| Summary: | 博士 === 國立臺灣海洋大學 === 河海工程學系 === 104 === As the development of remote sensing such as LIDAR (light detection and ranging), advanced measurement techniques have made high-resolution elevation datasets readily available in recent years. The DEM-based dataset has been applied to widespread interests especially for the establishment of rainfall-runoff models. Nevertheless, complicated procedures to cope with depressions in the existing DEMs are still criticized; on the other hand, different flow-direction algorithms have led to the variation on extracting the geomorphological properties as well as hydrological simulation results. Moreover, although the distributed simulation models are considered to be competent to capture the spatial and temporal variations of runoff transport by virtue of the employment of hydrodynamic equations, the high-resolution elevation dataset would decrease the computational efficiency. As to the aforementioned predicaments, this study focuses on: (1) developing a simple depression-filling method, (2) realizing the differences of geomorphologic factor extraction caused by using different flow-direction algorithms, and (3) establishing a highly efficient numerical model for runoff simulations.
This study proposed an intuitive concept to identify depressions and assign pseudo flow directions. The proposed method is readily applicable to either the raster or irregular datasets. The test results showed that the proposed algorithm can accurately identify the locations of depressions and efficiently assign pseudo flow directions across the filled depressions. Five most widely used flow-direction algorithms were implemented to investigate the distinctions of calculated geomorphological factors. A simple index was exploited to rank the effects of flow dispersion and flow concentration among the five algorithms. Moreover, this study developed an integrated numerical algorithm, in which a modified recursive formulation and a revised momentum equation were included, along with the change of routing sequence to permit a larger time increment for runoff simulations. The test results showed that the proposed method was not only with highly computational efficiency when applied to a steep area where the Courant-Friedrich-Lewy condition is quite limited, but also applicable to topographically complex watersheds or flat areas near stream networks where the numerical oscillation is frequently occurred.
|
|---|
