The solution to the problem of when and where observations should be taken in order to minimize forecast error requires knowledge of (a) the forecast error at the time the observations are taken (b) how the assimilation of observations will reduce the error in the estimate of the state of the atmosphere and c) how errors in the analysed state of the atmosphere will subsequently evolve. It is shown how a new type of Kalman filter called the ensemble transform Kalman filter (ET KF) incorporates these three factors. In particular, it allows the forecast error variance associated with a wide range of feasible deployments of observational resources to be rapidly estimated. A variant of the ET KF has been used in FASTEX, NORPEX 98 and NORPEX 99 to deploy aircraft borne dropsondes. While results from these experments show that target sites that significantly influence the verification region have been successfully identified by the technique, in a disturbing number of cases, the improvements rendered by the supplemental observations have been marginal or negative. By examining a simple case relevant to the problem of assimilating observations of hurricane like vortices, we find support for the suggestion that the isotropic error covariances assumed by today's operational data assimilation schemes may be responsible for these rather disappointing results. An application of the ET KF to the same problem suggests that the impact of adaptive observations will markedly improve once reasonable flow dependent data assimilation schemes are used in operational data assimilation schemes.
Close window or click on previous window to return to the Conference Program.
12th Conference on Atmospheric and Oceanic Fluid Dynamics