High frequency and mesoscale variability in SeaWiFS chlorophyll imagery and its relation to other remotely sensed oceanographic variables

Abstract

In its typical use for the study of large scale and relatively slow variability of phytoplankton biomass, ocean-color imagery is often binned in space and in time, and variability within the bin is discarded as noise. Since small- to mesoscale processes at time scales as short as a day may play a significant role in global cycles of carbon and nutrients, characterizing variability at these scales is necessary. With the first four years of nearly continuous daily imagery from the SeaWiFS instrument, we investigated patterns of variability at the mesoscales, operationally defined as that within a 2 � 2-degree neighborhood. We show that mesoscale variability of chlorophyll concentration (Chl) is high near the coasts, in dynamically active areas, and at the oligotrophic centers of subtropical gyres. High apparent variability over the oligotrophic ocean is a surprising contrast to the low variability in composite imagery at the same locations and may be due to increased relative noise at low mean Chl. Low correlation between pairs of images as little as 1 day apart in the oligotrophic ocean is consistent with a noise artifact, or alternatively may indicate that the observed variability is due to high-frequency phenomena. Spatial patterns of variability observed when data are binned into narrow ranges of mean Chl, suggest oceanographic origins. Patterns of variability in Chl and in sea-surface height have little correlation, suggesting that eddy pumping or turbulent diffusion along temporarily slanted isopycnal surfaces are not the major sources of Chl variability. The correlation between mesoscale anomalies of Chl and sea-surface temperature is not always negative as would have been the case if anomalies were produced mainly by the entrainment of colder, nutrientrich thermocline waters into the euphotic layer. Instead, we find roughly zonal bands of alternating negative and positive correlations determined by the relative directions of the background gradients of Chl and SST. Thus the most obvious influence of mesoscale motion on the distribution of Chl is advection of the existing gradients. Both long-term means and local anomalies of scatterometric winds from QuikSCAT are also correlated with mean Chl. Much of this correlation appears to be due to changes in the relationship between surface roughness and wind speed, brought on by factors like surface films, thermal stability of the air column, and surface currents. Our analyses show the feasibility of using ocean-color imagery to study mesoscale variability but also identify areas where there is room for major