A comparison of rainfall-runoff modelling approaches for estimating impacts of rural land management on flood flows

There is a requirement for predictive tools to assist in land management and flood risk planning, and a variety of tools have been proposed recently. We compare four tools developed under various UK research programmes. The strengths and limitations of the tools are reviewed, model performances on historic data are assessed, and the methods are applied to estimating flood flows of 5- and 10-year return periods, and flow peaks under both recent land management conditions and speculative scenarios (grazing intensification and tree planting), using the Pontbren catchment, UK as a case study. Overall, the models agree on the direction of change, so that heavy grazing increases, and afforestation and tree strips decrease the flood flows. However, the estimated effects vary significantly between methods. It is concluded that method selection needs to carefully consider the type and scale of land management scenario being examined, and the sources of data available to support the modelling. Using an ensemble of suitable models is proposed as a useful way to represent a multi-expert opinion and to characterise the structural error associated with a single model.