Monday, 24 January 2011: 2:00 PM
3B (Washington State Convention Center)
The response in the high-latitude Northern Hemisphere winter stratosphere to an increase in atmospheric CO2 depends strongly on the basic state of the control climate, which in turn is largely determined by the surface temperature. The majority of CO2-doubling or IPCC-like long-term climate change experiments undertaken with stratosphere-resolving GCMs prescribe monthly mean sea surface temperatures as a boundary condition; these are then interpolated linearly to give daily values. When surface temperatures are calculated interactively by a slab (or full) ocean coupled to the GCM, the atmosphere and ocean are in thermal balance and daily variability in the surface conditions is maintained. However, practical computational considerations generally preclude this approach. We investigate the effect of suppressing daily or monthly variability by undertaking five CO2-doubling experiments using a middle-atmosphere Chemistry-Climate Model (the IGCM-FASTOC). Each experiment consists of a control and a CO2-doubling simulation, run for 100 years (timeslice) each. Different combinations of land and sea surface temperatures are used: interactively generated by a coupled mixed-layer ocean and/or soil scheme; prescribed interannually varying; or prescribed climatological fixed. The surface temperature fields in all pairs of experiments are very similar by design, and are all equally realistic. However, they lead to different basic states of the atmosphere, which then respond in very different ways to CO2 forcing. When the sea surface temperature has no interannual variability, there is no significant dynamic climate change response in the polar stratosphere, regardless of land-surface temperature variability. A strong and statistically robust warming signal in the high-latitude lower and middle stratosphere is found only when the atmospheric GCM is coupled to a mixed-layer ocean.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner