台灣地區在夏秋季節颱風所帶來的豪雨，加上本島地勢崎嶇，往往對河川沿岸居民造成生命財產損失。因此目前全國各主要流域已建置各種型態的防洪預測模式，然而因各模式之架構、開發軟體、使用介面皆不盡相同，一但模式研發人員在人事上有所異動時，後續的維護人員難以替預測模式進行更新。在本研究，將建立一防洪預測模式整合平台來統一規劃管理已建置的防洪預測模式，並將預測結果以統一的格式來呈現給使用者，藉由整合平台的幫助，在颱洪時期，提供一個穩定的洪水預報系統，幫助決策人員掌握即時水文資訊，以做出正確應變措施。 本研究中所架構的整合平台系統包含了三個主要部份:資料庫管理模組、自動化預測程式模組、使用者管理介面模組；其核心部份為資料庫管理模組，在資料庫系統裡，規劃了六個主要的表格來分別管理水文即時資料、流域測站、預測模式、使用者權限等相關資料。藉由資料庫的幫忙，以提升整合平台的資料處裡效率，當資料需要更新異動時，藉由新增模式介面，將預測模式相關訊息填入網頁表格，如此可以直接更改資料庫各資料表裡欄位所記錄的值。自動化預測程式則是擷取即時資料，載入資料庫中所記載的模式參數設定，以架構預測模式並執行預測，再將結果自動傳入使用者介面模組的網頁中，使用者可以隨時透過網際網路連結來觀看，不再受到地域及時間上的限制。經由此三大模組的運作，往後欲納入新的預測模式或是對現有模式進行更新維護時，使用者無需去修改預測模式本身複雜的程式架構，如此可節省大量的時間成本與人力、物力資源。 Typhoons and storms bring heavy rainfall that frequently cause downstream flood. For the purposes of providing the warning systems, a variety of flood forecasting models have been constructed for many main rivers in Taiwan. However, these models varied in the structures, programming languages and user interface that made many obstacles for maintaining and updating these models. Especially, few people can take over the models when the designers quit or be transferred to another job. It is a must for developing a uniform and integrated platform to maintain and update the present flood forecasting models. The integrated platform also provides a uniform display of forecasting results to help operators and decision makers. The proposed integrated platform includes three parts: database management module, automatic forecasting module and user management interface module. The kernel is database management module, which includes six tables to mange real-time hydrological data, stations, predicted stations, user priorities and so on. The automatic forecasting module is a forecasting model program, which can extract the corresponding parameters of flood forecasting models from database and automatically build the models. The user management interface module is to provide users an interface of a remote control through website. Through the linkage of these tree modules, we just add or modify the database and do not need to rewrite any program code when in adding a new model or updating the existing models. Hence, it would save much time, manpower and material resources and users can maintain the flood models or view the forecasting results whenever needed.
