The uncertainty in hydrological model can be caused by various sources, such as the model input data, non-optimal model parameters and model structures. For regional studies, the model parameters and uncertainty of input data are acknowledged as the two important sources of uncertainty. Particularly the Imjin river basin has the unique geographical characteristics with river cross operation between North and South of Korea. The insufficient metro-hydrological information in the northern region may lead to significant errors on the flow simulations.
In the calibration and verification processes applied on the SURR (Sejong University Rainfall - Runoff) for rainfall - runoff simulations, it was found that the SURR simulated flows in the northern region of Imjin basin (Gunnam station) had the least accuracy as compared to the other two stations located in the southern area (Jeonkok and Jeogseong station) for all events (2007, 2008, 2009, 2010). This was due to the less precise calibrated parameters and the insufficient information of weather data in this station. Therefore the objective of the dissertation is to quantify the uncertainty of flow simulation based on GLUE (Generalized Likelihood Uncertainty Estimation) method model for the model parameter and precipitation input data in Imjin basin. To examine the uncertainty on streamflow simulation, two indexes are used: (i) p-factor, the ratio of the number of observations fall inside the uncertainty interval, and (ii) r-factor, the width of uncertainty interval.
The results showed that the uncertainties of the simulated flow in the northern station are always high for both parameter and input uncertainty estimations. This was caused by the variations of two parameters, ALPHABF and SURLAG in the Gunnam station were significant larger than those of the southern stations. Meanwhile, the interpolated rainfalls in this area were also less accuracy due to the extremely far distance to the observed rain gauges in southern area. This reason explains why the input uncertainty in the northern area of Imjin basin always higher than the southern area. In addition, the dissertation also stated that the parameter uncertainty in streamflow was also influenced by the different periods of the hydrograph. The peak flows had higher uncertainty than the flows at the beginning or the end of each event.