There are two main uncertainties in determining future climate: the
trajectories of future emissions of greenhouse gases and aerosols, and
the response of the global climate system to any given set of future
emissions (Meehl et al., 2007). These uncertainties are normally
explored through the use of global climate models, which provide
information at relatively coarse spatial resolutions. The application
of regional climate models that have been nested in the global models
presents another source of uncertainty, that associated with the
spatial scale of the simulations. In this program we focus on
understanding the uncertainty inherent to the global models and the
regional models, and do not directly concern ourselves with the
uncertainty related to the future emissions scenarios.
While we use 4 different AOGCMs to drive 6 different regional
models, we do not simulate all 24 possible combinations, but rather
only 12, sampling the full 4x6 matrix using a fractional factorial
design. Each RCM uses boundary conditions from half of the global
models, and each global model provides boundary conditions to half of
the RCMs:
|
ECPC |
HRM3 |
MM5I |
RCM3 |
CRCM |
WRFP |
GFDL |
X |
X |
|
X |
|
|
HADCM3 |
X |
X |
X |
|
|
|
CGCM3 |
|
|
|
X |
X |
X |
CCSM |
|
|
X |
|
X |
X |
While the 4 AOGCMs used form a relatively small subset of the total
set of 23 that produced climate change simulations for the Forth
Assessment Report of the IPCC, we are able to place these four
simulations of climate change in the context of the full suite of
global climate models using probabilistic statistical techniques
(e.g., see Tebaldi et al., 2004; 2005). Then the regional model
simulations, which essentially branch off from the AOGCMs used to
drive the RCMs, can also be placed in the probabilistic context. More
details on the development of the probabilistic methods will be
forthcoming as they are developed. The 12 scenarios themselves provide
a simple and straightforward measure of uncertainty that is still
potentially very useful. For example, in a climate impacts study, all
12 scenarios could be used in an impact model (e.g., a water resource
or crop model) to determine the potential range of impacts. The
eventual availability of probability distributions for the full suite
of experiments will also prove useful for policy analysis, impacts
analysis and so forth.
References
Meehl, G. et al., 2007: Global Climate Projections. In Solomon et
al. (eds.), Climate Change 2007: The Physical Science Basis.
Working Group I Contribution to the Fourth Assessment Report of the
IPCC. Cambridge University Press: Cambridge, 996 pgs.
Tebaldi, C., R. Smith, D. Nychka, and L. O. Mearns, 2005:
Quantifying uncertainty in projections of regional climate change: A
Bayesian approach to the analysis of multi-model ensembles.
J. Climate 18:1524-1540.
Tebaldi, C., L. O. Mearns, R. Smith, D. Nychka, 2004: Regional
probabilities of precipitation change: A Bayesian approach.
Geophys. Res. Lett. 31:L24213, doi:10.1029/2004GL021276.
|