National-scale hydrological modelling of high flows across Great Britain

: multi-model structures, regionalisation approaches and climate change analysis with uncertainty

Abstract

National scale hydrological modelling frameworks are required to underpin effective water management in the face of large-scale pressures such as climate change. Many challenges remain for the application of national-scale models including: 1) developing and selecting appropriate model structure(s), 2) estimating model parameters for gauged and ungauged catchments, especially for models which require spatially distributed parameter fields, and 3) incorporating and communicating model uncertainties. This thesis addresses these challenges through a focus on modelling median and higher flows for large samples (hundreds) of catchments across Great Britain (GB) within a nationally consistent framework that includes predictive uncertainties. The first research chapter evaluates the predictive capability of multiple lumped, conceptual models for over 1000 catchments within an uncertainty framework, providing a performance benchmark. Regions where models often failed were identified (mountainous catchments in northeast Scotland, catchments overlaying aquifers in southeast England), and model performance was related to catchment characteristics to better understand where/why models fail. Significantly, it was found that despite substantial human modifications to catchments across GB, poor model performance was often linked to more general hydrological processes, such as low annual total rainfall, high baseflow contributions, and the water balance not closing. The second research chapter develops a parameterisation scheme to estimate nationally consistent parameter fields for a distributed hydrological model by relating model parameters to spatial geophysical data. This is applied within a novel framework for the inclusion of uncertainties when constraining spatial parameter fields. The resultant parameter fields performed well (non parametric KGE > 0.75) across the majority (60%) of catchments, enabling nationally consistent simulations across gauged and ungauged catchments, and reflecting hydrologically meaningful variation in catchment characteristics. The third research chapter applied this nationally parameterised model, to provide the first evaluation of climate change impact on river flows across GB to include both climate and hydrological model parameter uncertainties. This indicated an increase in the magnitude and frequency of high flows for catchments along the west coast of GB and across Scotland, albeit with large uncertainties especially across the southeast. Overall, this thesis improves understanding of model performance variation across Great Britain and where targeted model improvements are needed, contributes a national modelling framework which enables spatially consistent predictions across gauged and ungauged areas, and demonstrates how uncertainties can be included in larger-scale and large-sample hydrological studies. Whilst this thesis is focused on modelling across GB, the methods and conclusions can be transferred elsewhere to improve large-scale and large-sample model applications.

Date of Award	21 Jan 2021
Original language	English
Awarding Institution	The University of Bristol
Supervisor	Jim Freer (Supervisor), Gemma Coxon (Supervisor) & Thorsten Wagener (Supervisor)