| Summary: | 碩士 === 國立成功大學 === 水利及海洋工程學系碩博士班 === 95 === Estimating design flow in ungauged basins is a task frequently occurred in the design and planning of hydraulic engineering. Regionalization is a way to deal with this issue. In this study, a regional formula is established using gauged flows and basin characteristics in order to estimate the design flows in ungauged sub-basins within the homogeneous region. At first, the principal component analysis and the cluster analysis were used to classify the small sub-basins, delineated by this work, in Western Taiwan into three homogeneous regions. Due to the limitation of data availability and the region characteristics, only two homogeneous regions, Region 1 and Region 2, were used to build regional design flow formulae. Besides, the regional formula with respect to the area crossing Regions 1 and 2, named Crossing Region, and the formula regarding both regions, named Whole Region, were established to assess comprehensively the performance of the proposed regionalization method. Regional regression functions of design flow with respect to different return periods were developed, with topographic inputs of basin area, length of main stream, mean elevation and mean slope. Moreover, the return period is found to have good correlation with parameters of regression functions. Therefore, this study integrated these regression functions with different return periods into a single regional design flow formula for easy implementation by engineers.
Statistical tests demonstrate that the design flows are significantly related to the topographic variables at the 5% significant level, pertaining to formulae of Region 1 and Crossing Region. But the significant test on the formula of Region 2 shows insignificance, primarily due to insufficiency of data from gauged stations in Region 2. The mean absolute relative errors in the simulation pertaining to Region 1 and Crossing Region are lower than the regional formula with respect to Whole Region. The results of cross validation show that the regional formulae in Region 1 and Crossing Region have good applicability, and prove that the delineation of homogeneous regions can enhance the performance of regional formulae to estimate design flow.
|
|---|
