Tests of schemes to infer stratospheric temperature from satellite measurements

• Main Author: Jackson, David R.
• Published: University of Edinburgh 1990
• Subjects: 520
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.652910

In this thesis we test a retrieval/analysis scheme for inferring stratospheric temperature from satellite observations of radiance. The scheme is similar to that used by the UK Meteorological Office. The retrievals are made by using a multiple linear regression model which regresses radiances against Planck function, whilst the analyses are made using a linear time/space interpolation method. In addition, we compare analyses made using time/space interpolation with analyses made using another analysis scheme which sequentially estimates Fourier coefficients at fixed latitudes using a version of the Kalman Filter. Because of the lack of 'ground truth' observations in the stratosphere, the schemes are tested in simulation experiments. Preliminary tests of the time/space interpolation and sequential estimation analysis schemes are made using idealised radiance fields which resemble observations made by a satellite radiometer in the northern hemisphere winter stratosphere. The regression retrieval scheme and the two analysis schemes arc also tested in a more sophisticated experiment in which the 'true' atmosphere is represented by an atmosphere simulated by a numerical model. Simulated observations are calculated by computing the radiance that would be observed from the 'true' atmosphere by a satellite instrument. The radiances are then retrieved and analysed and the resultant analyses compared with the corresponding 'true' fields. Tests are made using output from a day when a sudden wanning was present. The retrieval scheme is seen to perform less well within the area of the sudden wanning than outside it. However, this may be expected as the vertical structure within the sudden warming is generally too small to be resolved by a satellite instrument. The analysis scheme analyses the stratospheric field well, even in the area of a sudden warming. These results, and results from preliminary tests made using idealised radiance fields, suggest that the analysis is generally of better quality when the distance radius used to select observations for the scheme is small. Results of tests of the sequential estimation scheme reveal that this method also produces satisfactory analyses of idealised radiance and model fields. Constraints of time prevented more rigorous testing of the scheme, but suggestions for further research are given.