Abstract Motivated by apparent forecast improvements in the Met Office system with Spire‐processed Global Navigation Satellite Systems (GNSS) radio occultation bending‐angle observations, various experiments have been run to test the effect of increased vertical smoothing on forecast quality. The initial experiments were run with additional smoothing applied to Spire's observations as part of the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT) processing. In these experiments it was seen that increasing the smoothing decreased the standard deviation of the observation departures, but also increased the vertical correlation length‐scales. These observations with additional smoothing were then ingested within a low‐resolution version of the Met Office numerical weather prediction (NWP) system, and the forecast quality was seen to be improved with the observations using additional smoothing compared with the observations using the operational processing. A second set of experiments was run which applied additional smoothing as a pre‐processing step within the Met Office system. The smoothing is thus applied to the low‐resolution Binary Universal Form for the Representation of meteorological data (BUFR) observations that are normally assimilated operationally. This method has the advantage that it is applied to the whole observation dataset, but the disadvantage that it is applied to the low‐resolution observations, which posed some technical challenges. It also meant that it was possible to make the smoothing length‐scale proportional to the spacing between vertical levels in the Met Office model. Tests applying the additional smoothing in this way demonstrated improved forecast performance over a wide range of variables. However, using a large smoothing length‐scale produced degraded results, and the degradation was seen first in the tropical region, suggesting that less smoothing is beneficial there. Further experimentation is planned, which would demonstrate the impact of additional smoothing on a second NWP system.
Bowler et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: