Tuesday, 16 January 2007
Model analyses towards a robust and accurate physical-statistical Sea Surface Temperature (SST) algorithm for the use with the current and future satellite IR sensors
Exhibit Hall C (Henry B. Gonzalez Convention Center)
Sea Surface Temperatures (SSTs) are operationally retrieved from infrared measurements onboard polar and geostationary satellites. In the 1980s, Multi-Channel SST techniques were successfully tested with the Advanced Very High Resolution Radiometers (AVHRR) onboard NOAA satellites. A decade later, Non-Linear (NLSST) techniques were introduced. Both formulations were derived using physical, radiative transfer (RT) considerations. However their parameters (coefficients) are trained empirically against in situ SSTs using a limited match-up dataset early in the satellite mission, and then applied to retrieve global SSTs for the lifetime of a platform/sensor. Physical-statistical MCSST and NLSST-type techniques continue to form the basis of the operational SST retrievals from the current AVHRR/3 sensors flown onboard NOAA-16, 17, 18 and Metop-2 satellites and from the two MODerate-resolution Imaging Spectroradiometer (MODIS) sensors flown onboard the Terra and Aqua platforms. With the advent of the new sensors, such as the Visible and Infra-Red Radiometric Suite (VIIRS) and the Advanced Baseline Imager, to be flown onboard NPOESS and GOES-R, respectively, improvements to the MCSST and NLSST formulations are being explored. These analyses are expected to also benefit SST algorithms from the heritage AVHRR and MODIS sensors.
In this study, the development of improved SST algorithms is explored using theoretical RT calculations with MODTRAN 4.2. The focus is on the daytime retrievals using only split-window data in the spectral interval 10.5-12.5 μm. A simple RT model (monochromatic radiances, black surface, aerosol-free atmosphere, 6 standard atmospheres, nadir observations, Gaussian and non-correlated radiometric noise in bands) is used to concentrate on the analysis of the information content of multi-channel SST measurements in the presence of radiometric noise, and to determine the number of informative bands and predictors for SST retrievals. These preliminary analyses will be extended, to include more realistic ocean-atmosphere and sensor characteristics. Empirical analyses with AVHRR data are also underway to ensure consistency between the model and empirical SST retrieval results.