Hydrol. Earth Syst. Sci., 10, 903–912, 2006 www.hydrol-earth-syst-sci.net/10/903/2006/ Hydrology and © Author(s) 2006. This work is licensed Earth System under a Creative Commons License. Sciences Abrupt change in climate and climate models A. J. Pitman1 and R. J. Stouffer2 1Department of Physical Geography, Macquarie University, NSW, 2109 Australia 2Dynamics laboratory, Princeton, NJ, 08542, USA Received: 22 December 2005 – Published in Hydrol. Earth Syst. Sci. Discuss.: 19 July 2006 Revised: 9 November 2006 – Accepted: 10 November 2006 – Published: 28 November 2006 Abstract. First, we review the evidence that abrupt climate 1 Introduction changes have occurred in the past and then demonstrate that climate models have developing capacity to simulate many of If the Earth’s climate system was a linear system, with in- these changes. In particular, the processes by which changes puts and outputs proportional to each other, then modelling in the ocean circulation drive abrupt changes appear to be the climate system would be trivial and concerns over abrupt captured by climate models to a degree that is encourag- climate change due to increasing greenhouse gases in the at- ing. The evidence that past changes in the ocean have driven mosphere would be unwarranted. However, the Earth’s cli- abrupt change in terrestrial systems is also convincing, but mate system is dominated by a suite of non-linear phenom- these processes are only just beginning to be included in cli- ena (Rial et al., 2004) that make understanding the Earth’s mate models. Second, we explore the likelihood that climate climate, and how climate may evolve in the future, an enor- models can capture those abrupt changes in climate that may mous challenge. Some of these non-linearities are at the occur in the future due to the enhanced greenhouse effect. core of key components of the Earth System, such as the We note that existing evidence indicates that a major col- phase changes of water, the Clausius-Clapeyron equation, or lapse of the thermohaline circulation seems unlikely in the whether an organism is alive or dead. 21st century, although very recent evidence suggests that a The Earth’s climate system is effectively a closed system. weakening may already be underway. We have confidence Energy from the Sun is cycled through the various compo- that current climate models can capture a weakening, but a nents so that outputs of one component of the system be- collapse in the 21st century of the thermohaline circulation is come inputs for another, creating feedbacks. It is useful to not projected by climate models. Worrying evidence of insta- define “climate system” here: we do not mean the traditional bility in terrestrial carbon, from observations and modelling “mean state of the atmosphere” (e.g. Hann, 1908) or “aver- studies, is beginning to accumulate. Current climate mod- aged weather” (WMO, 1984). We use a more modern defini- els used by the Intergovernmental Panel on Climate Change tion of climate (e.g. Claussen, 2004) that includes the atmo- for the 4th Assessment Report do not include these terres- sphere, ocean, marine and terrestrial biosphere, cryosphere trial carbon processes. We therefore can not make statements and lithosphere and the interactions (the flows of mass, en- with any confidence regarding these changes. At present, ergy, momentum including biogeochemistry) between these the scale of the terrestrial carbon feedback is believed to be components. We define a threshold as a point beyond which small enough that it does not significantly affect projections the climate system responds to forcing in a nonlinear way of warming during the first half of the 21st century. How- and where the response is fast compared to the forcing (Al- ever, the uncertainties in how biological systems will respond ley et al., 2003). That is, over some time period the change to warming are sufficiently large to undermine confidence in in the response is much larger than the change in the forc- this belief and point us to areas requiring significant addi- ing. We will also use “abrupt” to describe a climate change tional work. that occurs when the climate system is forced to cross some threshold, triggering a transition to a new state at a rate deter- mined by the climate system itself and faster than the cause of the change (Alley et al., 2003). A smooth and gradual change in some determining quantity of the climate system Correspondence to: A. J. Pitman (e.g., radiation balance, land surface properties, sea ice, etc.) ([email protected]) can cause a variety of different responses depending on the Published by Copernicus GmbH on behalf of the European Geosciences Union. 