The intercomparison of radiation codes in climate models: An overview

Abstract

The recognition of the central role of radiative processes in many proposed climate change mechanisms and the perception of possibly significant uncertainties in the estimates of these fundamental processes led the Joint Scientific Committee of the World Climate Research Programme and the International Radiation Commission of the International Association of Meteorology and Atmospheric Physics to initiate the International Intercomparison of Radiation Codes in Climate Models (ICRCCM). The results from model calculations with specified clear‐and‐cloudy conditions show that many radiation algorithms may have unidentifiable but large errors that may significantly affect the conclusions of the studies in which they are used. This is true for climate modeling but may also be the case for other applications such as the estimation of radiation fluxes at the surface from satellite observations. As the study has progressed over a 4‐year period, there has been a narrowing of results as errors were found in some codes and as the understanding of many modeling problems increased. Many of the results, particularly for clear‐sky conditions, indicate that we are close to the range of (relative) accuracy for calculating flux quantities necessary for many climate programs. However, not all models will give such accuracy. It is recommended that the ICRCCM test cases be used to test radiation algorithms prior to their application to climate‐related problems. The participants feel that the rather large discrepancies revealed during ICRCCM cannot be decisively resolved by further calculation. Therefore the group recommends the organization of a program to simultaneously measure spectral radiance at high spectral resolution along with the atmospheric data necessary to calculate radiances.

The Radiative Forcing Model Intercomparison Project (RFMIP): experimental protocol for CMIP6

“…More subtly, parameterizations require so much effort to develop and maintain that they can lag behind current spectroscopic knowledge, especially for solar radiation where new absorption features continue to be identified (Rothman et al, 2013). These errors have been apparent in previous assessments of radiative transfer parameterizations for both gaseous absorption (Ellingson and Fouquart, 1991;Collins et al, 2006;Oreopoulos et al, 2012;Pincus et al, 2015) and aerosols . RFMIP will assess parameterization error in instantaneous radiative perturbations due to greenhouse gases and due to aerosols.…”

Radiative forcing by contrails

“…The solar and terrestrial versions of this code performed successfully in an intercomparison of radiation codes (Ellingson and Fouquart, 1990). Fischer and Grassl (1984) described the shortwave version.…”

“…Although the model is highly resolved for absorption and scattering by gases and aerosols in both SW (54 bands) and LW (67 bands) spectra, it uses only 6 SW bands and 12 LW bands to compute particulate scattering and absorption and surface emissivities and albedos. The model has also been tested during the intercomparison of radiation codes (Ellingson and Fouquart, 1990) and applied for cloud studies (Charlock et al, 1995). It uses spherical particles for liquid water at all wavelengths.…”

A springtime cloud over the Beaufort Sea polynya: Three‐dimensional simulation with explicit spectral microphysics and comparison with observations

“…The absorption spectrum of water vapor for longwave radiation is described by dividing the spectrum into four regions with different transparency and vapor absorption coefficients a vn in each n-th region. The spectral divisions and coefficients were tuned by comparison with the results of line-by-line calculations from the Intercomparison of Radiative Codes for Climate Models (ICRCCM) [Ellingson and Fouquart, 1990]. After tuning, an error in fluxes does not exceed 3 -8% through the entire troposphere, and an error in the LWR balance is less than 2 -5%, which is quite sufficient for the cloud model with a rather short time of integration.…”