale.

10 years?

Jim Hansen was widely quoted earlier this year stating that there were likely only 10 years left in which serious actions could be taken to prevent ‘dangerous anthropogenic interference’ on climate occurring in the future. He described this as a ‘tipping point’, but it should be clear that he was not using the term in exactly the same way as I defined above. He very specifically was not indicating that some irreversibly large change in climate would happen in 10 years. Instead he was pointing to the trajectory of increasing CO₂ emissions that continue to add to atmospheric concentrations. Actual and projected emission levels are already at the high end of Hansen’s ‘alternative scenario’ which was suggested as an achievable outcome (based on significant control efforts) that kept forcings (including Co₂, CH₄ and black carbon) below a level that Hansen considered would be ‘dangerous’ (specifically a level that would avoid the melting of any significant fraction of the WAIS or Greenland ice sheet). It is the inertia of societal infrastructure, the carbon cycle and the climate that implies that at any point there is a significant warming that is already ‘in the pipeline’ (and thus very difficult to avoid). We have estimated this at about 0.5 C. Hansen’s statement can therefore be read as a comment on a ‘point of no return’ of the human-climate system, rather than the climate system in a purely physical sense.

The ’10 year’ horizon is the point by which serious efforts will need to have started to move the trajectory of concentrations away from business-as-usual towards the alternative scenario if the ultimate warming is to stay below ‘dangerous levels’. Is it realistic timescale? That is very difficult to judge. Wrapped up in the ’10 year’ horizon are considerations of continued emission growth, climate sensitivity, assumptions about future volcanic eruptions and solar activity etc. What is clear is that uncontrolled emissions will very soon put us in range of temperatures that have been unseen since the Eemian/Stage 5e period (about 120,000 years ago) when temperatures may have been a degree or so warmer than now but where sea level was 4 to 6m higher (see this recent discussion the possible sensitivities of the ice sheets to warming and the large uncertainties involved). In 10 years time CO₂ levels will likely be greater than 400 ppm and the additional forcing combined with the inertia of the system will be make it increasingly unlikely that we will avoid a further 1 deg C or more warming. While the ’10 years’ shouldn’t be read as an exact timetable, it is surely in the right ballpark. 30 more years of business-as-usual will make it impossible to keep temperatures from rising beyond Eemian levels (see here for some discussion of stabilisation scenarios), and decisions (on infrastructure, power stations, R&D, etc.) that are being made now will determine the emissions for decades to come.

One point or many?

Much of the discussion about tipping points, like the discussion about ‘dangerous interference’ with climate often implicitly assumes that there is just ‘a’ point at which things tip and become ‘dangerous’. This can lead to two seemingly opposite, and erroneous, conclusions – that nothing will happen until we reach the ‘point’ and conversely, that once we’ve reached it, there will be nothing that can be done about it. i.e. it promotes both a cavalier and fatalistic outlook. However, it seems more appropriate to view the system as having multiple tipping points and thresholds that range in importance and scale from the smallest ecosystem to the size of the planet. As the system is forced into new configurations more and more of those points are likely to be passed, but some of those points are more globally serious than others. An appreciation of that subtlety may be useful when reading some of the worst coverage on the topic.