Multiresolution stochastic processing of Topex/Poseidon oceanographic altimetry

Remarkable advances in remote sensing are producing measurements on increasingly finer grids, offering the possibility of a far better understanding and characterization of the ocean surface. The sheer quantity of measurements has rendered traditional data smoothing/assimilation algorithms obsolete however, even on high performance workstations; the basis of this paper is the application of an extremely fast multiresolution smoothing framework to ocean altimetry data. Using the previously developed multiresolution framework, a simple multiscale model is developed which captures the characteristic power-law spectrum of the ocean surface. Smoothing altimetric data and computing error variances on a grid of 512 by 512 points is accomplished in about 10 seconds on a SUN Sparc-10. Finally, further applications of this framework within the oceanographic context are discussed, including model identification and anomaly detection