Synergy between remote sensing variables: Level 4 research products of Sea Surface Salinity and Chlorophyll-a

Abstract

Remote sensing imagery of the ocean surface provides a synoptic view of mesoscale signatures from different ocean scalars advected by the oceanic flow. The most probable origin of the observed structures is the turbulent character of the oceanic flow as they slowly evolve and are very persistent over time scales compatible with ocean mesoscale dynamics. At spatial scales of kilometers, turbulence is regarded as a two-dimensional phenomenon, with a complex geometry. Such complexity emerges in remote sensing images as filaments and eddies of different sizes. This is seen in images of surface chlorophyll-a concentration (Chl-a) and sea surface salinity (SSS), as well as the better-resolved sea surface temperature (SST) and sea surface height (SSH). A fusion technique has been recently proposed to exploit these common turbulent signatures between variables. This technique is theoretically based on the geometrical properties of advected tracers [Turiel et al., 2005b]. Coherent vortices in a turbulent flow strongly interact, leading to permanently stretch and fold small-scale filaments ejected from vortex cores, and generate small-scale tracer gradients between eddies. Therefore the spatial structure of a tracer inherits some properties of the underlying flow. This leads to an organized geometry of the flow as a hierarchy of fractal sets, called singularity manifolds, each one of them associated to a singularity exponent; this is the so-called multifractal formalism for fully developed turbulence. This geometrical arrangement of the flow is intimately linked to the energy cascade. A key point in this approach is the assumption of a multifractal structure in ocean images [Lovejoy et al., 2001]. It is assumed that singularity lines of ocean variables coincide [Umbert et al., 2013]. In turn, the gradient of both variables can be related by a smooth function. As a first and simple approach, the relating function is expressed as the identity, leading to a local regression scheme. This simple approach allows reducing the error and improving the coverage of the resulting Level 4 product of one variable using another variable as a template. Moreover, information about the statistical relationship between the two fields can also be obtained. This methodology is been applied to daily Aqua MODIS Level-3 chlorophyll maps using MODIS SST maps as template, to SMOS SSS using OSTIA SST as template, and to Aquarius SSS using SSH from AVISO as template. Resulting SSS and Chl-a Level 4 products contain the mesoscale structures seen in SST and SSH maps, exhibit a significant reduction of the uncertainty, and allow extrapolation to cloud-affected areas