Potential for natural flood management and land management practices to mitigate flooding in upland catchments

Abstract

There is an increasing uptake of Natural Flood Management (NFM) and land use management (LUM) schemes to tackle excessive, rapid runoff in rural catchments. At the local scale, there is a growing knowledge base regarding the impacts of NFM and LUM. However, evidence and understanding of how these local impacts manifest at a larger catchment scale is less well understood. There are many types of model that have been used for investigating NFM and LUM impacts at larger scales (>10 km2), ranging from the comparatively simple lumped conceptual approaches to more complex, physically-based, distributed models. How best to represent NFM and LUM impacts in models is ambiguous. This thesis presents research into impact modelling of flood mitigation measures from the hillslope to the catchment scale, using the lumped FEH rainfall-runoff model and a novel physicallybased, distributed model, Juke. A Flood Impact Modelling (FIM) methodology is proposed for rapid impact assessment using the FEH approach; FEH hydrographs are generated for sub-catchments and routed to the outlet. The impact of changes in timing and runoff generation in specific sub-catchments on the downstream hydrograph can be investigated to inform catchment planning. The Juke methodology is designed to make best use of field observations and existing GIS datasets for parameterising the runoff and routing components. Juke uses some of the knowledge embedded in the FEH approach regarding the timing and runoff generation and applies it spatially. Juke is capable of emulating the FEH, but also allows consideration of spatial changes in LUM. Two catchments in the north of England have been instrumented to characterise the rainfall-runoff behaviour and understand what causes the largest flood events, where NFM and LUM have taken place. This knowledge informs the LUM and NFM scenarios explored as well as for model parametrisation. Results from the lumped FEH modelling suggest that the mitigation of flood flow by managing the volume and timing of fast runoff will have the greatest impact on floods caused by short duration, high intensity rainfall events. The Juke modelling also suggests that the impact of NFM and LUM is likely to be minimal (<10 % flood peak reduction for 12 % coverage of riparian woodland) and depends on the duration and intensity of rainfall events and the internal synchronisation of the component sub-catchments. The flood peaks for some events ii may increase due to the effects of timing and synchronisation of flows from the landscape elements. The outcomes of this thesis recommends flood managers make field observations to better understand the causes of flooding within a catchment. Schemes using NFM and LUM are likely to be most beneficial for comparatively small catchments (<10 km2) that suffer from frequent flooding from short duration, high intensity rainfall.