| Home > Newsevents > Training > Rcourse_notes > PARAMETRIZATION > RADIATION_TRANSFER > |
Radiation Transfer
March 2000
By Jean-Jacques
Morcrette
European Centre for Medium-range Weather
Forecasts, Shinfield Park, Reading Berkshire RG2 9AX, United Kingdom
Table of contents >>
Next Section >>
1 . Introduction: An historical perspective
Among the various processes responsible for the diabatic heating of the atmosphere (convection, turbulence, large-scale condensation, heat, moisture and momentum transfer at the surface, radiation transfer), which have to be described by physical parametrizations in a general circulation model (GCM), this last one stands because its impact is felt at all times and over the whole 3-D domain. Compared to other processes, radiation transfer (RT) also stands as having a long history of theoretical developments. From the statistical and quantum mechanics at the end of the 19th and beginning of the 20th century, Boltzmann, Stefan, Wien, Planck and Einstein made pioneering advances in the spectral description of the radiation emitted by a black body. Afterwards, spectroscopic studies of the gases important for the radiative budget of the atmosphere, and the development of various approximations made the calculation of the radiation transfer in the atmosphere a tractable problem.
The very first such approximation is the separation between shortwave (SW) and longwave (LW) schemes due to the obvious wavelength difference between a black-body at Sun's temperature (around 5800 K, with most of its energy below 4 microns, and the energy source outside the atmosphere) and that of the atmosphere (globally averaged equivalent temperature of about 255 K as seen from the top of the atmosphere, with most of its energy above 4 microns, and the sources being the surface, the radiatively active gases (mostly H2O, CO2, O3), and the clouds within the atmosphere). Chandrasekhar (1935, 1958, 1960) provided the first approximations to the radiation transfer in a scattering atmosphere. Elsasser (1938, 1942), Goody (1952, 1964), Curtis (1952, 1956), Godson (1953), Malkmus (1967) defined the first usable simplifications to the spectral representation of the radiation transfer including its dependence on two key atmospheric parameters, pressure and temperature, allowing atmospheric heating/cooling rates to be computed (Rodgers and Walshaw, 1966).
To the point, that in the mid-70s, it was said (WMO/Global Atmospheric Research Program, 1975) that radiation transfer was a solved problem, provided big enough computers were available. The only developments required were for fast parametrizations. A good description of the various approximations used in the 80's at the different stages in the development of computer-efficient radiation schemes (and of the necessary trade-offs between accuracy and efficiency, two contradictory arguments) is given in Stephens (1984) and Fouquart (1987).
Unfortunately, the InterComparison of Radiation Codes for Climate Models (ICRCCM) at the end of the 80s (Ellingson et al., 1991; Fouquart et al., 1991, Baer et al., 1996) showed that, if an overall good agreement in clear-sky longwave computations by line-by-line models existed, large discrepancies were found in the results of both the LW parametrized schemes and all types of SW schemes. Hereafter, at least for clear-sky atmospheres, these were tracked down to deficiencies, mainly in the implementation of the approximations used to deal with the spectral and vertical integrations.
Since ICRCCM, a number of parametrized RT schemes have been developed from line-by-line (LbLs) models so that their accuracy, at least in clear-sky atmospheres, should be very close from that of LbLs (e.g., Ramaswamy and Freidenreich, 1992; Edwards and Slingo, 1996; Mlawer et al., 1997, to name just a few. As far as clouds are concerned, the situation is much more complex with an ongoing debate whether or not clouds might absorb more than what current parametrizations account for, with the somewhat related role of the spatial inhomogeneities in the 3-dimensional distribution of water condensates on the radiation fields within and at the boundaries of a cloud, with the role that the overlapping of cloud layers plays on the vertical distribution of the radiative fluxes and heating rates. Also the computational environment of a general circulation model might also impose some external limitations on the quality of the representation of the cloud-radiation interactions. In the following sections, an overview of these different questions will be provided.
Training Course Notes Front Page >>
Table of contents >>
Next Section >>
 |
12.06.2002 |
 |
© ECMWF |
