The upcoming EU Copernicus Imaging Microwave Radiometer (CIMR) satellite mission will measure Earth’s polarimetric microwave emission at five different frequencies between 1.4 and 37 GHz. It is the first time that this frequency range will be available on one satellite platform allowing simultaneous observations. Retrievals of surface and atmosphere parameters from measurements at these frequencies have a long-standing tradition. They provide cloud and daylight independent surface observations with full daily coverage of polar regions. For example, the more than 40-years long time series of sea ice area, retrieved from 19 and 37 GHz observations, forms one of the longest directly observed climate data records that exist today. Retrievals of sea surface temperature (SST) and water vapor have a similar history. However, current satellite microwave radiometers suffer from their coarse spatial resolution of at maximum 10 to 50 km depending on frequency. CIMR with its 8 m diameter deployable antenna reflector will be a major step forward and will offer spatial resolution better than 5 km at 37 GHz, better than 15 km at 7 GHz, and better than 50 km at 1.4 GHz.
Here we provide results of a multi-parameter retrieval using optimal estimation (OE). Measurements from the AMSR2 and SMOS sensors, which together cover the same frequency range as CIMR but at lower spatial resolution, are used to demonstrate the approach. The eight parameters (1) sea ice concentration, (2) thin ice thickness, (3) multi-year ice fraction, (4) ice surface temperature, (5) integrated water vapor, (6) liquid water path, and, over ocean, (7) wind speed, and (8) sea surface temperature and their uncertainties are retrieved simultaneously and in a consistent way. The accuracy of sea ice concentration is similar or better than current single-parameter retrievals. And the multi-parameter approach allows a smooth and physically consistent transition of all parameters from the open ocean to dense pack ice. For example, the “open ocean” parameters wind speed and SST, while having lower accuracy in our approach compared to single-parameter retrievals, can be retrieved close to the ice edge and in low ice concentration areas. The dominant frequency of 1.4 GHz for the thin ice thickness retrieval is currently measured from a different platform (SMOS) while measurements for all the other frequencies are from AMSR2. This poses some challenges for the consistency of brightness temperature measurements for the current AMSR-2/SMOS multi-parameter demonstrator product that will be resolved with the launch of CIMR.
Another challenge is the highly variable surface emissivity of sea ice, which depends on many factors including snow properties, salinity, roughness, temperature and more. Currently the sea ice emissivity variability is only considered in dependence of ice type and frequency dependent penetration depth. Here, we will present first results of an inclusion of a sea ice and snow emission model (MEMLS) in the OE forward model and its inversion. By this a higher, more realistic, variability of sea ice emissivity is obtained. The inclusion of a snow and ice forward model also allows to include snow depth as a 9th retrieved parameter in the OE scheme and preliminary snow depth results will be shown.
The quality of the retrieval for the single parameters will depend on how linearly independent they are from each other in the radiometric brightness temperature space the 10 channels (5 frequencies at vertical and horizontal polarization) span. Some parameters are radiometrically quite correlated and an information content analysis will show how much information, and thus trust, we can have in each retrieved parameter.