碩士 / 國立聯合大學 / 土木與防災工程學系碩士班 / 99 / Due to typhoon path in the Western Pacific, there were lots of serious disasters in Taiwan every year. During the typhoon period, because of the low air pressure and strong wind, the sea water moves to the land and causes the destruction of dike, resulting in inundation and impacts the safety of residents seriously. In the present study, the ANN models use hourly tidal data and meteorological forcing between 2007 and 2009 to predict the tidal level at the east coast including the Hualien harbor and Chengkung harbor and at the west coast including the Tanshui River mouth, Taichung harbor, and Kaohsiung harbor.
The input layer of ANN models includes the astronomical tide produced by harmonic analysis, air pressure, wind speed, and wind direction, whereas the output layer is the observed tidal level. The long-term tidal level and storm surge tide during the typhoon events was used for ANN training and testing. Three ANN methodologies including back-propagation neural network (BPNN), radial basis function neural network (RBFNN) and adaptive network-based fuzzy inference system (ANFIS) was adopted to predict tidal level in the coast of Taiwan. The data sets of tidal level from July to September, 2007 and October, 2007 were used for ANN long-term training and testing, respectively. The results indicate that the ANN models can faithfully simulate the storm surge tide during the typhoon KROSA occurred in October. ANN models also have the better predictions in astronomical tide than harmonic analysis method. The best predictions of tidal level among five stations are the Hualien and Chengkung harbors.
The data sets of storm surge tide during the typhoon events between 2007 and 2008 and in 2009, respectively, were served as ANN typhoon’s training and testing. The simulated results show that the best predictions of storm surge tide during the typhoon events are the Hualien and Chengkung harbors, especially during the period of typhoon MORAKOT. In general, the typhoon strength and path significantly affect the storm surge tide. However ANN models exhibit the better predictions in storm surge tide at tidal stations of east coast including the Hualien and Chengkung harbors. Because of the topographic effect, Taichung harbor is subjected to less surge height, resulting in bad prediction in storm surge tide during typhoon events. The results also reveal that BPNN and ANFIS models have the better predictions in tidal level than RBFN model. RBFN model predicts the large errors in peak compared to the observed tidal level during the typhoon events. Among three ANN models, ANFIS model can accurately simulate the astronomical tide and storm surge tide during training and testing phase for long-term and typhoon events.
| Identifer | oai:union.ndltd.org:TW/099NUUM1653003 |
| Date | January 2011 |
| Creators | Jian-Xiang Wang, 王建翔 |
| Contributors | Wen-Cheng Liu, 柳文成 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 116 |
Page generated in 0.0075 seconds