Deleersnijder, EricDelvenne, Jean-CharlesSemmah, AsmaaAsmaaSemmah2025-05-142025-05-142025-05-142022https://hdl.handle.net/2078.2/27704The goal of this master’s thesis is to cluster accurately the surface of the North Atlantic Ocean. For that, we constructed a Lagrangian flow network modeling the exchange of buoyant particles between subareas or boxes. Once we had our model, we selected two methods of clustering that suited well the particularities of the flow network : Infomap and Constant Potts model. We chose parameters for the clustering methods that gave us good quality metrics and then applied them onto the flow network. An extensive comparison of the clustering solutions is then done in order to assess the limitations of the two methods, in particular the resolution limit and the field-of-view limit.Flow-networkCommunity-detectionClusteringResolution-limitField-of-viewLagrangian-ocean-analysistext::thesis::master thesisthesis:37903