Advanced applications of Ice Cloud Imager data

The Ice Cloud Imager (ICI) will be the first operational satellite instrument operating in the sub-millimetre region.

By grants from SNSA and other funding opportunities we are and have been highly active in the preparations for the sensor. This project ensures that the tools developed and experience obtained are actually applied to real ICI data. ICI will be launched roughly at the midpoint of the project period.

The main objectives are to provide the science community with high-quality cloud products based on ICI, and support usage of ICI for weather forecasting purposes.A general retrieval chain will be developed. It will be based on QRNN (quantile regression neural network), a machine learning approach introduced by us.

The quality of the retrievals is governed by the scope of the simulations used to train QRNN.

We will create a diverse, basic version for initial processing, that also can be used to support the planned operational retrievals at EUMETSAT.

The ambition is to quickly continue towards more advanced versions, that explore synergies with other instruments and make optimal use of footprint overlaps.

Besides the production and validation of retrieval products, the constraints ICI gives on particle property assumptions, either alone or together with other instruments, will be identified.

This work should lead to better particle models, which in its turn will improve ice hydrometeor retrievals in general.Improved particle models are of special concern for making use of ICI in data assimilation.

We are already today supporting ECMWF on this subject and this project will provide input for strengthening this collaboration. We are suggesting this as a PhD student project (in Sweden extending over five years).