Abstract
First results are presented of 18 months' experience with a microwave profiler that has been in operation in an unattended mode. Profiles of temperature and water vapor were retrieved without bias by a statistical regression method that was more accurate as opposed to a neural network approach, in particular for water vapor. Cloud liquid water was estimated by a neural network. The accuracy of the retrieved profiles estimated against quasisimultaneous radiosonde measurements are of comparable quality to that of the retrievals of ground-based Fourier transform infrared (FTIR) measurements. For temperature, the accuracy is about 0.6 K near the surface and less or equal to 1.6 K up to 7 km in summer and 4 km in winter. For water vapor, the corresponding values are 0.2–0.3 g m−3 near the surface and 0.8–1.0 g m−3 from 1- to 2-km altitude. The vertical resolution, however, is worse than that of FTIR measurements.
Two case studies—a 1-week anticyclonic situation and 1 day with a cold front passage—demonstrate the capacity of microwave radiometry to sense the thermodynamic structure of the lower troposphere up to 3–4 km quasi-continuously with reasonable accuracy and height resolution, interrupted only during precipitation events. The diurnal course of temperature and humidity as well as the weakening of the amplitudes and decay with altitude was traced up to 4 km. Considerable structure was found, too, in the liquid water profile during the passage of the cold front. The cloud base retrieved from the microwave data corresponded well with collocated ceilometer measurements.
Corresponding author address: Dr. Jürgen Güldner, Deutscher Wetterdienst, Meteorologisches Observatorium Potsdam, Postfach 60 05 52, 14405 Potsdam, Germany. Email: juergen.gueldner@dwd.de