We propose a stochastic modelling framework to simulate the spatially lumped sediment and vegetation dynamics in a flood plain of a braided gravel bed river as an alternative to detailed morphodynamic modelling. The idea is that floods intermittently erode the river bed and expose sediment on gravel bars, while riparian vegetation continuously recolonises the exposed area. The exposed sediment and water stochastic dynamics can be analytically solved in terms of their probability density function (pdf), and later used to force a vegetation model, together forming a so-called coupled "Master-Slave" stochastic dynamical system in continuous time. We apply and validate the master model by using a historical record of aerial photographs of the Maggia River (Tessin, Switzerland) and daily streamfiow measurements. This approach is useful to statistically quantify the amount of sediment that is reworked by flood events, as well as the effects of changes in the flood disturbance regime on flood-plain dynamics.