2.5
Uncertainties in climate system properties and anthropogenic aerosol forcings based on climate change detection methods
Chris E. Forest, MIT, Cambridge, MA; and M. R. Allen, P. H. Stone, and A. P. Sokolov
We present a method for estimating uncertainties in climate predictions by comparing the modeled response to prescribed forcings over the 20th century against climate observations for that period. To determine these constraints, we use the MIT 2D climate model (Sokolov and Stone,1998) in conjunction with results from the Hadley Centre's coupled atmosphere-ocean general circulation model (A-OGCM), HadCM2. The MIT 2D model is a zonally-averaged version of a 3D GCM and was shown to accurately reproduce the global-mean transient response of any coupled A-OGCM through appropriate choices of the cloud feedback and the effective rate of diffusion of heat into the deep ocean. Vertical patterns of zonal mean temperature change through the troposphere and lower stratosphere also compare favorably with those generated by 3-D GCMs. We compare the height-latitude pattern of temperature changes as simulated by the MIT 2D model with observed changes, using climate change detection diagnostics, which yield an objective measure of model-observation goodness-of-fit (via the noise-weighted residual sum of squares, see Allen and Tett (1999)). For purposes of comparison, the forcings and fingerprint were chosen to match those used in Allen and Tett (1999), which follows Tett et al. (1996). The MIT model permits one to systematically vary global model parameters and determine how the goodness-of-fit with observations depends on these factors. Three such parameters are the cloud feedback (i.e. the model's climate sensitivity), rate of mixing of heat into the deep ocean, and the net direct radiative forcing by anthropogenic aerosols. This would not be possible with a coupled A-OGCM, owing to the computational requirements and lack of structural flexibility. Hence, we varied climate sensitivity from 0.4 K to 6.2 K, global-mean ocean diffusivity from 0.0 cm2/s to 160.0 cm2/s, and the anthropogenic aerosol forcing from -1.5 to +0.5 W/m2.
For a fixed aerosol forcing at -0.5 W/m2, two sets of model parameters are rejected, with a chosen confidence level, as being inconsistent when the model response is compared with observations. The first corresponds to high climate sensitivity and low heat uptake by the deep ocean; the second corresponds to climate sensitivity less than ~1 K for all values of ocean heat uptake. When the aerosol radiative cooling is increased, these rejection regions retain their shape but shift to compensate for the decrease in the net radiative forcing. Thus, to obtain matching temperature changes, either a higher climate sensitivity or a reduced ocean heat uptake are required for the same changes in greenhouse gas concentrations. When we turn off the anthropogenic aerosol forcing, we still find that simulations are not inconsistent with observed temperature changes. Although high climate sensitivities combined with low ocean heat uptake can now be rejected more strongly, we still reject regions with climate sensitivity less than ~1 K for all values of ocean heat uptake. This contribution provides an efficient framework for interpreting detection and attribution results, and it estimates quantitative uncertainty bounds for both physical properties and forcings of the climate system that are important for future climate predictions.
Session 2, IPCC TAR: Long-term Climate Variability and Change: Part 1 (Parallel with Session 3)
Monday, 10 January 2000, 1:30 PM-3:00 PM
Previous paper Next paper