Adding constraints by in situ informations to optimal estimation retrievals of tropospheric water vapour profiles from microwave radiometry

• Published in: Journal of quantitative spectroscopy & radiative transfer 2012-11, Vol.113 (16), p.1993-2003
• Main Authors: Bleisch, R., Kämpfer, N.
• Format: Article
• Language: English
• Subjects: Constraining; Covariance matrix; Inversion methodology; Microwaves; Optimal estimation; Optimization; Radiative transfer; Radiometry; Remote sensing; Retrieval; Spectra; Vapour; Water vapour
• ISSN: 0022-4073; 1879-1352
• DOI: 10.1016/j.jqsrt.2012.07.002

The optimal estimation method is a widely used method to invert species profiles from spectra observed by a microwave radiometer. The classical retrieval is constrained by the a priori profile and the corresponding covariance matrix, which is a “soft” constraining of the retrieved profile to a certain range of values. However, in some cases a “hard” constraining of the profile to a fixed value known from other measurements would be desirable. This work presents an approach to introduce such “hard” retrieval constraints (fixed-points) into optimal estimation retrievals by adapting the a priori covariance matrix. Its application is tested on the example of the retrieval of tropospheric water vapour volume mixing ratio (vmr) profiles from spectra of the MIAWARA radiometer operated by the Institute of Applied Physics, University of Bern. Thereby the cloud base height is one candidate to deliver a fixed-point, as the corresponding vmr value can be determined by assuming a relative humidity of 100%. As a test, the approach is applied to spectra simulated from balloon soundings. The cloud base height is derived from these same balloon soundings. The results show a significant improvement of the retrieval performance for all cases with liquid clouds except for fog. Afterwards the approach is also applied to real MIAWARA data. Thereby the measurements of a ceilometer and an infrared sensor (both installed close to the instrument) are used to derive a fixed-point. In principle, the application on real data also works. However the retrieval performance is limited, because we are currently not able to determine the vmr value at fixed-point altitude with suitable precision. The cloud base temperature, needed for the calculation of the vmr value at fixed-point altitude, is determined indirectly from measurements of an infrared sensor attached to the instruments or by for example interpolating data from ECMWF-reanalysis. In both cases the precision is not very high, with there being a significant uncertainty in the resulting vmr-value at cloud base. However, our new radiometer TEMPERA has the capability to deliver more reliable estimates of the cloud base temperature in near future. ► Classical OEM-retrievals constrain the retrieved profile to a certain range of values. ► This work discusses an approach to introduce retrieval constraints with a fixed value. ► The cloud base height is a promising candidate to derive a fixed-point (RH ∼ 100%). ► Retrieval simulations showe