Flood inundation phenomena typically occur over reach lengths of 5- 30 km and incorporate a number of complex flow mechanisms. These include a momentum transfer between the main channel and floodplain and turbulent mixing caused by the delivery of water to the floodplain from the channela nd its subsequenrte turn. However, currently available one dimensional schemes applicable at scales appropriate to floodplain inundation processes cannot effectively simulate such processes. This is due to both an incomplete description of the flow physics and a failure to treat floodplain areas in realistic fashion. More complex two and three dimensional models, which have these capabilities, have only been applied over very short reach lengths (c. 0.5 -2 km) and rarely to compound meandering channels. This thesis reports on the further development of a generalized two dimensional, finite element code (RMA-2) to meet this research need. This is achieved via a series of modifications to the numerical model and to the physical representation by finite elements that enable river channel/floodplain flow at the long reach scale to be effectively simulated. Evaluationo f the enhancedR MA-2 schemef ollows a three stages trategy. Firstly, the assumptions underlying the scheme are examined to identify possible inconsistencies. Secondly, tests are undertaken to assess whether the specified physical model has been correctly transferred into computer code. This is achieved via sensitivity analysis, examination of numerical stability issues and investigation of model response to abnormal parameterization. Thirdly, model predictions of flow field information are compared to observed field data in the context of an application of the enhanced model to an 11 km reach of the River Culm, Devon, UK. Results from this evaluation process indicate that the enhanced RMA-2 model is capable of simulating main channel/floodplain momentum transfer and the two dimensionale ffects associatedw ith compoundm eanderingc hannelsa t this scale. Model simulations compare favourably to field data, both for specific cross sections and over the entire mesh. Finally, extension of this core modelling capability is begun via the development of two model application scenarios. These demonstrate the likely utility of the enhanceds chemef or the assessmenotf flood risk and the investigationo f sediment depositionp rocessesin floodplain systems.