U51B-01 INVITED
Climate Sensitivity
Discussion of climate sensitivity requires careful definition of forcings, feedbacks and response times, indeed, foggy definitions have produced flawed assessments of climate sensitivity. The best information available on climate sensitivity comes from insightful interpretation of the Earth's history aided by quantitative information from climate models and understanding of climate processes. Climate sensitivity is a strong function of time scale, in part because of the nature of climate feedbacks. Unfortunately for humanity, the preponderance of feedbacks on the century time scale appears to be positive. The chief implication is the need for a sharp reversal in the trend of human-made climate forcing, if we are to avoid creating a planet that is dramatically different than the one on which civilization developed.
U51B-02 INVITED
Probabilistic inference for future climate change
The experiment we are currently performing on the Earth system is intrinsically non-repeatable, so the concept of "reliable probabilities" over a large ensemble of predictions cannot apply. Therefore, prediction of future climate change is strongly Bayesian in a way that numerical weather prediction is not: probabilistic predictions are necessarily a matter of informed belief conditioned on our interpretations of the data and models around us. The climate sensitivity (S) in response to anthropogenic and other forcings has long been one of the dominant uncertainties in predicting future climate change. Many observationally-based estimates have been presented in recent years, with many of them disagreeing with the original Charney/IPCC estimates, for instance by assigning substantial probability to extremely high sensitivity, such as P (S > 6C ) > 5%. However, there is now an abundance of observational evidence available to us covering a wide range of time scales and processes. The broad consistency of these data (and success of explicit predictions) suggests that we should rather have increased confidence in our scientific understanding. We will show that explicit probabilistic analysis supports this standpoint, and a much greater confidence in a moderate value for S is easily justified, with climate sensitivity very unlikely to be as high as 4.5C.
U51B-03 INVITED
State dependence of climate sensitivity and its implications
It has long been noted (Senior and Mitchell, 2000; Boer and Yu, 2004) that the climate sensitivity of global climate models evolves over time, changing by up to +/-20% in long runs following stabilization of greenhouse forcing. We explore the magnitude and origins of this effect in the ensemble of global coupled models contributed to the IPCC AR4 ensemble. Two possible parameterizations of this effect are discussed, the first in which the net feedback parameter, λ, depends on the degree to which global temperature is out of equilibrium and the second in which λ simply depends simply on global temperature itself. We show that the second parameterization provides a generally better fit to the GCMs' output and explore its implications for recent attempts to constrain long-term climate sensitivity using recent observed climate change. The possibility of an uncertain temperature- dependence of λ has even more profound implications for efforts to constrain climate sensitivity using the last glacial maximum, but this is not the focus of this talk. The possibility of state-dependent feedbacks is a further nail in the coffin of attempts to provide a useful, objectively constrained, upper bound on the long-term equilibrium response to stabilization of atmospheric greenhouse gases at any currently conceivable level. We note, however, that this does not preclude our use of observations to constrain equally policy-relevant quantities, such as the familiar transient climate response, or TCR, the equivalent carbon dioxide concentration consistent with a two degree warming, or C2K, or the transient warming commitment, or TWC. The TWC is the maximum warming we expect to result from a linear ramp increase in radiative forcing followed by a linear decrease: at current values of the atmospheric airborne fraction for carbon dioxide, this is approximately equivalent to the forcing that would result from a complete cessation of emissions at any given time. Although this is not a feasible emissions scenario, it represents a much more natural (and better constrained) measure of the warming attributable to greenhouse emissions to date than the more familiar stabilization commitment scenario. The support of the modeling groups contributing to the WCRP CMIP-3 model intercomparison is gratefully acknowledged.
U51B-04 INVITED
Estimating the sensitivity of the climate system to CO2 forcing from observed climate change
Observations of climate change over the 20th century and over recent paleoclimatic periods have been used to estimate the equilibrium climate sensitivity, which is the global mean equilibrium warming in response to a doubling of CO2 in the atmosphere. This talk compares the evidence and the uncertainties in different lines of evidence, including 20th century warming, Northern Hemispheric climate changes over the last millennium and data for the Last Glacial maximum. While most lines of evidence render very small climate responses or negative feedbacks unlikely, constraining the equilibrium climate sensitivity at the upper range remains difficult. However, the available information, taken together, yields very high climate change less likely than each individual line of evidence suggests. However, when using multiple lines of evidence of climate change, important questions arise about the independence of information used, and if the feedbacks that are sampled in the observed climate change fully span the feedbacks important to drive the climate systems response to CO2 doubling. The talk concludes with a discussion of limitations of equilibrium climate sensitivity, and ideas how to move beyond it.
