|
This paper focused on analyzing different sources of uncertainties in design flood estimation. First, in the analysis of methodology uncertainty, Regional Synthetic Method (RSM) and L-moments Method (LMM) were used to calculate design rainfall depths, thus to estimate design flood. Also, three simple 24-hour-hyetographs were considered, with peak rainfalls at different times. Second, a Monte Carlo simulation based on Latin Hypercube Sampling (LHS) was employed to estimate the input and parameter uncertainty. A small ungauged basin in Zhejiang Province was used as an example. Results showed that for return periods more than 10 years, the peak flows derived through RSM were larger than that through LMM, indicating that RSM was more conservative and contained more uncertainties. Besides, small change of peak flows for different hyetographs could also be observed. With regard to input and parameter uncertainty, the model was run 1000 times and it was found that the peak flow distribution could be represented by the normal distribution. The statistical characters of simulated peak flow also presented that as the return periods increased, the coefficient of variation grew, slowly though, indicating increased uncertainty. |
|
Keywords:Hydrology and water resources; design flood; ungauged basin; Regional Synthetic Method; L-moments; Monte Carlo simulation; Latin Hypercube Sampling |
|