The Research on Summarizing Flood Forecasting Messages (II)

瀏覽統計 Email 通知 RSS Feed

簡歷

基本資料

Project title

Code/計畫編號

MOST103-2221-E019-036

Translated Name/計畫中文名

洪氾預警資訊之摘要化研究 (II)

Project Coordinator/計畫主持人

Ya-Hui Chang

Funding Organization/主管機關

National Science and Technology Council

Department/Unit

Department of Computer Science and Engineering

Website

https://www.grb.gov.tw/search/planDetail?id=8344057

Year

2014

Start date/計畫起

01-08-2014

Expected Completion/計畫迄

31-07-2015

Bugetid/研究經費

649千元

ResearchField/研究領域

資訊科學--軟體
防災工程

台灣常年因颱風與豪雨事件所帶來的降雨，釀成嚴重的淹水災害，因此在淹水發生前的預警措施成為重要的研究議題。我們在去年度的科技部計畫裡，提出了「基於地標的警示訊息」，也就是將水理模組所輸出的淹水預警區塊，以地標為主加以精簡，如此可使訊息簡短易於傳送，且容易通知相關人員疏散。去年度我們提出了C2L方法，事先記錄與地標距離小於最小距離門檻的演算區塊並建立B-tree，雖然能夠有效地產生摘要訊息，但是需要極大的前處理成本。本年度提出「VC」方法，根據Voronoi diagram的概念來切割空間，並利用R-tree索引來輔助查詢。為了減少該方法搜尋R-tree的次數，更進一步設計「VC+」方法，利用合併淹水區塊的方式來提昇效率。我們透過一系列的實驗以及各種資料集，評估這些方法的效率。結果顯示，在不考慮儲存空間和前處理的花費時，以C2L為佳。但是綜合空間和查詢效率來評估，則是VC+最佳。 The disaster brought by heavy rain has become more and more serious in Taiwan, and it has been an important research issue to provide warning messages before flood. In the project carried out last year, we proposed to summarize flooded cells based on landmarks, to produce short and clear messages. The method being proposed is called the C2L method, which pre-calculated the distance between cells and landmarks to preclude those cells whose distances are more than a given threshold. A B-tree index was also constructed to support the online querying process. Although that method was effective, it required a lot of preprocessing efforts. This year, we propose the VC method, which utilizes the concept of Voronoi diagrams to divide the space based on landmarks, and also uses the R-tree index to efficiently perform spatial join. We further design the VC+ method to improve efficiency by merging nearby flooded cells before traversing the R tree. We have implemented these methods and performed a series of experiments based on a variety of datasets to evaluate these three methods. Experimental results show that the C2L method is most efficient at the cost of space requirement and preprocessing efforts. On the other hand, the VC+ method has the best overall performance.

Keyword(s)

洪氾預警系統
地標
資訊摘要化
地理資料處理
最短路徑規劃
flood forecasting system
landmark
warning message summary
spatial data processing
shortest path planning

DSpace CRIS

The Research on Summarizing Flood Forecasting Messages (II)

基本資料

Description