Why Do Water Masses Successfully Describe a Spatially-Complex, Process-Rich Ocean?
Why Do Water Masses Successfully Describe a Spatially-Complex, Process-Rich Ocean?
Abstract:
An inversion of physical and biogeochemical property distributions reveals that the sea surface fills the interior ocean according to a simple scaling law. More specifically, the Nth most important surface region (when discretized at 2 by 2 degree spatial resolution) contributes about 1/N times the volume filled by the most important box (reminiscent of the 80-20 statistical law in economics). This shallow power law appears to indicate that many thousand surface locations
all significantly influence the interior ocean, but traditional water mass analyses have been successful with just a handful of water masses. If optimal sets of water properties are chosen (i.e., water masses), how many water masses are necessary to explain oceanic spatial variability: just a handful or many thousand? The waters at any interior location are composed of a distribution of surface waters, and the optimal water masses will be the combinations of surface waters that recur the most often, usually due to homogenization in isolated ocean basins or geometrical constrictions that favor the passage of particular waters. In satellite image analysis, the mathematical problem of extracting the mixture of constituent materials in the visible spectrum, called spectral unmixing, is shown to be equivalent to the optimal water mass problem. Using a nonlinear optimization method borrowed from engineering, the ocean is spectrally unmixed and the most important water masses are identified, ranked by importance, and their formation mechanism is described. This method shows the maximum fraction of ocean variability explained by any given number of water masses. In particular, we find the number of water masses needed to explain 95% of ocean property variability, and reconcile the differences to the previous studies that suggested the number should be very small or very large.
all significantly influence the interior ocean, but traditional water mass analyses have been successful with just a handful of water masses. If optimal sets of water properties are chosen (i.e., water masses), how many water masses are necessary to explain oceanic spatial variability: just a handful or many thousand? The waters at any interior location are composed of a distribution of surface waters, and the optimal water masses will be the combinations of surface waters that recur the most often, usually due to homogenization in isolated ocean basins or geometrical constrictions that favor the passage of particular waters. In satellite image analysis, the mathematical problem of extracting the mixture of constituent materials in the visible spectrum, called spectral unmixing, is shown to be equivalent to the optimal water mass problem. Using a nonlinear optimization method borrowed from engineering, the ocean is spectrally unmixed and the most important water masses are identified, ranked by importance, and their formation mechanism is described. This method shows the maximum fraction of ocean variability explained by any given number of water masses. In particular, we find the number of water masses needed to explain 95% of ocean property variability, and reconcile the differences to the previous studies that suggested the number should be very small or very large.