The radiative impacts of implementing RRTMG on NASA's GEOS-5 Atmospheric GCM

Description

It has been demonstrated that Monte Carlo spectral integration can be combined with Independent Column Approximation (ICA) calculations to replace CPU-intensive explicit spectral integration with negligible influences on GCM simulations. Such a "McICA" approach has been implemented in Goddard's GEOS-5 GCM in conjunction with the ARM-supported RRTMG radiation package. GEOS-5 with McICA can adopt horizontally variable clouds that are allowed to overlap arbitrarily both in terms of cloud fraction and layer condensate distributions. In our presentation we will show radiative and other impacts of the combined horizontal and vertical cloud variability on multi-year simulations of an otherwise untuned GEOS-5 with fixed SSTs. Introducing cloud horizontal heterogeneity without changing the mean amounts of condensate reduces the reflected solar and increases the emitted thermal radiation to space, but disproportionate changes may yield increased radiative imbalance at TOA. The net radiation at TOA can be modulated by allowing the parameters of the generalized overlap and heterogeneity scheme to vary, a sensitivity whose limitations we will discuss. Results from our investigation on the extent to which other mean cloud properties need to be retuned in order to get closer to net radiation flux balance at TOA will also be presented.