Estimation of uncertainty of rating curves and its impacts on the calibration of hydrological models

The subject of this thesis is to investigate the uncertainty which affects the study and the treatment of floods due to a series of mistakes and especially the presence of unsteady flow. It is true that in most hydrological and hydraulic applications, the study of floods is done considering a one-to-one relationship between stage and discharge and assuming that the flow is steady (Manning equation, 1-D, 2-D models for unsteady flow), as this consideration is easier to use, requires less computational time and is more conceivable. However, the above approach causes systematic error which tends to underestimate the peak flow causing problems in the design and construction of the flood protection works. In this study, it was examined a theoretical and a real problem. In the theoretical problem, three standard channels were designed in the 1-D hydraulic model HEC-RAS with fixed width 30 m and length of 10, 25 and 50 km in order to examine the unsteady flow. The study entails the examination, for various values of slopes, from mild to steep ones, for various values of Manning coefficient, corresponding to a river channel, and for various hydrographs, of the differences between the rating curves of steady and unsteady flow. In particular, it was investigated the underestimation of the peak flow when using the curve for steady flow and the uncertainty, surrounding the steady curve, due to the loop created by the curve of the unsteady flow. Also, it was examined the effect of the shape of the hydrograph in the uncertainty between the steady and the unsteady flow and the use of Manning and Chezy equations in the extrapolation of the curves. In the real problem, the Evinos River was chosen as the study area and specifically the section between the two gauging stations (Poros Riganiou and Bania Bridge). In this case, they were conducted the analysis described as follows: (1) the drawing of rating curves, (2) the calibration of the HEC-RAS for the section of the river between the two stations, (3) the calculation of the error in the extrapolation curves constructed with hydraulic methods and with the graphical expansion method in the already drawn curves, (4) the calculation of the standard deviation of the residual errors of the interpolation when drawing curves in order to perform a Monte Carlo analysis for a 95% confidence interval. Finally, it was examined the effect of the incorporation of uncertainty in the calibration of a hydrological model for the subbasin of Poros Riganiou. To this end, the conceptual model Zygos was used in daily time step. Initially, it was calibrated based on the values of the estimated flow from the rating curves on the outlet of subbasin, while in the second phase the model was calibrated based on the values of the flow counted the estimated uncertainty for the period from November 1992 to May 1995 in order to investigate whether the use of uncertainty plays a role in the process of calibration or not. In other words, whether the model is more flexible in the process of calibration, it will provide more robust solutions.