The role of physics based models for simulating runoff responses to rural land management scenarios
Ballard, Caroline Elizabeth
PublisherImperial College London
MetadataShow full item record
Recent floods in the UK have focused attention on the effects of rural land use and land management change on flood risk. Over recent decades agricultural intensification has been widespread across the uplands of the UK, with increases in stocking density, ploughing, reseeding and drainage of fields, use of heavy machinery, and the removal of trees from the landscape. A key scientific question is whether or not these changes in land use and land management in the uplands are increasing flood frequency and magnitude. Although land use and land management changes have been observed to change local surface runoff, attempts to isolate these responses at the catchment scale have failed due to limitations of data sets and modelling capability. While hydrological modelling is a well advanced field of science, a key methodological challenge that remains is how to upscale information about local scale changes. This Thesis evaluates the role of physics based hydrological models for upscaling local scale hydrological process knowledge and data to catchment scale flood flow responses. A model upscaling procedure that aims to quantify the changes in peak flows at multiple scales related to localised land use management changes is presented. The procedure divides the catchment into a number of runoff generating elements, which are each classified based on soil types and land management. For each runoff generating element, a physics based model is developed, incorporating understanding of hydrological processes and properties. This permits the investigation of local scale impacts, but cannot be applied at the catchment scale due to excessive computational burden. Therefore, the outputs from these physics based models are used to train simpler “metamodels”, which are then incorporated into a semi-distributed catchment model. In this way, the understanding of local changes in physical properties can be incorporated into a more flexible and computationally efficient catchment scale conceptual model. This procedure has previously been tested to a limited extent on a 12km² experimental catchment in upland Wales, which provided multi-scale hydrological data sets. The applicability of the procedure is now examined for a 25km² upland subcatchment of the Hodder in north-west England for an extended range of land management questions. This catchment is currently undergoing a number of land management changes, including: the blocking of open drains in the peatlands that cover the upper extent of the subcatchment, changes to an existing coniferous plantation and extensive deciduous riparian planting. The catchment does not include supporting multi-scale monitoring; without local data, physics based models are developed a priori using information from the literature, qualitative field observations and a proxy catchment. The significance of the uncertainties due to this lack of data and also uncertainties related to the upscaling procedure itself are explored, particularly examining the identifiability of the predicted effects at multiple scales. Based on the findings, the strengths and limitations of physics based modelling and the upscaling procedure in terms of ability to predict catchment-scale impacts of local land management interventions are assessed. The outputs from the multi scale modelling are also used to increase conceptual understanding of the hydrological processes and their relative importance under different land use and land management scenarios at the local scale, and also to quantify the impacts of land management scenarios at the catchment scale, taking into account the limitations of the modelling procedure.