Satellites that record imagery of the same sea surface area, at times separated by a few hours, can be used to estimate ocean surface velocity fields based on the apparent motion of patterns observed in a pair of images. Human interactive, statistical, model inversion, and feature correspondence methods have all been applied to this problem in the past. Previous methods used Advanced Very High Resolution Radiometer (AVHRR) data, which offered only long time separations, and geolocation inaccuracies that were often detrimental to the accuracy of the retrieved velocity vectors. Also, the previous methods were developed as scientific studies, and as such, require scientific sophistication or computing facilities that make them poor candidates for commercialization. This proposal addresses the development of a new method that uses genetic algorithms to minimize a cost function based on conservation laws and dynamical constraints. The method will utilize Moderate-resolution Imaging Spectroradiometer (MODIS) imagery that has important improvements over AVHRR imagery. Surface current estimates are important to forecasting drift of harmful algal blooms, oil spills, downed pilots, lost boaters, and free-floating mines. Many of these applications are crucial to decision support systems that NASA is currently supporting or investigating for future support.
