The binary characterization of the surface is valid only if the observations allow this to be possible and the discrimination of sea ice and open water can be done unambiguously. This assumes that within the footprint of the sensor, the radiances from 100% ice and 100% liquid water are distinct enough to make it possible to estimate or infer from the data the concentration on a measurement by measurement basis. This is generally true for microwave data (and also for visible, infrared which are usually used for validation studies) but it is not always the case and there are at least two situations in which there is a special concern, namely: (a) new ice/thin ice regions as in polynyas and marginal ice zones during autumn and winter and (b) melt/meltponding regions as in Arctic basin during the spring and summer. In new/thin ice regions, the emissivity of sea ice is not well defined since it changes constantly from close to that of liquid water to that of thick sea ice. Other physical processes associated with the growth of sea ice that may affect the radiative signature are described in detail in Weeks and Ackley (1986), Tucker et al. (1992), and Eicken et al. (1991). Also, when the surface temperature goes above freezing in spring and summer, the brightness temperature of the surface goes up as the emissivity increases to almost that of a blackbody as liquid starts to form around the ice crystals, but as the snow continues to melt and forms slush and then melt ponds, the brightness temperature goes down to that of an ice free water (Eppler et al., 1992).