U51B-05
Linking glacial and future climate through ensembles of GCM simulations
Relatively few GCMs have been integrated for both doubled carbon dioxide (2xCO2) and Last Glacial Maximum (LGM) conditions, making it hard to quantify any relationship between these two climates in the models. Here we use ensembles of the MIROC3.2 GCM with varied parameters, whose members all agree reasonably well with present day climatology. By running the ensembles for a variety of experiments, including LGM and 2xCO2 we can explore some of the uncertainties in the relationship between these two climates on global and regional scales. The MIROC3.2 model shows an asymmetry in its response to changes in greenhouse gases, with 80% of the ensemble having a weaker cooling for low CO2 than warming for high CO2. This asymmetry would suggest that direct estimates of climate sensitivity from the LGM based on first-order energy balance considerations could be underestimated by the order of half a degree. Additional analysis of the differences in climate sensitivity at LGM and pre-industrial climate states is now underway and first results will be presented. Regionally, our results lend support to the idea that Antarctic is a useful place to look for historical data which can be used to validate models used for climate forecasting of future greenhouse gas induced climate changes, at regional and global scales. Good results may also be obtainable using tropical temperatures, particularly those over the ocean. Our result for Greenland is not so strong, possibly due to difficulties in accurately modelling the sea ice extent.
U51B-06
Low Climate Sensitivity to Atmospheric CO2 during the Ice Ages
Ice core records from Antarctica show that atmospheric CO2 has varied between 180 and 280 ppm during the ice ages of the last 500,000 years. The ice cores also show that there is a strong correlation between the temperature of the air over Antarctica and atmospheric CO2 in which the air temperature jumps up by 8-10 deg C as the pCO2 increases at the end of each cold glacial stage. So, what are the ice cores trying to tell us about the climate system's sensitivity to CO2? The point of this presentation is that the ice cores are telling us more about local changes around Antarctica than global sensitivity. In this regard, the two hemispheres seem to be doing different things: the Southern Hemisphere is responding primarily to an internal mechanism with a 100,000-yr period that is focused on Antarctica; the ice sheets in the Northern Hemisphere are responding to the Milankovitch forcing at 23,000 and 41,000 years. d18O records derived from foraminiferal tests incorporate both signals. If the combined record is viewed simply as "ice volume" then one would erroneously conclude that the global climate is varying more strongly with CO2 than it really is. The sawtooth-shaped cycles in the ice cores are a response to an internal feedback that alters the stratification and circulation of the ocean and the winds around Antarctica. CO2 comes out of the ocean when the stratification breaks down and the surface temperature warms (Toggweiler et al., Paleoceanography, 21(2), 2005PA001154, 2006). Thus, there are two components to the warming recorded in the ice cores, one associated with local SST and wind changes and one associated with the greenhouse effect from CO2. It would seem that most of the 8-10 deg. warming, perhaps 6 or 7 deg C, is due to the SST and wind changes. The remainder, 3 deg C or less, is the greenhouse response. The small remainder puts an upper limit on the greenhouse response for the planet as a whole.
