Prediction of cloud ice signatures in submillimetre emission spectra by means of ground-based radar and in situ microphysical data

Abstract

Submillimetre down-looking radiometry is a promising technique for global measurements of cloud ice properties. There exist no observation data of sufficient size that can be used for detailed pre-launch studies of such an instrument and other means must be found to obtain data to optimise the instrument design and similar tasks. Several aspects of the observations make traditional retrieval methods not suitable and nonlinear multidimensional regression techniques (e.g. Bayesian Monte Carlo integration and neural networks) must be applied. Such methods are based on a retrieval database and to be successful the database must mimic relevant real conditions closely. A method to generate such databases of high quality is described here. Correct vertical distributions of cloud ice are obtained by basic data from ground-based radars. Cloud ice particle microphysical properties are generated randomly where statistical parameters are selected to mimic in situ measurement data closely. Atmospheric background fields from ECMWF are perturbed to account for variation on sub-grid scales. All these data, together with sensor characteristics, are fed into a state-of-the-art radiative transfer simulator (ARTS). The method was validated by a successful comparison with AMSU data. Copyright © 2007 Royal Meteorological Society