Please use this identifier to cite or link to this item: http://theses.ncl.ac.uk/jspui/handle/10443/3290
Title: A robust multi-purpose hydrological model for Great Britain
Authors: Lewis, Elizabeth Ann
Issue Date: 2016
Publisher: Newcastle University
Abstract: Robust numerical models are an essential tool for informing ood and water management and policy around the world. Physically-based hydrological models have traditionally not been used for such applications due to prohibitively large data, time and computational resource requirements. Given recent advances in computing power and data availability, this study creates, for the rst time, a robust, physically-based hydrological modelling system for Great Britain using the SHETRAN model and national datasets. Such a model has several advantages over less complex systems. Firstly, compared with conceptual models, a national physically-based model is more readily applicable to ungauged catchments, in which hydrological predictions are also required. Secondly, the results of a physically-based system may be more robust under changing conditions such as climate and land cover, as physical processes and relationships are explicitly accounted for. Finally, a fully integrated surface and subsurface model such as SHETRAN o ers a wider range of applications compared with simpler schemes, such as assessments of groundwater resources, sediment transport and ooding from multiple sources. In order to develop a national modelling system based on SHETRAN, a large array of data for the whole of Great Britain and the period 1960-2006 has been integrated into a framework that features a new, user-friendly graphical interface, which extracts and prepares the data required for a SHETRAN simulation of any catchment in Great Britain. This has vastly reduced the time it takes to set up and run a model from months to seconds. Structural changes have also been incorporated into SHETRAN to better represent lakes, handle pits in elevation data and accept gridded meteorological inputs. 306 catchments spanning Great Britain were then modelled using this system. The standard con guration of this system performs satisfactorily (NSE > 0.5) for 72% of catchments and well (NSE > 0.7) for 48%. Many of the remaining 28% of catchments that performed relatively poorly (NSE < 0.5) are located in the chalk in the south east of England. As such, the British Geological Survey 3D geology model for Great Britain (GB3D) has been incorporated for the rst time in any hydrological model to pave the way for improvements to be made to simulations of catchments with important groundwater regimes. This coupling has involved development i of software to allow for easy incorporation of geological information into SHETRAN for any model setup. The addition of more realistic subsurface representation following this approach is shown to greatly improve model performance in areas dominated by groundwater processes. The sensitivity of the modelling system to key inputs and parameters was tested, particularly with respect to the distribution and rates of rainfall and potential evapotranspiration. As part of this, a new national dataset of gridded hourly rainfall was created by disaggregating the 5km UK Climate Projections 2009 (UKCP09) gridded daily rainfall product with partially quality controlled hourly rain gauge data from over 1300 observation stations across the country. Of the sensitivity tests undertaken, the largest improvements in model performance were seen when this hourly gridded rainfall dataset was combined with potential evapotranspiration disaggregated to hourly intervals, with 61% of catchments showing an increase in NSE as a result of more realistic sub-daily meteorological forcing. Additional sensitivity analysis revealed that the slight over-estimation of runo using the initial model con guration which has a median water balance bias of 5% was reduced in 62% of catchments by increasing daily potential evapotranspiration rates by 5%. Similarly, model performance was also found to improve by universally decreasing rainfall rates slightly, which together indicate the possibility of slight under-estimation of potential evapotranspiration derived from available data. In addition to extensive sensitivity testing, the national modelling system for Great Britain has also been coupled with the UKCP09 spatial weather generator to demonstrate the capability of the system to conduct climate change impact assessments. A set of 100 simulations for each of 20 representative catchments across the country were processed for a medium emissions scenario in the 2050s, in order to establish and demonstrate the methodology for conducting such an assessment. The results of these initial simulations suggest that higher potential evapotranspiration rates, combined with modest increases in rainfall under this climate change projection, lead to a general decrease in mean annual river ows. Changes in mean annual ow across the country vary between -26% to +8%, with the biggest reductions in ow found in the south of England and modest increases in runo across Scotland. This work represents a step-change in how the physically-based hydrological model SHETRAN can be used. Not only has this project made SHETRAN much easier to use on its own, but the model can now also be used in conjunction with external applications such as the UKCP09 spatial weather generator and GB3D. This means that the modelling system has great potential to be used as a resource at national, regional and local scales in an array of di erent applications, including climate change impact assessments, land cover change studies and integrated assessments of groundwater and surface water resources.
Description: PhD Thesis
URI: http://hdl.handle.net/10443/3290
Appears in Collections:School of Civil Engineering and Geosciences

Files in This Item:
File Description SizeFormat 
Lewis, E 2016.pdfThesis61.6 MBAdobe PDFView/Open
Lewis, E 2016.pdfThesis61.6 MBAdobe PDFView/Open
dspacelicence.pdfLicence43.82 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.