Geocalibrating millimeter-wave spaceborne radiometers for global-scale cloud retrieval

Millimetre-wave radiometers will be on board of the future operational Eumetsat Polar System Second Generation (EPS-SG) satellites with the primary objective to support weather and climate models. These radiometers, and in particular the Ice Cloud Imager (ICI), will provide channels from 183 up to 664 GHz at a spatial sampling of 16 km, greatly enhancing ice cloud retrieval capability at global scale to validate and improve microphysics parameterization. At millimetre wave the emissivity of surface water bodies increases with frequency with values comparable to land targets so that the ICI brightness temperature contrast between water and land itself diminishes. This ICI feature implies that it is not possible to easily discriminate the surface coastlines, usually exploited for imagery geolocation purposes. This work proposes a methodology to evaluate the expected ICI geolocation error using the 183.3 GHz channel. Data from existing conically-scanning radiometers, such SSMIS (Special Sensor Microwave Imager Sounder), are used to emulate ICI observations. The idea is to extract Earth natural targets with an identifiable contour at 183.3 GHz to be compared with a reference one through a cross-correlation technique in the spatial and spectral domains. Such targets are quite rare and detectable only in dry regions/seasons. One of these regions is here found to be the Antarctic coastlines, which can seasonally change their contour due to temperature variations or ice collapses (creating iceberg). For these Antarctic targets, we propose the use of synthetic aperture radar (SAR) images as reference lines, sufficiently accurate for this purpose thanks to their high spatial resolution. To test this methodology the selected areas are the Ross and Ronne ice shelfs from May to September, the driest period in Antarctica. Ice shelfs are thick suspended platforms of ice with a sufficient brightness temperature contrast to extract the contour of the ice coastline. Using SSMIS data, results will be presented showing the geolocation error sensitivity to spatial resolution and target features. Open issues and future developments will be also discussed.