機率淹水潛勢圖之研析

Investigation on Flood Inundation Probability Maps

賴彥維(Yen-Wei Lai)

指導教授 ： 張倉榮

國立臺灣大學/生物資源暨農學院/生物環境系統工程學研究所/碩士(2009年)

https://doi.org/10.6342/NTU.2009.10196

摘要

台灣的地理位置特殊，位於西北太平洋颱風行進路徑上，加上氣候屬於熱帶和亞熱帶交界之處，全年雨量充沛，夏秋之際時尤為明顯，屬於高淹水潛勢區域。在水文(降雨量)、水理(渠道曼寧粗糙係數與下游邊界)及地文因子(地表曼寧粗糙係數)本身所具有不確定性的情況下，淹水災害的影響範圍也會有變動值存在，根據災害防救法第22條第7款，為了減少災害的擴大，應該以科學方法進行災害潛勢、危險度及境況模擬之調查分析，並適時公佈其結果。目前的淹水潛勢之分析係採用定率之方式，給予淹水或者不淹水這兩種不確定性過高之答案，因此，為了改善此不確定性過高之情況，將不確定性觀念應用於機率淹水潛勢圖裡面，除了能夠告知會不會淹水之外，還能表示出各區域淹水的機率高低，若颱風豪雨發生之時，可以作為一個緊急疏散優先考量的參考依據。
本研究以台灣西南沿海低窪地區之嘉義縣荷苞嶼排水系統作為機率淹水潛勢圖實例分析，把具有不確定性因子之資料輸入值與參數進行敏感度分析。由分析結果得知，可將降雨量、曼寧粗糙係數以及下游邊界分成數個不同之區間，再以拉丁高次取樣法給予不確定性分析。經過拉丁高次取樣法衍生出之200組資料，帶入一套以一維排水渠道變量流與二維地表漫地流為主體之淹水模式，模式研擬統合了當地曾經歷過之洪颱事件及沿海低漥之水利特性，將淹水模擬結果經過計算之後繪製成機率淹水潛勢圖，可取代從前只能判定淹水或者不淹水的淹水潛勢圖，作為台灣水利機關往後設計防洪設施之參考規範，並可成為民眾自主防災的資訊來源。

關鍵字

不確定性 ； 拉丁高次取樣法 ； 淹水模擬 ； 機率淹水潛勢圖

並列摘要

Taiwan is located at a special point, which is in the path of typhoons from northeast Pacific Ocean. Taiwan is also situated in a tropical-subtropical transition zone. As a result, rainfall is abundant all the year round, especially in summer and autumn. For flood inundation analysis in Taiwan, there exist a lot of uncertainties in hydrological, hydraulic and land-surface topography characteristics, which can change flood inundation characteristics. According to the 7th work item of article 22 in Disaster Prevention and Protection Act, for preventing flood disaster being deteriorating, investigation analysis of disaster potentials, hazardous degree and situation simulation must be proceeded with scientific approaches. However, the flood potential analysis uses a deterministic approach to define flood inundation without considering data uncertainties. This research combines data uncertainty concept in flood inundation maps for showing flood probabilities in each grid. It can be an emergency evacuation basis as typhoons come and extremely torrential rain begin.
The research selects Hebauyu watershed of Chiayi County as the demonstration area. Owing to uncertainties of data used, sensitivity analysis is first conducted by using Latin Hypercube sampling (LHS). LHS data sets are next input into an integrated numerical model, which is herein developed to assess flood inundation hazards in coastal lowlands, base on the extension of the 1-D river routing model and the 2-D inundation routing model. Finally, the probability of flood inundation simulation is calculated, and the flood inundation probability maps are obtained. Flood Inundation probability maps can be an alternative of the old flood potential maps for being a regard of building new hydraulic infrastructure in the future.

並列關鍵字

Uncertainties ； Latin Hypercube sampling (LHS) ； Flood simulation ； Flood Inundation probability maps

參考文獻

16. 陳宜清等，2006年，探討於生態風險評估之不確定性，科學與工程技術期刊，第二卷，第一期。

19. 王嘉和，2007年，氣候變遷與地層下陷對台灣西南沿海地區淹水之衝擊評估，國立台灣大學生物資源暨農學院生物環境系統工程學系碩士論文。

30. Aronica, G., Bates, P. D., Horritt, M. S. (2002), Assesing the uncertainty in distributed model predictions using observed binary pattern information within GLUE, Hydrol. Process. 16, 2001–2016.

31. Bates, P. D., Horritt, M. S., Aronica G. and Beven K. (2004), Bayesian updating of flood inundation likelihoods conditioned on flood extent data，Hydrol. Process. 18, 3347–3370.

32. Chang, T.J., Hsu, M.H., Teng, W.S., and Huang, C.J. (2000), A GIS-Assisted Distributed Watershed Model for Simulating Flooding and Inundation, Journal of the American Water Resources Association, 36, No. 5, 975-988.