Hjem
Geofysisk institutt

GFI/BCCR Seminar: Two cases for a higher climate sensitivity

Hovedinnhold

Trude Storelvmo (Yale University, New Haven, CT, US):

Two cases for a higher climate sensitivity

Abstract
Earth’s climate sensitivity, i.e. the global mean surface temperature increase for a given atmospheric CO2 increase, remains an elusive quantity, and arguably has come to represent the “holy grail” of climate science. The lack of progress on this issue can partly be attributed to the difficulty of deducing the climate sensitivity to CO2 based on observations. Such efforts have been hampered by the fact that aerosol particles, which have a cooling effect on climate, have been increasing along with CO2, and therefore “masked” some unknown proportion of CO2-induced warming to date. Representing the cooling effect of aerosol particles in global climate models (GCMs) has proven notoriously challenging, and GCM estimates of aerosol cooling continue to diverge.

Realizing that insights from the field of econometrics could be of value to this particular problem, we recently published a study in Nature Geoscience, which takes advantage of statistical methods that are commonly applied on economic time series but largely unknown to the climate science community. Applying these methods on a rich observational data set of climate variables, we found that ~1/3 of the CO2 warming of continents to date has likely been masked by aerosol cooling. Studies not accounting for this cooling would falsely conclude that climate is less sensitive to CO2 than it really is. The study therefore supports climate sensitivities at the upper end of the range published, for example by the last report from the Intergovernmental Panel on Climate Change (IPCC).

A second and very different study began with the realization that most GCMs underestimate the amount of supercooled liquid in clouds compared to satellite data that has recently become available. We followed up with experiments in which we forced a GCM to reproduce the observed cloud phase, and tested the implications for the simulated CO2-induced warming. Intriguingly, we were able to establish an intimate relationship between the representation of cloud phase and the simulated climate sensitivity – bringing cloud phase into agreement with satellite observations rendered the model much more sensitive to CO2 increases. It turned out that because the GCM simulated too much ice in mixed-phase clouds, it exaggerated the cloud phase changes induced by a warming atmosphere. This in turn caused the clouds to brighten and reflect much more sunlight back to space. The latter had a strong cooling effect on climate that ended up damping much of the initial CO2 warming (a negative feedback). Because most GCMs exhibit a similar cloud phase bias, we expect all GCMs to exaggerate this negative feedback, and thus underestimate climate sensitivity. These findings were recently published in Science, and further strengthen the above argument for a climate sensitivity at the upper end of the IPCC range, but here by taking a completely different research approach.