OSSEs for the stratosphere

Richard Swinbank
with Mike Keil, David Jackson, William Lahoz, Roger Brugge and Cecilia Girz

Met. Office
Exeter, U.K.


The use of Observing System Simulation Experiments (OSSEs) is a well-developed technique to assess the benefit of new types of observations. In an OSSE, a numerical weather prediction model is used to produce a "nature run". A complete set of observations is simulated from the nature run, then assimilated into a second numerical model. A pair of assimilation experiments are run, one including and one excluding the particular type of observation that is under study. The two experiments can then be compared with the nature run, to assess the benefit of using that particular observation type. This procedure can be extremely resource intensive.

There is a major gap in the stratospheric observing system, namely the lack of stratospheric wind data. Using the Met Office stratospheric data assimilation system, two separate studies have been carried out to assess the likely impact of stratospheric wind observations. First, we assessed the likely impact of wind measurements from a proposed satellite wind instrument, SWIFT, using a "reduced OSSE", based on use of simulated retrievals rather than radiances. In a second study we assessed the impact of winds inferred from a set of stratospheric balloon trajectories, using a "Partial" OSSE, or POSSE. Both studies indicated that stratospheric wind observations would be beneficial. We discuss the robustness of the results, bearing in mind the effect of using cheaper versions of the OSSE method.