HHT與熵理論應用於荖濃溪流域颱風降雨特性之分析

Analysis of Typhoon Characteristics for Laonong River Basin based on Hilbert-Huang Transform and Entropy Theory

羅凱文(Kai-Wen Lo)

指導教授 ： 黃富國

淡江大學/工學院/水資源及環境工程學系碩士班/碩士(2013年)

https://doi.org/10.6846/TKU.2013.01158

摘要

台灣位於北太平洋西部熱帶氣旋活躍之海域，近年的莫拉克颱風為台灣創下50年來最嚴重的水災，因此鑒於颱風對台灣造成的災損如此嚴重，實有必要對颱風特性有進一步的了解。本研究為了解荖濃溪流域颱風降雨時空分布之非線性及非穩態等特性，採用具有完全自適性，且適合處理突變訊號之希爾伯特－黃轉換(HHT)來進行分析。另有別於一般時頻圖使用目視判定法，本研究引用度量訊息不確定性之熵值(entropy)來量化水文訊號之時頻特徵，並將其應用於荖濃溪流域之颱風降雨特性分析，展現了探究水文時頻特性之另一可行途徑。
本研究根據莫拉克等10個颱風在荖濃溪流域之降雨特性分析結果發現，颱風雨量變化在熵值上所展現之特性亦是影響坡地災害發生因素之一。且透過颱風熵值與崩塌潛勢圖之套疊結果可得知哪些地區屬高危險區，並可以此做為一防災指標。在熵值與降雨參數之關係中可歸納出，颱風降雨延時和累積雨量越大則熵值就越高，以及熵值與降雨強度關係不明顯。而由熵值與颱風降雨分布之特性顯示，颱風熵值高是由於其降雨的不確定性高，也就是雨量變化劇烈；熵值低則代表降雨不確定性低，雨量較集中。雨型分析則顯示荖濃溪流域颱風降雨之主要雨型為中峰型降雨，而流域內多山坡地與丘陵，由於中峰雨型有易蓄積地下水之特性，配合上高熵值之影響，將容易導致坡地之穩定性降低，進而發生山崩、地滑與土石流等災害，值得注意。

關鍵字

荖濃溪流域 ； 颱風 ； 希爾伯特-黃轉換 ； 熵 ； 時頻分析

並列摘要

Taiwan is located in the western North Pacific–East Asian (WNP-EA) sector. Over the WNP-EA region, typhoon is one of the most important systems for producing heavy precipitation that will cause serious damages. The exploration of typhoon characteristics becomes an importment issue thus. In order to investigate the characteristics of typhoon rainfall with non-linear and non-stationary in nature, a relatively new method of Hilbert-Huang transform (HHT) will be used to carry out the time-frequency analysis. Traditionally, the HHT spectrogram is explained by visual inspection. To improve the problems, the entropy that can measure the signal uncertainty will be utilized to quantify the time-frequency signal characteristics of typhoon rainfall. The combined HHT-entropy method is used to examine the typhoon rainfall characteristics of the Laonong River Basin in the southern Taiwan in this study.
The result according to the rainfall characteristics of 10 typhoons in Laonong River Basin shows that the variation in entropy of typhoon rainfall is a good indicator for slope disaster prevention. By the overlay map between typhoon entropy isopleth map and landslide potential map, we can know where the high-risk areas are easily. The relationship between entropy and rainfall parameters can be summarized that the greater entropy comes the greater typhoon rainfall duration and cumulative rainfall will be, but the relationship between entropy and rainfall intensity is less obvious. According to characteristics of the entropy and the typhoon rainfall distribution, it shows that high entropy is due to the high uncertainty in rainfall, which means the rapid fluctuation in rainfall; and low entropy is due to the low uncertainty in rainfall, which means the rainfall is more concentrated. In addition, the analysis of rainfall pattern shows the main rainfall pattern in Laonong River Basin is peak at the center. This rainfall pattern has a characteristic the groundwater is easy to accumulate with high rainfall entropy, which in turn will easily lead to landslides and debris flows because of the reduction of slope stability. Thus, more attention should be paid on this situation.

並列關鍵字

Laonong river basin ； typhoon ； HHT ； entropy ； time-frequency analysis

參考文獻

40. 楊元鎮（2002），集水區可能最大洪流量之研究，碩士論文，中原大學土木工程研究所。

39. 陳振雄 （2010）， 應用希爾伯特-黃轉換之訊號濾波研究，科學與工程技術期刊，第 6卷，第1期，第75-84頁。

51. 藍嘉偉(2006)，利用HHT之EMD方法分析SSM/I資料估算之客觀指數與颱風強度年際變化關係，碩士論文，中央大學大氣物理研究所。

37. 陳志豪（2009），集集地震後荖濃溪流域崩塌發生特性分析，碩士論文，國立成功大學地球科學系碩博士班。

31. 李明熹（2006），土石流發生降雨警戒分析及其應用，博士論文，國立成功大學水利及海洋工程學系碩博士班。