The outline of the colloquium will be as follows. In the beginning water challenge in Turkey will be introduced and the motivation for rainfall-runoff modelling will be presented. Research objective and research inquiries will also be mentioned. Secondly, Buyuk Menderes and subcatchments will be described. Afterwards, two rainfall-runoff models will be introduced. Additionally, necessary data and calibration techniques will also be mentioned. Following slides will be about results based on research questions and discussion of the results. Finally, the colloquium will be concluded with conclusion and recommendations.
Application of WALRUS and HBV-light in the Buyuk Menderes catchment, Turkey and the role of groundwater abstraction
Uneven water distribution, groundwater abstraction and lack of data are challenges concerning the water management in Turkey. Rainfall-runoff models can be used in order to provide missing hydrological data. In the present thesis, two lumped rainfall-runoff models were implemented: The Wageningen Lowland Runoff Simulator (WALRUS) and the Hydrologiska Byrans Vattenbalansavdelning (HBV-light).
The WALRUS was designed for lowland catchments having shallow groundwater table and it just requires five calibrated parameters. On the other hand, HBV-light was designed for cold catchments and relatively elevated catchments and it requires ten parameters including both catchment and vegetation zone calibrated ones. The main objective was to investigate the performance of rainfall-runoff models in an area with slight topography, dry climate, catchment size and groundwater abstraction and compare WALRUS to HBV-light. The Adiguzel subcatchment (2285 km2), which is the part of the Buyuk Menderes catchment in Turkey was used as a research area in order to answer the main research question. The Ikizdere (42 km2) subcatchment was also chosen for the comparison between WALRUS and HBV-light.
First, WALRUS parameters were calibrated for the current situation with groundwater abstraction and wastewater treatment effluent. After validation, an average Nash-Sutcliffe efficiency of 0.53 was found. When the same model parameters were used for the naturalised situation, without groundwater abstraction and wastewater treatment effluent, the model efficiency remained at 0.48. In order to ensure the effect of groundwater abstraction, the simulated discharges, the actual evapotranspiration and groundwater level variations were compared. After the end of the process, it was concluded that a) WALRUS underestimates observed discharge in two conditions, b) The effect of the groundwater abstraction and surface water supply is limited and c) There is not a single parameter set for the entire year.
Another research question was to compare two rainfall runoff models. WALRUS was calibrated for three different hydrological years, namely a) dry b) average and c) wet in terms of precipitation. After validation, it was seen that model parameters calibrated for dry conditions were the most optimal ones. Furthermore, it was found that WALRUS performs better when simulating peak discharges during just wet seasons. However, HBV can simulate better recessions, the timing of the peaks and the first peaks after a long dry period. In terms of Nash-Sutcliffe efficiency, average values of 0.56 and 0.72, using the WALRUS and the HBV-light respectively, were found for validation period in the Ikizdere subcatchment.
It can be concluded that the performance of the WALRUS is acceptable both in the Adiguzel subcatchment and Ikizdere subcatchment with provided data and assumptions. The performance of HBV-light is better in Ikizdere catchment. However, it is not easy to compare two models which have been designed for different conditions with different structure. Finally, more research should be conducted and the problems related to data should be solved to enable us to gain better results.