rpt1
Report 1: The Validation of Atmospheric Models
Gates, W. Lawrence
March 1992, 17 pp.
The validation of atmospheric models is a key part of the modelling
enterprise, but one to which increased attention needs to be given if systematic
progress is to be made in the development of predictive climate models.
The validation of current AGCMs in terms of the mean seasonal distribution
of primary variables such as pressure, temperature, and wind shows a reasonable
ability to simulate the observed large-scale features, while at the same
time identifying a number of systematic errors. More recent validations
have included the simulation of variability, which reveals a modest level
of skill but with further systematic errors. Recent results from mesoscale
models nested within AGCMs, however, have shown substantial skill in the
simulation of regional climate. In addition to conventional data sources
of various resolutions, current model validation is enriched by the use
of satellite observations and other special data sets, as well as by the
analyses from operational models.
A comprehensive atmospheric model validation program includes examination
of not only the mean and variance, but of the complete frequency distribution.
Moreover, in addition to the primary dynamical and physical variables,
the various derived quantities associated with fluxes and processes and
the occurrence of specific events should also be evaluated. A complete
validation would also include evaluation of a model's ability to simulate
more than just the present climate and/or its ability to simulate observed
climate change (the latter aspect necessarily including the oceans). This
effort will require the acquisition, calibration and processing of global
observational data sets specifically for the purpose of model validation.
Useful approaches toward such a program include the reanalysis of recent
decades with modern AGCMs and advanced data assimilation systems, and the
extension of the various atmospheric model intercomparison projects now
underway to include a wider variety of diagnostics. (pdf
file)
UCRL-MI-123395