With recent advances in dynamical system theory and Lagragian stochastic (LS) 
models, it is now possible to address the role of Lagrangian structure in 
transport and mixing. Observations or modeling result in sampled data of a given 
flow, thus small features beneath the level of observation or discretization, 
i.e. sub-grid scale motions, will not be captured. Several authors have explored 
the use of random walks to ascertain the effect of the sub-grid scale motions in 
oceanic applications ( e.g. Dutkiewicz et al. [1993], Lacorata et al. [1996], 
Buffoni et al. [1997]). However, these efforts were limited to modeling 
small-scale turbulence which is both stationary and homogeneous. Much 
theoretical work has been done recently (from Thomson [1987] to Reynolds [1998]) 
to overcome these limitations by developing random walks into more general LS 
models which can simulate the effect of nonstationary and inhomogeneous 
turbulence. Our work is focused on how to implement a modern LS model in a 
sub-grid scale context and how to merge it with the lobe dynamics approach, so 
that a unified framework to transport and mixing is developed which allows us to 
better understand the roles of coherent Lagrangian structures and turbulent 
diffusivity, i.e. dynamical systems theory tells us which particles to track, 
while LS models tell us how to track the particles.