Reducing systematic errors in cold-air outbreaks
Forbes, Richard ; Geer, Alan ; Lonitz, Katrin ; Ahlgrimm, Maike
Models and observations both have uncertainties. Characterising and representing uncertainty and its growth in time is a significant activity at ECMWF with the development of ensemble forecasts (ENS) and more recently the Ensemble of Data Assimilations (EDA). At the same time, an ongoing challenge of research and development is to reduce uncertainties in both the initial state and the forecast model through improvements in the use of observations, data assimilation methodologies and the fidelity of the model dynamics and physics.
Comparing model forecasts with observations on medium-range to seasonal timescales can highlight systematic errors, but non-linear interactions and feedbacks make it difficult to attribute the errors to a particular source. Using short-range forecasts in the data assimilation system, on the other hand, means the model state is reasonably close to reality, and systematic departures of these forecasts from the observations are more easily assessed. Such assessments may even make it possible to identify regime-(or flow-)dependent systematic errors and to attribute these errors to specific physical processes.<br>In this article, we describe how this approach has enabled us to identify supercooled liquid water cloud in convective cold-air outbreaks as the source of regime-dependent systematic model errors in simulated microwave radiance and shortwave radiation. A proposed solution reduces the model errors at all forecast lead times. This is seen in reduced bias against assimilated microwave observations and reduced shortwave radiation error.
Accès à la notice sur le site du portail documentaire de Météo-France