Monitoring the observation impact on the short-range forecast

This paper describes the use of forecast sensitivity to observations as a diagnostic tool to monitor the observation impact on the 24-hour forecast range. In particular, the forecast error is provided by the control experiments (using all observations available) of two sets of observing system experiments performed at ECMWF, a month ill summer 2006 and a month in winter 2007, respectively. In Such a way, the observation data impact obtained With the forecast sensitivity is compared with the observing system experiment's data impact; differences and similarities are highlighted. Globally, the assimilated observations decrease the forecast error; locally, some poor performances are detected that are related either to the data quality or to the suboptimality of the data assimilation system. It is also found that the synoptic situation can affect the measurements or can produce areas of large field variability that the assimilation system cannot model correctly. Copyright (C) 2009 Royal Meteorological Society