OCEAN SURFACE ROUGHNESS SPECTRUM AND MICROWAVE BRIGHTNESS TEMPERATURE

After a long period of planning and development, global remote sensing of sea surface salinity (SSS) from space is becoming a reality with the launch of SMOS (Soil Moisture and Ocean Salinity) satellite by the European Space Agency (ESA) in November 2009 (Font et al. 2010). A similar mission (AQUARIUS/SAC-D), jointly sponsored by the United States National Aeronautics and Space Administration (NASA) and Argentine Space Agency, CONAE, is scheduled for launch later this year. The microwave L band centered on 1.413 GHz and with a bandwidth of 25 MHz is selected for soil moisture and ocean salinity retrievals because this frequency band is protected for use by radio astronomy. For this band, it has been determined that the ocean surface roughness is the leading geophysical error source for sea surface salinity retrieval (Lagerloef et al. 2008). The spectrum of ocean surface roughness relevant to microwave remote sensing remains poorly defined. Many roughness models have been published but very few intercomparison studies reported. The difficulty of specifying an optimal roughness spectrum is highlighted in the comparison between field measurements and analytical computations. For example, Yueh (1997) selects Durden and Vesecky (1985), denoted DV, for his two-scale model computation but finds it necessary to double the roughness spectral densities in order to obtain a reasonable agreement with field measurements. Camps et al. (2004) use the spectrum models of Durden and Vesecky (1985) and Elfouhaily et al. (1997) and also have to double the roughness spectral densities of both models to achieve approximate agreement between calculated and measured wind speed sensitivity of brightness temperature. Similar difficulty of specifying the ocean surface roughness spectrum is also encountered in active radar scattering computations. In a recent study, Hwang and Plant (2010) present a comparison study of calculated normalized radar cross sections (NRCS) using three spectrum models: Donelan-Banner-Plant (Plant 2002), Elfouhaily et al. (1997), and Hwang (2008), abbreviated as D, E, and H, respectively. The H spectrum is constructed from the parameterization function of field measurements of short scale surface waves (wavelengths between 0.02 and 6 m or wave number k between 1 and 300 rad/m) (Hwang and Wang 2004; Hwang 2005) following the analysis of source term balance by Phillips (1984). Analytical expressions are de-