Influence of observations on the operational ECMWF system

Editorial ECMWF regularly reviews the impact that various observing system data has on the forecast. A recent series of such observing system experiments are reported in the first article (page 2). In January 1997, there were severe floods in Greece due to torrential rainfalls over 12/13 J a n u a r y. The ECMWF forecast for that period is discussed on page 9. Three more reports in the ECMWF Te c h n i c a l Memoranda series are summarised on pages 7,8 and 9. The relentless march of computer technology improvements over the years has been, and still is, impressive. The article on page 12 highlights this; it reviews the changes that have occurred in the computer equipment at ECMWF over the past 17 years. As part of these ongoing changes in the Centre's computer equipment, the Fujitsu VPP700 will shortly be upgraded, the article on page 14 summarises the changes. F i n a l l y, remember that ECMWF's telephone number is changing (see box on the inside front cover), this is due to an area code change for R e a d i n g. Cover Two plots from a recent observing system experiment, showing Meteosat water vapour channel data (bottom panel) and the corresponding simulated model output (top panel). Such data will prove valuable when the 4D-var analysis is implemented as it is available every 1 ⁄2 h o u r. (False colours – blue dry – red wet – white clouds) See the article on page 2. Introduction The global meteorological observing system is extremely expensive and in the present economic climate conventional observations such as radiosondes are beginning to be severely reduced. At the same time improved satellite systems are becoming available. The operational observing network, which uses both conventional and satellite measurements, influences how accurately the initial atmospheric state can be prescribed and therefore to a large extent the resulting forecast accuracy. There is therefore an urgent need to investigate the importance of different observing systems on Numerical Weather Prediction performance. In this work we quantify through Observing System Experiments (OSEs), the contribution made by the main ground-based and satellite-based operational systems to medium range forecasting. In an OSE the impact of a specified observing system is assessed by comparing extended data assimilation and regular forecasts based on the total operational system with those generated excluding the particular observing system …