With rapid flood extent mapping capabilities, Synthetic Aperture Radar (SAR) images of river inundation prove to be very relevant to operational flood management. In this context, a recently developed method provides distributed water levels from SAR images. Furthermore, in view of improving numerical flood prediction, a variational data assimilation method (4D-var) using such distributed water level has been developed in Part I of this study. This method combines an optimal sense remote sensing data (distributed water levels extracted from spatial images) and a 2D shallow water model. In the present article (Part II of the study), we also derive water levels with a ±40cm average vertical uncertainty from a RADARSAT-1 image of a Mosel River flood event (1997, France). Assimilated in a 2D shallow water hydraulic model using the 4D-var developed method, these SAR derived spatially distributed water levels prove to be capable of enhancing model calibration. Indeed, the assimilation process can identify optimal Manning friction coefficients, at least in the river channel. Moreover, used as a guide for sensitivity analysis, remote sensing water levels allow also identifying some areas in the floodplain and the channel where Manning friction coefficients are homogeneous. This allows basing the spatial segmentation of roughness coefficient on floodplain hydraulic functioning.