The ultimate goal of the prediction of Sea Surface Temperature (SST) from satellite data is to attain an accuracy of 0.3°K or better when compared to floating or drifting buoys located around the globe. Current daytime SST algorithms are able to routinely achieve an accuracy of 0.5°K for satellite zenith angles up to 53°. The full scan swath of VIIRS (Visible Infrared Imaging Radiometer Suite) results in satellite zenith angles up to 70°, so that successful retrieval of SST from VIIRS at these higher angles would greatly increase global coverage. However, the accuracy of present SST algorithms steadily degrades to nearly 0.7°K as the satellite zenith angle reaches 70°, due mostly to the effects of increased atmospheric path length. We investigated the use of T<sub>field</sub>, a gap-free first guess temperature field used in NLSST, as a separate predictor to the MCSST algorithm in order to clearly evaluate its effects. Results of this new algorithm, TfieldSST, showed how its rms error is heavily dependent on the aggressiveness of the pre-filtering of buoy matchup data with respect to Tfield. It also illustrated the importance of fully exploiting the a priori satellite-only information contained in T<sub>field</sub>, presently tamed in the NLSST algorithm due to the fact that it shows up as a multiplier to another predictor. Preliminary results show that SST retrievals using TfieldSST could be obtained using the full satellite swath with a 30% improvement in accuracy at large satellite zenith angles and that a fairly aggressive pre-filtering scheme could help attain the desired accuracy of 0.3°K or better using over 75% of the buoy matchup data.
The orthodox approach to designing an underwater imaging system with artificial illumination has been to consider
only the unscattered target photons as useable signal while looking at scattered photons as a nuisance to be mitigated.
Photons scattered from the target towards the receiver cause blurring of fine target details in the collected imagery, while
photons backscattered by the water column as the artificial source illuminates the target act as a veiling luminance that
reduces overall image contrast. Typical performance for the Laser Line Scanner and Pulsed Range-Gated imagers can
reach up to 6 attenuation lengths, which can still represent very short ranges in the turbid waters of coastal regions. In the
early 1970's, with the goal of extending these performance ranges, the Visibility Laboratory explored an unconventional
concept that was called imagery by means of Time Varying Intensity (TVI). TVI uses both scattered and unscattered
photons from the laser-scanned target as useable signal. This novel approach enabled high-quality imagery to be
collected over 20 attenuation lengths between the target and receiver. Although this system was eventually shelved, it
has been resurrected by using a modulated laser illuminator to communicate critical information about the laser scan to a
distant receiver via both the scattered and unscattered photons. With this knowledge, a high-fidelity image of target
detail can then be recreated. In this paper, a real-time interactive simulation of TVI's expected imaging performance is
presented and model predictions are compared with experimental imagery acquired when laser and receiver are both