904 A. J. Pitman and R. J. Stouffer: Abrupt change in climate and climate models nature of the system. If the system contains more than one and flows southward out of the Atlantic basin. Both paleo- stable equilibrium state then transitions to structurally differ- studies (e.g., Broecker, 1997, 2000) and modelling studies ent states are possible and we will highlight examples from (e.g., Manabe and Stouffer, 1988, 1997; Vellinga and Wood, the ocean and terrestrial systems. Upon the crossing of a bi- 2002) suggest that disruptions in the THC can produce abrupt furcation point the evolution of the system is no longer con- climate changes. Some modelling studies (Rahmstorf, 1995; trolled by the time scale of the forcing, but rather determined Tziperman, 1997; Rind et al., 2001) suggest that there are by its internal dynamics. In these definitions, the magnitude thresholds where the THC may suddenly weaken or collapse of the forcing and the response to the forcing as well as the causing abrupt climate changes. time scale are important. If non-linear change, thresholds The palaeoclimate community has established, over sev- and feedbacks are crucial to the Earth’s climate we should be eral decades, a convincing case for the existence of non- able to identify clear examples in the observational record. linear and abrupt climate changes centred on the stability We demonstrate that this is indeed the case and show that to of the THC. Data from sediments, tree rings, speleothems, explain the observational record we need to use non-linear ice cores and corals show large, widespread, abrupt climate processes, thresholds and feedbacks. It is therefore reason- changes have occurred repeatedly throughout the past glacial able to believe that these will also be key to explaining those interval (see Rahmstorf, 2002). The most dramatic of these changes that could occur in the future. abrupt climate changes are the warm Dansgaard-Oeschger The tool to project how the Earth’s climate will change events, characterised by a warming in Greenland by 8 to in the future is the climate model (McAvaney et al., 2001). 16◦C within about a decade (Huber et al., 2006) and the Climate models are mathematical representations of com- cold Heinrich events where cooling occurred over the cen- ponents of the Earth’s climate system. Until recently, cli- tury time-scale, but the warming that ended them took place mate models tended to focus on the atmosphere and oceans within decades (Cortijo et al., 1997; Voelker, 2002). While with relatively little attention to the cryosphere or terrestrial the strongest impact of Dansgaard-Oeschger and Heinrich systems and biogeochemical cycles were rarely represented events were centred on the North Atlantic, significant change (McGuffie and Henderson-Sellers, 2001). More recently, cli- also occurred in tropical wetlands (Chappellaz et al., 1993; mate models (also called Earth System Models to identify Brook et al., 2000) and in the Asian monsoon region (Wang their increasingly broad scope) have substantially increased et al., 2001). their investment in cryospheric, biospheric and biogeochem- There is now good evidence of a link between these abrupt ical processes and are increasingly reliable tools for large- surface climate changes to aspects of the ocean circulation scale climate projection (McAvaney et al., 2001). To be con- (Clark et al., 2002). Heinrich events are likely caused by ice- fident in the capacity of climate models it would be impor- sheet instability (Hemming, 2004) and the exceedence of in- tant to show that they can simulate those changes, thresholds ternal thresholds that maintain ice-sheet mass. The resulting and feedbacks that are important in the climate system. We iceberg discharge provides an influx of fresh water sufficient present evidence that climate models can capture elements of to shut down or significantly reduce deepwater formation in non-linear change thresholds and feedbacks and then point to the North Atlantic. Cold events at the end of the last ice age some key areas where non-linear change is possible in the fu- were likely caused by meltwater from land-based ice sheets ture. (Teller et al., 2002). It is important to note that these ma- This paper therefore provides an overview of the evidence jor changes in ocean circulation, due to an increased influx for abrupt change in the Earth’s climate system. We high- of fresh water, indicate either that the ocean circulation hap- light those processes that are important to climate change pened to be close to a threshold (Ganopolski and Rahmstorf, mainly on time scales of a century or so (Rial, 2004, discuss 2001) or is typically sensitive to changes in fresh water forc- abrupt climate change on millennial timescales). Our focus ing. is to establish the capacity of climate models to capture these Climate model studies have been performed in which fresh processes as a basis for assessing our confidence in climate water discharge from ice sheet instability (Heinrich event) models as a tool for projecting future climates. or a meltwater release (e.g.