We implemented in IFS a recently developed technique that estimates fluxes from randomly-chosen spectral subsets designed to minimise error in surface fluxes. This approach introduces substantial uncorrelated noise in surface fluxes and heating rates but, reducing the cost associated to the spectral integration, it allows to update radiation more frequently and on a denser spatial grid.
We evaluate how this approach applies to medium range (< 10 days) weather forecasts by testing different combinations of temporal and spectral sampling at high spatial resolution. We assess the impact of each configuration, by comparing the surface parameters forecast to observations from synoptic stations and the errors are evaluated as function of the computational effort.
The results show that a combination of spectrally sparse but temporally and spatially dense radiation calculations has the potential to reduce the forecast errors, particularly those associated with under-resolved topography and stable nocturnal boundary layers, with affordable computational costs.