Uncertainty assessment in projection of the extreme river flows, the case of Ömerli catchment, İstanbul

Thesis Type: Postgraduate

Institution Of The Thesis: Middle East Technical University, Turkey

Approval Date: 2015

Thesis Language: English

Student: Batuhan Eren Engin



The average temperature at the surface of the Earth has been increasing over the past century due to the increased greenhouse gases concentrations in atmosphere through anthropogenic activities. Rising temperature leads to an increase in evaporation and thus intensifies the components of water cycle which results in extreme flows in different parts of the world through changes in globally averaged precipitation. Projection of extreme flows is very important in this aspect, yet obscurity about many factors that would influence future climate causes great uncertainty in climate variable prediction. Therefore, in this research, it is intended to quantify relative contribution to the uncertainty in extreme flow (high and low) projection change for the future period by two factors, the parameterization of a hydrological model and temperature and precipitation inputs from fifteen different Regional Climate Models (RCM), for Omerli Catchment area, in Istanbul. The uncertainty due to the precipitation and temperature inputs is investigated by using 15 different RCMs and also by applying two different statistical downscaling (SD) methods to the RCMs outputs for the vi reference (1961-1990) and future (2071-2099) period, “Bias Correction in Mean” and “Change Factor”. The uncertainty due to the hydrological parameterization (HP) of the hydrological model is assessed by using 25 different parameter sets generated by Monte-Carlo simulation technique using 5 different Nash functions as the objective function during the calibration of the hydrological model, which are NSE_Normal, NSE_BL, NSE_p3, NSE_Viney and NSE_Weighted. Observed daily precipitation and temperature records are provided by Turkish State Meteorological Service for the period 1961-2004, while daily discharges are obtained from State Hydraulic Works for the period 1978-2004. In converting the precipitation and temperature from RCMs into discharges, the HBV Hydrological model is used, which is calibrated to the period 1978-1985 and validated for the period 1986-2004. Main finding is that the relative contribution to the uncertainty by the temperature and precipitation inputs from different regional climate models is greater than the uncertainty caused by hydrological model parameterization in prediction of extreme high flow events in each data type: Original RCM, BC and CF methods. BC and CF increase the total mean variance in the changes in extreme high flow and low events from reference to future period. It is found that the observed dominant high flow events mostly occur during autumn/winter season for the reference period. Simulations using Original RCM, BC and CF data overestimate the seasonality index for the reference period. For the future period, simulations by parameter files using BC and CF data projected that the seasonality in high flow events will be stronger than their reference period, which means that in the future the Omerli catchment would likely to observe more dominant high flow events in autumn/winter.