4D-variational data assimilation with an adjoint air quality model for emission analysis

The problem of analyzing the chemical state of the troposphere and the associated emission scenario on the basis of observations and model simulations is considered. The method applied is the four-dimensional variational data assimilation method (4D-var) which iteratively minimizes the misfit between modeled concentration levels and measurements. The overall model–observation discrepancy is measured in terms of a cost function, of which the gradient is calculated for subsequent minimization by adjoint modeling. The model applied is the University of Cologne EURopean Air pollution Dispersion model (EURAD) simulating the meso-alpha scale. The forward and adjoint components are Bott’s horizontal and vertical advection scheme (Bott, Mon. Wea. Rev. 117 (1989), 1006), implicit vertical diffusion, and the RADM2 gas phase chemistry. The basic feasibility of the adjoint modeling technique for emission rate assessment is demonstrated by identical twin experiments. The objective of the paper is to demonstrate the skill and limits of the 4D-var technique to analyze the emission rates of non-observed precursor constituents of ozone, when only ozone observations are available. It is shown that the space–time variational approach is able to analyze emission rates of NO directly. For volatile organic compounds (VOC), regularization techniques must be introduced, however.