| 摘要: | 台灣是個位於亞熱帶地區的海島國家，約有三分之二以上的面積屬於丘陵及高山地，地形變化起伏大，降雨的時間及空間分布上極不均勻。為了解降雨時空分布之非線性及非穩態等特性，本研究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)來進行時頻分析。有別於一般HHT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降雨特性探討，展現了探究水文時頻特性之另一可行途徑。
研究結果顯示，熵值確實為量化時頻能量分布集中或分散之良好指標。在長時間尺度之歷年降雨特性方面，根據荖濃溪流域各測站自1992年至2009年間之降雨參數分析結果，大部份測站之年累積降雨量、年降雨強度、最大1日及2日與3日降雨量等，近年皆有明顯上升之趨勢；而最大日降雨量越大使得整年降雨趨勢越趨於集中，造成熵值越低，顯示近年降雨分布趨於集中，故應多加注意短延時、強降雨之氣象事件，以降低可能災損。另外，在短時間尺度之颱風降雨特性方面，本研究篩選1996年賀伯、2004年敏督利、2008年辛樂克，及2009年莫拉克等四個颱風進行研究。根據颱風事件熵值變化與其降雨延時、降雨量及降雨強度關係之分析結果顯示，熵值與累積降雨量及平均降雨強度之關聯性較不明顯，但與降雨延時則呈現正相關性，亦即颱風事件降雨延時越長，強降雨時頻能量分布之區間越廣，致災可能性越大，因此未來應更加留意長延時、強降雨之颱風所帶來之可能危害。
Taiwan is an island located in the subtropical regions with more than about two-thirds of the area belonging to the hills and mountains. Because of the terrain, the temporal and spatial distribution of rainfall is very uneven. In order to explore the characteristics of rainfall with non-linear and non-stationary nature, a relatively new method of Hilbert-Huang transform (HHT) will be used to carry out the time-frequency analysis in this study. Traditionally, the HHT spectrogram is explained with the eyes. To improve the problems, the entropy that can measure the signal uncertainty will be utilized to quantify the time-frequency signal characteristics of rainfall. The combined HHT-entropy method is used to investigate the rainfall characteristics of the Laonong River Basin in the southern Taiwan in this study following.
The results show that the entropy is indeed a good indicator to quantify the time-frequency energy distribution with concentrated or dispersive trend. In the analysis of rainfall parameter for long-time scale, the rainfall data from 1992 to 2009 in the stations of Laonong River Basin are analyzed. It is shown that the year cumulative rainfall, annual average rainfall intensity, and the 1-day, 2-day, and 3-day annual maximum rainfall, etc., are obviously with rising trend for the majority of stations. Because larger maximum daily rainfall makes the year-round rainfall more concentrated, the associated entropy of rainfall is lower. It should be paying more attention to the meteorological event that is short duration with heavy intensity to reduce the possible disaster. On the other hand, in the analysis of rainfall parameter for short-time scale, four typhoon events, including the 1996 Herb, 2004 Mindulle, 2008 Sinlaku, and 2009 Morakot, etc. are evaluated. According to the analysis results, it shows that the relationship between entropy and cumulative rainfall and the average rainfall intensity is less obvious, but with positive correlation for typhoon duration. It means that the longer the typhoon duration, the wider the range of the energy distribution of the typhoon event with heavy intensity, and the greater the likelihood of hazard may occur. Thus, more attention should be paid to typhoon events with long rainfall duration and heavy rainfall intensity that may bring damage in the future. |
