ASSIMILATION OF SEVIRI INFORMATION IN THE MET OFFICE'S FORECAST MODELS: INITIAL EXPERIENCES

• Main Authors: Ruth Taylor, Peter Francis, Roger Saunders
• Other Authors: The Pennsylvania State University CiteSeerX Archives
• Format: Text
• Language: English
• Subjects: Nimrod; Saunders; North Atlantic
• Online Access: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.550.9714

SEVIRI cloud information is currently assimilated operationally into the 12 km North Atlantic European (NAE) model at the Met Office using an analysis correction scheme, with derived cloud information provided by a nowcasting system, NIMROD (Golding, 1998). As well as work on assimilation of clear-sky radiances and assimilation of all-sky radiances to extract dynamical information from cloud evolution, research into variational assimilation of cloud products is being carried out. Both NIMROD products and cloud products derived from SEVIRI radiances by the AUTOSAT system (Saunders, 2006) are being tested; this poster presents results using the latter. Cloud top height and effective cloud amount from the AUTOSAT system are converted into column cloud by specifying a cloud amount on each model level. Below the cloud top it is assumed that nothing is known about cloud amount. A diagnostic relationship based on Smith (1990) is then used to convert cloud amount into relative humidity information. The data assimilation scheme compares this to model relative humidity and calculates increments to the model state which best fit the cloud-derived relative humidity, given errors on both observed and model-background humidities. For this poster, we compared the moisture increments produced by 3D-Var analysis for a single case (15.08.2007, 00Z), assimilating two sets of data: a “control ” set including surface observations, sonde and aircraft data, atmospheric motion vectors (satellite winds), scatterometer winds, ATOVS and SSMI radiances and ground-based GPS observations over a six-hour window centred on the analysis time; the control set + SEVIRI cloud information, taken from a single image half an hour before analysis time, and thinned to a resolution of 12 km.