Heterogeneous background-error covariances for the analysis and forecast of fog events

An ensemble assimilation, which is based on the operational cloud-resolving model Applications de la Recherche a lOperationnel a Meso-Echelle (AROME) and its 3D-Var assimilation system, is used to diagnose background-error covariances separately in areas with and without fog. The fog and haze analysis system Cartographies des Analyses du RIsque de BrOUillard ( CARIBOU) is used as reference to calibrate the best fog predictor from model fields, which was found to be a low-level nebulosity. It appears that the physical processes in fog layers lead to very specific balances between control variables as well as much shorter vertical correlation length-scales at low levels in background-error covariances. In order to spread the information from surface and satellite observations with adequate structures in fog areas, a binary heterogeneity based on theuse of geographical masks is added to the background-error covariances. After the elimination of discontinuities at the mask borders, the positive impact of this formalism on the analysis-increment structure is discussed. Impact studies based on long-term real cases indicate that the global impact is closely related to the quality of the fog mask, for which future improvements are awaited. Copyright (C) 2011 RoyalMeteorological Society