Uncertainty sources of hydrological models and their impacts to the simulation of hydroelectric reservoirs

The complex nature of hydrologic activities make the hydrological planning of a hydraulic work (or a system of works) a demanding procedure, which is guided by the objectives, the requirements and the restrictions of each specific study, regarding the management and exploitation of water resources. The background of hydrology is intensely based on experience, and it is revealed through the applied procedures that may follow simplified approaches, under the regard of macroscopic view, or is based on experimental drainage basins, that are verified with standard hydroclimatic and geomorphological conditions. A additional factor that prevents the reliable planning is the lack of quantity and quality of filed information that are used for the calibration and the verification of simulation models. Therefore, the frontiers are continuously extended, on one side technologically, via the exploitation of computational systems and relative hydrologic and hydraulic models, and one the other scientifically, via the permanent research and search of improved methodologies, which incorporate if it becomes more realistically the interactive hydrologic components. The variability of the processes that are associated with the hydrological cycle introduces a probabilistic – statistical framework of approach, in a specified spatial and time scale of the hydrosystem under study. The time scale can range from a fine analysis, of few minutes and certain days, for storm and floods studies, to monthly or annually time scale, which is usually employed as a rule water resource management problems. Τhe drainage basin constitutes the most usual spatial unit of study. The main effort of the present study is to link the influence of the hydrologic uncertainties that come with the hydrologic parameters of a river basin model in the decision-making simulation model that is related with the operation of a hydroelectric reservoir. As study area was selected the hydroelectric system of Mesochora, in Acheloos river, and its upstream catchment. Basic core of the developed methodology constitute the two simulation models, one applied to the river basin and one to the hydroelectric system. The models have been implemented in a computer environment (MATLAB), while in the same environment was developed an innovative procedure for parameter optimization, also allowing for quantifying the sample uncertainty imposed by the time period of data. Specifically, by accounting the time series of areal rainfall, potential evapotranspiration and the observed runoff of the basin (during the period 1960-2012) as inputs to the hydrological model, we developed a calibration procedure that uses subsets of varying length. By solving the calibration problem across all the subsets, a wide range of alternative optimized parameter sets and the corresponding simulated runoff series were produced. Afterwards, the simulated runoff data were imported within the reservoir simulation model, key objective of which is the optimization of the reliability of the two water uses that are served by the reservoir, i.e. the energy production and the environmental flows. Then, we analyzed the correlations of the model parameters and model responses, via an original graphical method. From all the above analyses it becomes clear that the understanding of the impacts of hydrological uncertainties within the planning and the management of a hydroelectric system provides useful conclusions and tools of the design parameters, giving the opoortunity for further research in a research filed that is still in its early stages.