Observing‐system impact assessment using a data assimilation ensemble technique: application to the ADM–Aeolus wind profiling mission

Abstract

AbstractEnsembles of parallel 4D‐Var data assimilation cycles have been used to assess the impact of two observing systems: the existing network of radiosonde and wind profilers, and the future spaceborne ADM–Aeolus wind‐profiling LIDAR. We demonstrate that this new technique for impact assessment provides a practical alternative to the traditional observing system simulation experiments (OSSEs), with the particular advantage that real existing observations are assimilated exactly as in operational practice, and do not need to be simulated artificially. It is only the future observing system under test (ADM–Aeolus in our case) that is generated through simulation. Furthermore, in contrast with OSSEs, there is no need to generate an artificial reference atmosphere (‘proxy truth’ or ‘nature run’), and the problems normally associated with identical‐twin experiments are thus avoided. Our results, based on detailed simulation of the ADM–Aeolus wind‐measuring capabilities and expected data quality, show that ADM–Aeolus will provide benefits comparable to the radiosonde and wind‐profiler network, with analysis impact particularly over ocean and in the tropics. The impact is retained up to the medium range of forecast (around day 5). Our results for radiosonde and wind‐profiler impact agree qualitatively with those obtained with the well‐established observing system experiment (OSE) technique; this agreement gives some confidence in the usefulness of the ensemble‐based technique for impact assessment. Copyright © 2007 Royal Meteorological Society