AMIL2DA
Changes in atmospheric composition are important in the context of stratospheric ozone depletion, global change and related environmental problems. The Michelson Interferometer for Passive Atmospheric Sounding (MIPAS), which is a core instrument of the ENVISAT polar platform successfully launched on 1st March 2002 by the European Space Agency (ESA), is a powerful tool to measure vertical profiles of trace species on a global scale. While operational data processing by ESA covers only analysis of pressure, temperature, and the mixing ratios of the species H2O, O3, HNO3, CH4, N2O and NO2, MIPAS infrared spectral limb emission measurements contain information on a bulk of further species relevant to environmental problems mentioned above. The goal of AMIL2DA was to generate data analysis tools for these supplemental species along with thorough validation of these algorithms.
Instead of merging the contributions of all participants to one data analysis algorithm which fits all purposes, the AMIL2DA strategy was to maintain the diversity of different computer codes of each group which are custom-tailored to their specific scientific needs.
Generally, data analysis consists of forward modeling of radiance spectra and inversion of measurement data. As a first step, forward radiative transfer algorithms and retrieval processors were adapted to the physical and computational needs of the MIPAS experiment. This includes adaptation to high resolution limb emission measurements, acceleration of numerical methods, and automated provision of input data as well as generation of spectroscopic line data not included in the current databases but presumably important to MIPAS applications.
In a second step, these codes were cross-validated by a blind-test intercomparison to reveal potential weaknesses of models. In particular the relevance of breakdown of thermodynamic equilibrium in the atmosphere was emphasized. After successful cross-validation of forward radiative transfer models and subsequent upgrading, these were operated in the context of an inversion computer code, which inferred atmospheric constituent abundances from measured spectra.
For purpose of cross-validation, different inversion algorithms were applied to a common set of synthetic measurement data in a blind-test mode. After upgrading the inversion models and fine-tuning of processing parameters, a common agreed set of real MIPAS measurements was used for further testing. Residuals between measured and best-fitting modeled spectra were analyzed for systematic patterns. Emphasis was put on candidate explanations such as inappropriate predictions on instrument characteristics; different use of initial guess and a priori data; over or under-regularization of the retrieval, and possible deficiencies in spectroscopic data.
These activities are the basis to better exploit MIPAS data by inferring vertical profiles of species relevant to ozone destruction and global change. Deficiencies in forward radiative transfer as well as inversion algorithms were detected and removed, and confidence in retrieval strategies and data products was strengthened. Completeness and appropriateness of physical effects included in the involved radiative transfer models were proved. Standardization of data products was gained while the diversity of data analysis strategies used by different European groups was maintained.
