University of Twente Student Theses


Sensitivity of channel-size estimations on flood inundation

Leummens, Leo (2021) Sensitivity of channel-size estimations on flood inundation.

[img] PDF
Abstract:Currently, almost 30% of total world population live in close proximity to the locations which are exposed to floods of 100 years return period (Rentschler and Salhab, 2020). The ongoing coast habitation combined with the climate change effect leads to the increased potential flooding damage both to human health and infrastructure. However, much of the damage caused by flooding could be alleviated using various safety measures which often include inundation modelling. More accurate, dense and widespread data of ground elevation and flood extents have enabled the rapid development of hydrodynamic modelling across the globe (Teng et al., 2017). Unfortunately, due to inability of modern remote sensing instruments to penetrate water, channel size measurement techniques have not benefited from these advances, therefore channel size approximations are widely used in hydraulic modelling in data-scarce regions. It is a common assumption that the river channel size or rather channel conveyance corresponds to the dominant discharge value (Wolman and Miller, 1960). Most commonly, for modelling purposes, the dominant flow rate is adopted as an annual maximum flow of 1-in-2 years frequency (Woodyer, 1968; Williams, 1978; Wilkerson, 2008). However, this is in spite of understanding that some rivers may flow out of bank seasonally (e.g. the River Niger) or much more infrequently due to human modifications (e.g. the River Thames in central London). In accordance with the defined problem and present knowledge and technology gap, the main research objective of this master thesis project is to increase the understanding and to quantify how different channel size estimates influence the performance of flood inundation modelling. In order to reach the objective of this research, the commonly used channel size estimation should be tested first. Thus, the LISFLOOD-FP model has been set up, calibrated and validated on 2 independent hydrology data series for December 1993 – January 1994 and January 1995 flood events. The value for 1-in-2 years annual peak flow was found to be equal to 6160 m3/s. After the 1-in-2 years annual maximum discharge was transited into bankfull depth approximation for river channel, the built model was used to perform a simulation run for the assessment of the adopted assumption. Good alignment of the model built on proposed bankfull discharge estimate with the measurement data was observed. The maximum deviation of the water depth data was limited by 5%, while the also only 5% error was made in prediction of the inundation area. Then, other channel size estimates were tested in a similar way to find which of them shows the best-fitted prediction for water level series over the Lower Rhine basin. Those assumptions for bankfull discharge value considered annual maximum and mean discharges of various frequency, regionally defined value, effective and half-yield discharge estimates. It was found that the best-fitted simulation results were obtained for the initial guess of 1-in-2 years annual maximum discharge as well as for the regionally defined discharge value obtained from the hydraulic geometry relationships. The latter was estimated to be equal to 6100 m3/s. Additionally, annual mean discharges of 10 and 50 years recurrence period were found to perfectly balance each other in terms of the river sections where their simulation run results have shown sufficiently good alignment with measurements. Thus, the symbiosis of those two assumptions could have also been judged as one of the best fitted estimates. As the result of this study, the means for channel approximation in modelling cases of rivers with relatively flat and densely populated large watershed areas with Western and Central hydrometeorological conditions were provided. The 1-in-2 years annual maximum flow and hydraulic geometry assumptions were considered best for such situations. Moreover, the sediment-related assumptions have shown really poor performance throughout this research, which could be related to the uncertainties introduced in sediment regional curves calculation by urbanization, climate change or other sources. This implies caution while using such type of channel size estimations in hydrodynamic modelling for areas similar to Lower Rhine basin. Finally, it was found that the annual mean discharge assumption performance is highly dependent on the average channel width and its variability along the specified river reach. Thus, it could be concluded that Qk,mean assumptions are too unstable to be used on large urbanised areas, where various aspects influence the river channel up to high extent, without any supporting parameters. The solution would be representing the large basin as a set of the smaller catchments with distinctive channel approximation techniques.
Item Type:Essay (Master)
Fathom, Bristol, United Kingdom
Faculty:ET: Engineering Technology
Programme:Civil Engineering and Management MSc (60026)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page