In context of increasing observation of the earth by satellites and airborne sensors, this contribution investigates information extraction from those data combined with in situ data for hydrologicalhydraulic modeling of river networks. This work is based on recent multi-satellite datasets over two relatively large catchment river networks with contrasted and complex hydrological variabilities and flow features (anabranching reaches, confluences, tidal effects) : the Maroni River basin in French Guyana (SWOT Cal/val site) and the Adour River basin in metropolitan France. The dataset contains Multimission altimetry, optical and radar image of rivers in addition to in situ data and recent SWOT data. The forward modeling approach consists in dynamic shallow water flow models of river networks, in 1D following or multi-D inflowed by catchment scale hydrological models. The inverse modeling approach uses variational data assimilation and enables to optimize spatially distributed bathymetry-friction patterns to reduce the misfit to in-situ and satellite observables as well as to optimize simultaneously inflow hydrographs.