University of Twente Student Theses


Impacts of climate change on flow composition using a model tailored to runoff components

Schreur, W.J. (2019) Impacts of climate change on flow composition using a model tailored to runoff components.

[img] PDF
Abstract:Over the past decades various studies have indicated that climate change and its accompanying hydrological impacts are a prevalent issue. Without proper measures the impact of floods and droughts can extend to various water dependent sectors including agriculture, forestry, fishing, hydropower and tourism. Generally, climate change impact on river discharge is assessed based on projected changes to the total runoff. It is however argued that looking exclusively at the total river discharge is not a sufficient indicator and additional insights in the runoff would be a step towards improvements on runoff modelling and gaining insight in climate change impacts. The goal of this research is to gain insight in the runoff components that together make up the total runoff and how to calibrate a hydrological model such that the skill at which the individual runoff components are simulated improves. These new insights will then be applied to calibrate a hydrological model tailored towards simulating runoff components and assess the impacts of climate change not only on the total runoff but also the impact it has on the runoff components and the composition of the total runoff. For this research, hydrological modelling will be done using the Hydrologiska Byråns Vattenbalansavdelning (HBV) model. Three Polish catchments: Biala Tarnowska, Dunajec and Narewka catchments are used as test cases in this research. To find if there is a relationship between the criteria that are used in model calibration and the skill at which the HBV model simulates the runoff components two objective functions have been used for model calibration. The two criteria or objective functions that are used for model calibration are the Nash-Sutcliffe efficiency criterion (NS) that emphasised high flows and the logarithmically transformed Nash-Sutcliffe (NSL) which emphasises baseflow. The two objective functions are used in a weighted manner by assigning a weight to each function to be able to later identify a trade-off between the objective functions. The weights assigned to both objective functions are varied between 0% to 100% at increments of 10% while keeping the sum of the weights equal to 100%. Calibration was done using the SCEM-UA calibration algorithm which resulted in 11 unique parameter sets for each catchment, each corresponding to a combination of NS and NSL function weights. The contributions of the runoff components are derived from the total runoff using the Wittenberg baseflow filter (Wittenberg, 1999). The skill at which the runoff components are simulated is assessed by using the NS as performance indicator and comparing the individual runoff components derived from simulations to those derived from the observations. Correlation between the skill at which runoff components are simulated and the NSL weight used in model calibration indicates that, for all three catchments, there is a significant correlation between the NSL weight and the HBV model its ability to skilfully simulate the baseflow component. A significant correlation between the NSL weight used in calibration and the skill at which the HBV model simulates the fast runoff component was only found for the Narewka catchment. Because the goal is to improve on the skill at which runoff components are simulated, the final calibration has been done using exclusively the NSL as performance indicator for model calibration. Using the HBV model that is calibrated for the three catchments using the NSL as performance indicator, climate change impacts on the total runoff have been assessed and additionally the impact that climate change has on the runoff components and the composition of the runoff are assessed. For climate change (re)assessment synthetic climate data is used from the EURO-CORDEX initiative. The datasets consist of precipitation and temperature simulations generated by seven combinations of General Circulation Models (GCM) and Regional Climate Models (RCM) Assessment of climate change impacts on projected flow resulted in findings that correspond to earlier modelling studies that have used the considered catchments as case studies (Osuch et al., 2016; Piniewski, 2017). The annual total runoff is projected to increase with the largest projected increase observed in the winter (December, January and February) and spring (March, April and May) periods. Separation of the runoff in the fast runoff component and the baseflow component indicated that there are differences in the composition of the total runoff between the three catchments for the observational period. When projected flows are assessed differences between the catchment are still present but differences between the simulations that are done using the GCM/RCM combinations are also present. These differences between the seven GCM/RCM combinations appeared to be consistent between the catchments and are assumed to be related to the individual GCM and/or RCM models that are used in simulating the climate data. Looking at the impact that projected climate change will have on the composition of the runoff indicated that the composition of the total runoff is not projected to change in the future for either of the climate models. Previously, climate change projections have shown that the intensity of precipitation evens is projected to increase which results in the expectations that more fast runoff will occur. Findings from this research however do not support this expectation. This might be caused by not taking into account changes that the catchments themselves may undergo over the upcoming decades. Wang & Cai (2010) have shown that the main driving force in changes in the flow composition is human interference (e.g. urbanisation or canalisation), an aspect that is not taken into account in this study. Another explanation for the absence of these changes can be found in the climate data that has been used in this research. Simulations from the GCM or downscaling to a RCM could have resulted in climate data that does not contain this increase in intensity but rather display a gradual uptrend in the precipitation. These possible causes however were outside of the scope of this research and have not been further pursued. In conclusion, there appears to be a significant relationship between objective functions that are used in model calibration and the skill at which runoff components can be simulated. This again shows the importance of selecting the most suited objective functions according to the research objective. But even while new insights have been gained in this area it displays that the impacts of climate change on river runoff might be restricted to changes in the total runoff after all. And that the composition of the runoff is not projected to change when compared to runoff simulations of the present situation. To conclude this research, several recommendations have been made that might be a worthwhile endeavour to pursue in order to be able to make these results more applicable wider variety of cases and to further improve on modelling runoff components.
Item Type:Essay (Master)
Faculty:ET: Engineering Technology
Subject:56 civil engineering
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