The paper compares observations from a small-scale model of a planned debris basin trapping a massive bedload deposition with results of a 2D depth-averaged hydraulic and morphologic numerical model. It is used as a validation case of its capability to reproduce flow and bed changes in this steep and laterally unconfined case with coarse sediment. The numerical model used a fractional transport approach enabling to compute grain size sorting and segregation processes. The fractional transport rates were computed with Meyer-Peter Muller equation. Water depth was computed with the “variable power equation” developed by Ferguson to account for both low, intermediate and high submergence (depth/grain size). Comparison of the numerical and physical models was done qualitatively (based on animations of the numerical results compared to videos of the physical model) and quantitatively computing an indicator of ratio of active channels (i.e., transporting sediment) measured using the large scale image velocimetry technique applied to time-lapse videos. The results show that fractional transport in itself does not bring significant improvement. However, dramatic improvements of the modelled channelization pattern and dynamics, both qualitatively and quantitatively, emerged when coupling the fractional transport with the friction law (through the active layer D84). In essence, when computing local grain size, it should be taken into account in the hydraulics too. This approach is a step forward to achieve better predictions of flow and channel dynamics on unconfined active morphologies and depositional systems.