Abstract. The second generation of the EUMETSAT Polar System (EPS-SG) will include the
Ice Cloud Imager (ICI), the first operational sensor covering sub-millimetre
wavelengths. Three copies of ICI will be launched that together will give a
measurement time series exceeding 20 years. Due to the novelty of ICI, preparing
the data processing is especially important and challenging. This paper
focuses on activities related to the operational product planned, but also
presents basic technical characteristics of the instrument. A retrieval
algorithm based on Bayesian Monte Carlo integration has been developed. The
main retrieval quantities are ice water path (IWP), mean mass height (Zm)
and mean mass diameter (Dm). A novel part of the algorithm is that it fully
presents the inversion as a description of the posterior probability
distribution. This is preferred for ICI as its retrieval errors do not always
follow Gaussian statistics. A state-of-the-art retrieval database is used to
test the algorithm and to give an updated estimate of the retrieval
performance. The degrees of freedom in measured radiances, and consequently
the retrieval precision, vary with cloud situation. According to present
simulations, IWP, Zm and Dm can be determined with 90 % confidence at
best inside 50 %, 700 m and 50 µm, respectively. The
retrieval requires that the data from the 13 channels of ICI are
remapped to a common footprint. First estimates of the errors introduced by
this remapping are also presented.