Recent progress in all-sky radiance assimilation
Geer, Alan ; Bormann, Niels ; Lonitz, Katrin ; Weston, Peter ; Forbes, Richard ; English, Stephen
Satellite observations make a major contribution to the Earth system data which are routinely assimilated into models to determine the initial conditions for weather forecasts. Since the beginning of satellite data assimilation in the 1980s, most cloud-affected observations have been rejected following the 'clear-sky' approach. This is because, in areas of cloud and precipitation, neither model forecasts nor the conversion of model values into satellite observation equivalents (observation operators) have been accurate enough. The machinery for using cloud and precipitation in data assimilation has needed decades of development, but the work is starting to pay off. A decade ago, ECMWF introduced direct 'all-sky' assimilation of satellite radiances in the presence of cloud and precipitation. The aim was to extract more information in sensitive and under-observed areas, particularly in midlatitude fronts. We saw that four-dimensional variational data assimilation (4D-Var) was able to infer updates to winds, temperatures and pressures from the location of cloud and precipitation in the observations, resulting in improved medium-range forecast quality in ECMWF's Integrated Forecasting System (IFS). Recent progress in exploring the full potential of assimilating observations of cloud and precipitation has been substantial. In one stream of work, we have expanded the coverage of all-sky assimilation from a handful of microwave sensors with limited impact to now nine sensors that are a major part of the observing system. We aim to expand all-sky assimilation to the rest of the operational microwave and infrared sensors over the next few years, and we hope to add entirely new types of sensors aimed primarily at cloud and precipitation, such as the upcoming Ice Cloud Imager on EUMETSAT's next generation of polar satellites. Progress has also been made in properly representing observation error correlations; using more observations over land surfaces; and exploiting the information provided by cloud and precipitation-affected radiances to further develop the modelling of moist processes in the atmosphere.</p>
Accès à la notice sur le site du portail documentaire de Météo-France