臺灣國際港區船舶動態管理特性及颱風波浪資料補遺研究(四)

Abstract

本計畫延續96年度之研究結果，利用花蓮港港務局所提供之船隻動態記錄，從波浪之觀測與船舶動態資料分析，於本年度考慮考量中央氣象局所定義之颱風路徑來改善已建立之基本預警模式，成為較原模式穩定且精確高的船隻異動類神經網路模式，另外本模式考慮實際操作時之風險因素，本年度特別將風險分析應用於預報時之決策，同時也加強GUI模組，使得爾後港灣管理單位簡易的操作使用，可做即時的正確判斷與反應處理。另外96年度已建立之颱風波浪模式，並以此進行模式之測試，由測試結果可獲得較佳的補遺精度，依其結果評估可應用為颱風波浪實測資料遺失之補遺之用。此外，本年度計畫針對台北及安平港觀測波浪資料，進行統計分析與分類，以期能建立有系統之颱風波浪統計模式，進而能提高船隻預警管理之能力。

Based on the original ANN model that was developed in the first three years of a four-year sequential project by connecting typhoon’s paths and scales to conditions of ship escape from Hua-Lien harbor, the project in this year will modify the last model considering typhoon paths based on the definition of the central weather bureau. The corrected ANN model will be expected to be more accurate and stable than the original one. Risk analysis will be also applied to determine the possible predicted time delay at a confidence level of 95%. Furthermore, the model will be constructed in modules by GUI for easily operating the model by engineering staffs of Hua-Lien harbor bureau. Another aim of this project is to remedy the possibly missing tide and wave data during the period of an approaching typhoon due to instrument loss or damage. The missing data often happens at the time of the strongest typhoon that can cause the maximum wave heights. Thus, it is very important to evaluate the possible maximum wave height during the period of each typhoon, even when the instrument for waves is beyond work. The statistical properties, such as wave- height and wave-period their for both An-Ping and Taipei harbors will be also studied in this project. The well developed model in the future can provide the ships in the Hua-Lien harbor with duly alarms for escaping from severe wave impacts. The proposed integrated ANN alarm model can provide a good reference for harbor managements on ship mooring safety and is available for reducing harbor damages due to the typhoons.