Tropospheric Chemical State Estimation by Four-Dimensional Variational Data Assimilation on Nested Grids

The University of Cologne chemistry transport model EURAD and its four-dimensional variational data assimilation implementation is applied to a suite of measurement campaigns for analysing optimal chemical state evolution and flux estimates by inversion. In BERLIOZ and VERTIKO, interest is placed on atmospheric boundary layer processes, while for CONTRACE and SPURT upper troposphere and tropopause height levels are focussed. In order to achieve a high analysis skill, some new key features needed to be developed and added to the model setup. The spatial spreading of introduced observational information can now be conducted by means of a generalised background error covariance matrix. It has been made available as a flexible operator, allowing for anisotropic and inhomogeneous correlations. To estimate surface fluxes with high precision, the facility of emission rate optimisation using scaling factors is extended by a tailored error covariance matrix. Additionally, using these covariance matrices, a suitable preconditioning of the optimisation problem is made available. Furthermore, a module of adjoint nesting was developed and implemented, which enables the system to operate from the regional down to the local scale. The data flow from mother to daughter grid permits to accomplish nested simulations with both optimised boundary and initial values and emission rates. This facilitates to analyse constituents with strong spatial gradients, which have not been amenable to inversion yet. Finally, an observation operator is implemented to get to assimilate heterogeneous sources of information like ground-based measurements, airplane measuring data, Lidar and tethered balloon soundings, as well as retrieval products of satellite observations. In general, quality control of the assimilation procedure is obtained by comparison with independent observations. The case study analyses show considerable improvement of the forecast quality both by the joint optimisation of initial values and emission rates and by the increase of the horizontal resolutions by means of nesting. Moreover, simulation results for the two airplane campaigns exhibit outstanding characteristics of the assimilation system also in the middle and upper troposphere region