On the addition of microwave sounders and numerical weather prediction skill

Abstract

AbstractThe utility of microwave sounding for numerical weather prediction (NWP) is well established. However, the volume of sounder data assimilated in NWP systems may change substantially in upcoming years, as there is an ageing constellation of satellites but also the prospect of proliferating small satellite constellations. This study examines the addition of temperature and humidity sounders from a baseline that includes no microwave sounders at all, aiming to elucidate the incremental benefit gained from adding sounders to the assimilation system. Framed as a series of observing‐system experiments (OSEs), large improvements in forecast skill and background fits to independent observations are gained from the first sounders added. Significant further benefit is observed from additional sounders, with the impacts largest at higher latitudes and no saturation evident in the maximal setup. In the second part of this study, the relative importance of updating background errors for OSEs is investigated to give greater context to the sounder addition results. The ensemble of data assimilations (EDA) underlying the background‐error specification is rerun for this purpose to be consistent with the observation usage in the OSE. Comparing experiments with updated and unchanged background errors shows that the effect of updating background errors is secondary to that caused by the observing‐system change itself. Rerunning the EDA can reduce some of the forecast skill apparently lost in a data‐denial experiment or gained when assessing data addition, but this is roughly 10% of the overall change and should not alter conclusions resulting from typical OSEs, in which background errors are not tailor‐made.