Quantifizierung und Optimierung der radiometrischen Genauigkeit des Fourierspektrometers MIPAS-B2

Abstract

The cryogenic balloon-borne Fourier transform spectrometer MIPAS-B2 (Michelson-interferometer for passive atmospheric sounding - balloon version 2) is a limb sounder measuring emission spectra of the atmosphere in the mid-infrared. From these spectra, trace gas profiles of numerous species are deduced which are relevant to the ozone chemistry and the greenhouse effect. Measurements of atmospheric emission spectra demand a very good instrument performance and a precise data processing. Although the instrument is cooled, its thermal self-emission is of the same order of magnitude as the atmospheric signal. Therefore, a profound characterisation of the instrument's radiometric properties is necessary to obtain a high accuracy in the calibrated spectra. The objective of this thesis is the quantification and optimisation of the radiometric accuracy of MIPAS-B2. For this purpose the various error sources leading to a radiometric error in the calibrated spectra are examined. These errors are: noise in the calibration measurements, changes in the instrument's self-emission and responsivity due to thermal drifts, errors in the baseline determination due to residual atmospheric signatures, uncertainties of the temperature and emissivity of the blackbody which is used for calibration, errors in the phase correction, and residual errors after correction of the detector non-linearity. Between 1997 and 2002, seven balloon flights have been performed under arctic winter conditions and in mid-latitudes in spring- or summertime. The analysis of these flights reveals that due to the thermal drift of the instrument, frequent calibration measurements are mandatory for a high accuracy. However, a long measurement time was necessary in the earlier flights to properly distinguish residual atmospheric signatures from the instrument's self-emission and to provide a good phase correction. Within this thesis the instrument characterisation and data processing could be improved so that the time needed for calibration measurements is reduced significantly. This allows to calibrate more often. In spite of the more frequent calibration measurements, the overall time for calibration could be reduced compared to earlier flights. The remaining errors in the calibrated spectra have been quantified using simulated spectra. For the correction of the detector non-linearity, a new method has been developed, allowing a precise determination of the non-linearity parameters from spectra that are measured during flight. An estimate of the total systematic error in the calibrated spectra is given for the flight on 12th February 2002, which represents the current state of the instrument. The software, that has been developed within the framework of this thesis, will allow an operational determination of the systematic error in the calibrated spectra for future flights. Finally, the influence of systematic errors in the spectra on the retrieved trace gas profiles is investigated. It is shown that one has to distinguish between multiplicative and additive error contributions, because their influence on the trace gas profiles is different. A scaling error leads to an error in the profiles in the same order of magnitude, whereas additive errors barely alter the retrieved profiles, as long as an offset is treated as retrieval parameter and the spectra are not affected by clouds or aerosols. For the interpretation of correlations between different trace gases, the requirements for the radiometric accuracy are higher than for isolated profiles.