Data assimilation and use of observations in NWP
Convener: A. Cress 
Oral Programme
 / Wed, 14 Sep, 09:45–13:00  / Room Princeton
Poster Programme
 / Attendance Thu, 15 Sep, 16:00–17:00  / Poster Hall (Ground Floor)
The quality of a forecast from a numerical weather prediction model depends on the accurate determination of the initial state of the atmosphere and a sophisticated forecast model to simulate the subsequent atmospheric evolution realistically. In recent years, variational data assimilation systems have been successfully developed combining the incomplete and erroneous observations of the present with a forecast model that contains all the information about the atmospheric state of the past to achieve the “best estimate” of the true atmospheric state at the start of the analysis. The variational technique was essential to assimilate satellite data successfully. An extension of variational techniques including longer assimilation windows and weak constraint methods to allow for the inclusion of model error estimates are current research areas. Additionally, ensemble based assimilation systems allowing for a flow dependent estimation of background error variances and co-variances are under development.

Besides a sophisticated assimilation and forecasting system, high quality, reliable observations are essential for a successful forecast at all ranges. Impact studies, mostly in the form of OSEs (Observing System Experiments), of the relative benefit of current observing systems to data assimilation and forecast skill at short and medium range are important to estimate the value of specific data types and to optimize cost-effective use of observation resources. This information can be used to improve utilization of current observation systems, to locate target areas where additional observations could provide the most benefit and to determine priorities for new observing systems. New diagnostics rating the impact of any observation subset on the analyses and forecast, based on the adjoint of the data assimilation mathematical operators and of a short range forecast model, have recently been developed. Design studies on current and future observing networks, e.g. satellite missions, can be performed using simulated data and OSSEs (Observing System Simulating Experiments).

This session will accept papers about current progress and results from all different kind of assimilation techniques, studies related to the impact of observations on NWP analyses and forecasts as well as targeting strategies including, but not limited to, the following

• Progress in data assimilation techniques
• Results of ensemble based assimilation methods
• Global and regional observing system experiments (OSEs)
• Design studies of future observing systems (OSSEs)
• Assessment of the impact of targeted observations on NWP models
• The use of targeted data in field experiments (COPS, ETReC, IPY, etc.)
• Strengths and weaknesses of current sensitivity computation methods

Of particular importance to this session will be presentations on studies documenting the impact of innovative observing systems on meso-scale NWP models, not necessarily operational.