The role of balance in data assimilation
R. N. Bannister (University of Reading, Department of Meteorology)
Data assimilation is one half of the problem of numerical weather prediction, modelling the atmosphere being the other. Data assimilation estimates the initial conditions of the model by bringing together information from observations and a forecast. The measure of the uncertainty (or errors) of such data is very important in the way that the assimilation treats the data. The atmosphere is governed by a set of dynamical equations which influence strongly how errors of meteorological variables are correlated. In the computer-intensive operational data assimilation problem, these error correlations are quantified largely with balance relationships. In this paper the current methodology is discussed, and future challenges are highlighted.