This paper considers the erodible river corridor, which is the area in which the main river channel is free to migrate over a period of time. Due to growing anthropogenic pressure, predicting the corridor width has become increasingly important for the planning of development along rivers. Several approaches can be used to predict the future erodible corridor width but the results possess a large degree of uncertainty in all cases. The work presented here addresses prediction of the erodible corridor width of a reach of the River Irwell, UK, taking into account the uncertainty that arises from input parameters such as representative discharge, channel width, sediment and so on.The work adopts a probabilistic framework for assessment using Monte Carlo type simulations. Future river corridor width predictions, based on a model calibrated on past observations, are presented in a probabilistic manner using confidence levels. The results indicate the necessity of capturing input variability in the modelling process.Furthermore, the understanding gained from a relatively simple model used in a probabilistic framework is greater than a more complex one where only a few runs are feasible.

This paper considers the erodible river corridor, which is the area in which the main river channel is free to migrate over a period of time. Due to growing anthropogenic pressure, predicting the corridor width has become increasingly important for the planning of development along rivers. Several approaches can be used to predict the future erodible corridor width but the results possess a large degree of uncertainty in all cases. The work presented here addresses prediction of the erodible corridor width of a reach of the River Irwell, UK, taking into account the uncertainty that arises from input parameters such as representative discharge, channel width, sediment and so on.The work adopts a probabilistic framework for assessment using Monte Carlo type simulations. Future river corridor width predictions, based on a model calibrated on past observations, are presented in a probabilistic manner using confidence levels. The results indicate the necessity of capturing input variability in the modelling process.Furthermore, the understanding gained from a relatively simple model used in a probabilistic framework is greater than a more complex one where only a few runs are feasible.