An evaluation of the long-wave radiative transfer code used in the Met Office Unified Model

Abstract

A detailed evaluation of the radiative transfer code used in the Met Office Unified Climate/Forecast Model is performed, comparing it with a line-by-line model and testing the climatic effects of errors in a radiative--convective model. The radiative forcing at the tropopause due to CO₂ changes within SRES scenarios and across Quaternary glacial cycles is represented with reasonable accuracy, suggesting that surface temperature will be correctly predicted. However, this is achieved by partial cancellation of opposing errors in upward and downward fluxes. The changes in the vertical profiles of radiative fluxes and the changes to surface and top-of-atmosphere fluxes all show significant errors, even at twice pre-industrial CO₂. This causes a sign error in the change in the convective flux in the radiative--convective model. Performance of the code deteriorates rapidly above four times pre-industrial CO₂. For less-abundant greenhouse gases, CH₄ and N₂O, the errors are larger as a proportion of their radiative forcings. Errors for surface and top-of-atmosphere fluxes for CO₂ are similar to those from the mean of the general circulation model (GCM) codes submitted to the intercomparison of radiation codes for IPCC AR4, implying that errors as found here may not be uncommon in climate models. A renewed emphasis on accuracy in radiative transfer calculations and openness in intercomparison studies is necessary to improve the modelling of climate change. Copyright © 2009 Royal Meteorological Society