U51B-07
Synchronization of Glacial-Interglacial SST changes in the Tropical Oceans: Evidence for Amplification of Climate Sensitivity by Carbon Dioxide Over the Past 2.7 Ma
Understanding the links between tropical sea surface temperatures (SST) and changes in high latitude climate across ice age cycles has challenged paleoclimatologists for some time. On the one hand, the tropical oceans should be shielded from processes that produce large temperature sensitivity in the high latitudes (ice-albedo feedback, sea-ice feedback, steering of wind fields by ice sheets, etc.). However, the high latitudes exert an influence on tropical ocean temperatures through the thermocline, and, perhaps via the atmosphere by modulating the CO2 greenhouse effect. The tropics, in turn could provide important feedbacks to glacial cycles via their influence on water vapor and clouds. Work with recently developed SST proxies (Mg/Ca, alkenone Uk'37) has proven that substantial sensitivity exists in tropical SST on a glacial-interglacial timescale over the course of the last ~ 1Ma. Over the extent of the EPICA ice core record, this sensitivity is coherent with glacial-interglacial variations in CO2. In this study, we have extended the tropical SST record to the mid-Pliocene (3.5 Ma) in tropical locations in the Atlantic, Indian and Pacific oceans. We use the alkenone unsaturation index, recorded at ~ 3kyr resolution at each of three sites, in conjunction with benthic foraminiferal δ18O measured at the same sites, which synchronizes the records and allows for a direct comparison of tropical SST to ice volume changes. We find that all three tropical sites show long-term cooling over the last three million years, although the rate of cooling is the steepest in the eastern equatorial Pacific. Glacial-interglacial SST variation is strikingly similar in the three tropical basins after 2.7 Ma, with a succession of 41 kyr (obliquity) cycles persisting up to the transition to 100 kyr cycles at about 800 ka. Prior to 2.7 Ma, temperatures at the three sites behave idiosyncratically on the orbital timescale. A further intriguing observation is the existence of coherent 300-500 kyr cycles in SST in all three basins (also observed in Mg/Ca data) that may not be related to orbital forcing. We evaluate and reject the case that the tropical SST cycles originate dominantly from upwelling/thermocline- driven processes, and argue instead that the SST data suggest that a coherent, substantial CO2-glaciation feedback began at about 2.7 Ma, synchronizing to first order the tropical SST variations on orbital timescales in late Pliocene and all of Pleistocene time. The key event at 2.7 Ma, therefore was the phase locking of CO2 and ice/sea ice feedbacks to orbital forcing. Furthermore, the large tropical SST memory observed (the 300-500 kyr wavelength) argues for the importance of a long response time element in the climate system, most likely ocean biogeochemical cycles and CO2, in modulating tropical SST on supra-orbital timescales. Prior to 2.7 Ma, the combination of glacial and CO2 feedbacks may have been much weaker, diminishing the similarity of tropical SST patterns between the different ocean basins, and lessening their sensitivity to high latitude processes.
U51B-08 INVITED
Geologic Constraints on Climate Sensitivity
A firm understanding of the relationship between CO2 and temperature is critical for interpreting past climate change and for making predictions of future climate change. A recent synthesis of data spanning the Last Glacial Maximum to present-day suggests a 5-95 % likelihood range of 1.5 to 6.2 °C warming for each doubling of CO2 (ΔT(2x)). However, some evidence is not consistent with this range; further, CO2 concentrations and global temperatures have been similar to or lower than the present-day for most of the last several hundred thousand years, and so this past time interval also may not represent the best analog for predicting globally-warm future climates. In an effort to reduce the uncertainty of ΔT(2x), we turned to the Phanerozoic record, an interval that includes times when the Earth was both colder and substantially warmer than the present-day. The GEOCARB and GEOCARBSULF long-term carbon cycle models have been used to calculate multi-million-year patterns of Phanerozoic CO2. A critical factor in this approach is the effect of atmospheric CO2 level on the rate of CO2 uptake by calcium and magnesium silicate mineral weathering. A rise in temperature, accompanying a rise in CO2, increases the rate of silicate weathering, which in turn accelerates atmospheric CO2 consumption, forming a negative feedback loop. Using the logarithmic relation between temperature change and CO2, we examined how different values of ΔT(2x) affect calculated Phanerozoic CO2 levels for best estimates, and physically reasonable ranges, of all other factors affecting CO2 in the long-term carbon cycle. We then compared the model calculations against an independent proxy dataset for atmospheric CO2 that spans the past 420 Myr. Our results that best fit the proxy data are broadly consistent with most climate model calculations and greatly restrict the possibility of a weak radiative forcing by CO2 (ΔT(2x) < 1.5 °C) over multi-million year timescales. A ΔT(2x) of at least 1.5 °C has likely been a robust feature of the Earth system for the last 420 Myr, independent of temporal scaling.