Ocean province classification using ocean colour data: observing biological signatures of variations in physical dynamics


Wayne E. Esaias, fax + 1/ 301 614 5644, e-mail wayne.esaias@gsfc.nasa.gov


We have used satellite colour data to classify ocean environments for monitoring interannual changes in the ocean. The unsupervised classification method is based on our observation that the frequency distributions of Coastal Zone Color Scanner (CZCS) annual pigment means and standard deviations are nonuniform and contain distinct clusters. The frequency distributions are used to objectively determine ocean areas with similar pigment statistical characteristics. A major separation between high variance, high pigment and lower variance, lower pigment waters is observed in terms of global ocean area. The ocean areas determined with our method reflect different bio-logical responses to variations in ocean physical dynamics. Pigment means and variances around the Joint Global Ocean Flux Study (JGOFS) Time Series stations are used as fiducial characteristics. Hawaii Ocean Time-series (HOT) station is associated with the low-variance portion of the global annual pigment distribution characteristic of the central gyres, but shows slightly higher mean and variance than the minima in the central Pacific gyre. The Bermuda Atlantic Time Series (BATS) pigment associations comprise a transitional region between the gyres and high-variance pigment areas, and circumscribe the HOT pigment associations. Together, these associations encompass 23% (HOT-like) and 48% (BATS-like) of the Northern Hemisphere open ocean. The Pacific regions delineated by the JGOFS station pigment-based patterns are similar to distributions described historically for Pacific zooplankton communities. Interannual variation for the northern hemisphere gyre area is on the order of by 10% for the 11/78–10/81 period.