Search results

Presents the development and implementation of a four‐dimensional variational (4D‐var) data assimilation technique for a comprehensive Eulerian chemistry‐transport model. The method aims at analysing the chemical state of the atmosphere on the basis of trace gas observations with arbitrary distribution in time and space, a chemistry‐transport model, and a priori knowledge as available from climatological records or preceding model runs. The model under consideration is the University of Cologne EURAD‐CTM2 with the full RADM2 gas phase mechanism. Describes the storage and recalculation strategy of a parallel implementation of the 4D‐var method and first experiences of its performance, when model generated data are provided as artificial observations. The problem of pre‐scaling the minimization problem is discussed in some detail. It is found that the algorithm is well suited to adapt the model trajectory to the observation data.