Variational assimilation of stratospheric remote sounding data by an adjoint chemistry-transport-model

In recent years high resolution data have become available due to the deployment of satellite born instruments observing the state of a large number of stratospheric constituents with unprecedentedly high horizontal and vertical resolution. These measurements are valuable for accessing the state of the atmosphere and for helping to develop guidelines for its preservation as a protection layer for terrestrial life. Nevertheless, the measurements alone offer limited direct information. They require interpretation and combination with other information sources to accurately describe the state of the atmosphere. By combining measurements with atmospheric models, the measurements' scatter in time and space can be mitigated and their interpretation improved. A chemistry-transport model (CTM) version of the Cologne model of the middle atmosphere (COMMA) has been developed to model transport and chemical transformation of atmospheric trace gases. In this thesis, the method of four-dimensional variational data assimilation is used to realize the above-mentioned combination in a mathematically rigorous way. The goal is to identify the most probable chemical state of the atmosphere using all available information. As chemistry-transport modelling is an initial value problem, those initial conditions are sought which result in best compliance of the model state with available information during the time period considered. As a measure of compliance, a cost function is specified, whose gradient is needed for minimising the cost function, and which can be obtained by means of an adjoint model. To this end, the adjoint code to the CTM was developed. As a last building block of the data assimilation system, the L-BFGS algorithm for the minimization of the cost function was added. The adjoint model of the middle atmosphere of Cologne CTM (AMMOC-CTM) data-assimilation system thereby developed was tested by applying it to data measured by the cryogenic infrared spectrometer and telescopes for the atmosphere (CRISTA) during the first campaign conducted in November 1994. In this thesis, the first 4D var chemical data assimilation system is presented that takes into consideration horizontal correlations by means of a new penalty term in the cost function. The usefulness of the AMMOC-CTM for the analysis of chemically active constituents from satellite measurements was demonstrated by the synoptic maps which it produced. The model results were tested for consistency with the underlying assumptions by means of statistical investigations that for the first time included evaluation of the background term. After parallelization, the developed model system can be used to routinely create synoptic maps assimilating for example Envisat (environmental satellite) data. CTMs such as the AMMOC-CTM are not only useful for scientific investigations, but can provide essential information for environmental policy making and supervision as well, for example concerning the interdiction to produce certain CFCs. (orig.)With CD-ROMAvailable from TIB Hannover: Rn 1818(155) / FIZ - Fachinformationszzentrum Karlsruhe / TIB - Technische InformationsbibliothekSIGLEDEGerman