Phytoplankton assemblages exhibit strong variations at the mesoscale and sub-mesoscale. But the processes which govern this organization, and the influence of sub-mesoscale structures (front, eddies, filaments) are still not fully understood. This is in part due to the difficulty to retrieve synoptic taxonomic data on phytoplankton concentrations. A new method using a neural-network trained on in-situ data was recently developped (El Hourany 2018). It allows to retrieve the concentrations of different phytoplanktonic diagnostic pigments from satellite measurements and thus can provide a synoptic view on the phytoplanktonic assemblages. This method is used to compare observations from the Gulf-Stream with measurements obtained with a biogeochemical model of an idealized western boundary current. The variation in repartition of phytoplanktonic functional types and in diversity is studied. A distinction is made between frontal and background areas. Lagrangian diagnostic tools are applied along with surface velocity data in an effort to decouple the biologic response of the phytoplanktonic communities to a change in its environment and its horizontal advection by oceanic currents.