| Summary: | 碩士 === 淡江大學 === 水資源及環境工程學系 === 87 === Mean rainfall in a river basin is generally very important and useful information for hydrologic design. Estimation of mean rainfall relies not only on accurate recordings of each raingage but also on the appropriate distribution of raingages over the entire river basin. To serve the latter purpose, Kriging method, developed from the theory of regionalized variables, has deen widely applied to raingage network design. The main purpose of this research is hence to study the effect on area mean rainfall estimation as Kriging method being applied to determine raingage removals.
Ten-day rainfall data from February to August were collected by the study from 25 raingages in Tan-Sui River watershed. Dimensionless variogram model of ten-day rainfall was implemented due to the time-varying nature of the rainfall data. To study the synthetic data, we first randomly generate the distribution of raingages with the variogram of exponential model and power model. Raingages are then removed successively from the sample in an order determined by the estimated Kriging variance. Area mean rainfall was estimated from both the arithmetical averaging method and Kriging method. This study then examines the estimation errors of mean area rainfall after the raingage removal treatment.
Results of this study indicate that the erring line of the mean rainfall estimated with raingage removal treatment varies significantly across different distributions of raingages over space. After classifying raingage distributions into uniform and non-uniform distributions, we find that the application of Kriging method for raingage removals improves the mean rainfall estimation in the case of non-uniform distribution of raingages. Furthermore, the erring lines of the mean rainfall from the arithmetical averaging method can be used to find the maximum number of raingages to be removed.
|
|---|
