Study on Stream Level Forecasting by Using the GMDH Model Coupled withDistance~Level Relation Model in Storm Period

• Main Authors: Yi-Lin Shieh, 謝易霖
• Other Authors: Pei-Hwa Yen
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/14000782015254278800

碩士 === 國立成功大學 === 水利及海洋工程學系碩博士班 === 91 === Because of urban over-development, shorten time of concentration, unequal rainfall distribution in space and period and the steep slope of streambed in Taiwan environment, heavy storm usually cause serious damages both in human life and properties. Take the Herb typhoon (occurred in July 1996) for example, which caused the most serious disaster included almost all over Taiwan area since 1961. Hence, flood prediction, control and mitigation in order to prevent the suffering from flood and reduce the flooding calamity are the imperative duties of flood control authorities. Levees construction along the flood plain and stream flowing level prediction by hydrologic models cooperated with the flood warning systems could be the solutions for flooding prevention in engineering hardware and software aspects. Most of stream level prediction model need stream velocity, depth, slope, cross section area et. which could only be obtained by field surveying and other uncertainty parameters such as Manning, eddy viscosity, momentum and Chezy coefficients as the model input. The predicted results might get fair estimations with accuracy poor field measuring data and the uncertainty parameters. A framework based on GMDH (Group Method of Data Handling) is proposed in this paper to establish the I/O model as the alternative by using the relatively simple field measuring water/tidal level and rainfall data as the model input to predict prior 1 to 6 hours’ water level of specific river during storm period. The update stream level and rainfall data were collected to organize the Sequential GMDH modified model to match the time variant properties in stream level forecasting steps. The distance-level relation model based on the Muskingum formula and range data simply measured from a paper map were established in this paper also. The GMDH stream level forecasting model as mentioned above cooperated with this distance-level relation model could be used as the coupled model for prior 1 to 6 hours level forecasting at any specific stream sections. Finally, Tam-Shui River Watershed has been chosen as the case study to verify this GMDH stream level forecasting/distance-level relation coupled model. Data length for model construction could be determined by trial and error procedure. The GMDH stream level forecasting model then, was calibrated by Herb and other 5 typhoon events. The calibrated RMSE of 4 hours prior is less than 10% compared with the measured data and the variations of discharge peak time and peak level are within 1 hour and 10％ respectively. Only 72 hours measurement data of stream level and rainfall were collected to construct the GMDH model that means this model can be applied in less data-measuring area for level prediction purpose. The optimum GMDH stream level forecasting model established by a single storm event (Zeb typhoon) was applied as well to forecast another 5 typhoon events. The forecasting results then coupled with distance-level relation model to predict the prior 1 to 4 hours water level of Taipei Bridge station which assumed as the data nil station in Tam-Shui River.　Analysis results show that the forecasting errors are within 30cms both in rising and regression lamb of the hydrograph and obtained satisfied forecasting results.