Tuesday, 11 February 2003
Sea surface temperature determination during day-time using 3.9 and 11 micron top-of-atmosphere radiance observations
The frequent observations by geostationary satellites of evolving sea surface temperatures (SSTs) have important roles in understanding air-sea interactions, defining the lower boundary condition of the atmosphere for numerical weather prediction, and in ocean modelling and data assimilation. The sensor on NOAA's GOES-8 has been shown to have SST accuracy approaching that of AVHRR (predicated on high quality cloud screening), and we have already developed new radiative-transfer based coefficients for sea surface temperature retrieval, which are being implemented in the operational NESDIS product. GOES-12 and N to Q, however, will have the surface-sensitive channel at 12 microns reallocated to a lower-troposphere channel at 13.3 microns, leaving channels at 3.9 and 11 microns only for SST determination. This presents an additional challenge to day-time SST retrieval, since the top-of-atmosphere near-infrared radiance includes scattered / reflected solar radiance. Previously, the 3.9 micron channel has not been used for day-time SST to avoid bias from this solar 'contamination'. Two options are available for obtaining day-time SST. The first is single-channel retrieval, using additional information (e.g. forecast NWP fields) to estimate SST from the 11 micron brightness temperature. The second is retrieval using both channels, with the 3.9 micron radiance corrected for the estimated solar contribution. We present our investigation of this latter option. The solar contribution comes from atmospheric scattering and ocean-surface glint, the latter being concentrated around near-specular geometries. Much of the glint region is eliminated by cloud screening tests, but we show that the region of significant additional radiance from sun glint at 3.9 microns extends beyond the screened area. We then characterize the contribution by atmospheric scattering using radiative transfer modelling and empirical observations. The associated retrieval methods are under investigation and results will be available on the conference poster.