A new algorithm to retrieve the sea ice concentration using weather-corrected 85 GHz SSM/I measurements
Zum Verlinken/Bookmarken: http://dx.doi.org/10.23689/fidgeo-415
Sea ice is a very important component of the climate system. While the Arctic sea ice extent has retreated during the past 20 years, it has remained constant in the Antarctic. In order to better understand the role of sea ice in the climate system in the context of global warming currently used coupled Atmosphere-Ocean Global Circulation Models have to be improved. This requires to know the sea ice concentration C for a long period for both hemispheres and at the best possible spatial resolution. Currently used methods to obtain C like the NASA Team (NT) algorithm are based on data acquired by the Special Sensor Microwave/Imager (SSM/I) at 19 and 37GHz. The SEA LION (SL) algorithm presented here allows to infer C from the polarization P at 85GHz taking advantage of the higher spatial resolution at this frequency. However, the decrease of P caused by the weather influence leads to an overestimation of C. Therefore, P is corrected using a radiative transfer model and atmospheric data taken from Numerical Weather Prediction models and/or derived from SSM/I measurements. The various sea ice and snow properties are considered calculating monthly sea ice tie points. The average standard deviation of C derived with the SL algorithm is 12% for C < 50% and below 5% for C > 90%. The SL ice edge agrees within 10km with the one evident in VIS/IR images. The SL ice concentration gradient across the marginal ice zone (MIZ) agrees much better with the one evident in SAR images compared to results of the NT algorithm. Using the higher spatial resolution at 85GHz the SL algorithm allows to detect smaller open water areas than known algorithms. A major limitation of the SL algorithm arises from the quality of atmospheric data needed for the weather correction. A spatial resolution of these data lower than the 85GHz SSM/I channels and/or a time lag larger than half an hour between both data sets can cause relative errors above 100%, particulary in the cloud-covered parts of the MIZ.