Geoscience Reference
In-Depth Information
climate models are used to drive macro-scale hydrology models (
<
1km
resolution) to provide information, such as stream-flow, surface runoff, or
subsurface water storage, needed for management of irrigation, flood control,
hydropower production, municipal and industrial supply, navigation, and
recreation.
9.2.6 Model intercomparison and validation
Given the large uncertainties in GCMs, it is clear that results from a single
model cannot be interpreted too literally and that an estimate of the model
reliability has to be included. This consideration has led to the Atmospheric
Model Intercomparison Project (AMIP), which was initiated in 1989 under
the auspices of the WCRP with the aim to systematically validate, diagnose,
and intercompare the performance of AGCMs in various simulated aspects
of the climate system (Gates et al.
1992
,
1999
). During AMIP-I, over 30
AGCMs around the world were organized to carry out simulation of the
evolution of the Earth climate from 1979 to 1988, subject to identical
prescribed observed monthly sea surface temperature, sea ice, CO
2
concen-
tration and solar constant. A large number of model output variables are
archived and standardized and made available to the scientific community.
Thanks to the AMIP climate model, users have gained a better appreciation of
the strengths and weaknesses of climate models. More importantly, AMIP
allows modelers to learn more about their own model from having indepen-
dent examination of their own model outputs in comparison with other
models. It is the driving force behind many efforts in model improvement
at research institutions. Moreover, AMIP results have shown that even if an
individual model does not perform well, the ensemble means of all models
can do a better job than individual models in simulating the evolution of
climate. This is because model errors tend to cancel out in large model
ensembles, so that the signal-to-noise ratio can be increased. The use of
super-ensemble techniques (Krishnamurti et al.
2000
), whereby statistical
weights are assigned to each model variable, at each grid point, hold promise
for more reliable simulations, and climate projections on regional scales.
Following on the success of AMIP-I, an expanded AMIP-II is now underway
to include a wider range of variability, to accelerate model physics improve-
ment and to improve the infrastructure for model diagnostics, validation,
and experimentation. Various model intercomparison projects (MIPs), tai-
lored to various modeling communities, have emerged in recent years. These
include the Coupled Model Intercomparison Project (CMIP), the Seasonal
Model Intercomparison Project (SMIP), the Project of Intercomparison
of Land Parameterization Schemes (PILPS), the Paleoclimate Model
Intercomparison Project (PMIP), and many others.
Search WWH ::
Custom Search