請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/7236
標題: | 結合關聯結構與卜松過程之乾旱分析架構 A Drought Analysis Framework based on Copula and Poisson Process |
作者: | Pei-Yu Wu 巫佩諭 |
指導教授: | 游景雲 |
關鍵字: | 乾旱分析,關聯結構,卜松過程,非定常性, Drought analysis,Copula,Poisson process,Non-stationarity, |
出版年 : | 2019 |
學位: | 碩士 |
摘要: | 乾旱是一種經常性與反覆出現的氣候現象。由於近年來氣候變遷與全球對於水資源需求的增加,我們經歷了更嚴重的乾旱。為了評估乾旱趨勢的影響,本研究旨在提出一個完善的分析架構,利於流量乾旱分析並模擬未來情境的乾旱特徵。基於本研究提出的關聯結構(Copulas)與卜松過程(Poisson process)之乾旱模型架構。首先,分析歷史水庫入流數據,估算最佳乾旱事件之參數,並定義乾旱事件。接著從乾旱事件中萃取三個乾旱特徵,分別為乾旱持續時間,乾旱缺水量以及乾旱發生時間。此三個乾旱特徵具有一定相關性,故本研究使用三維高斯關聯結構,建構隨機模型來模擬乾旱事件之發生,此外,為了更恰當地模擬乾旱事件,本研究進一步採用改良之卜松過程來描述兩乾旱事件間之間隔年份,此做法能有效地降低合成乾旱事件時的困難度。至此,上述分析過程與模擬結果皆是基於定常性假設,而為進一步考慮氣候變化的影響並分析乾旱特徵的趨勢,本研究於最後提出一個非定常性乾旱分析架構,能繁衍可信之未來乾旱事件。這項研究的結果可以應用在施政方針的參考、提供未來的乾旱政策或乾旱調節模式的檢驗,並在後續中、長期有更好的規劃。 Drought is a regular and recurring climatic phenomenon. Because of climate change and increasing water demand, we have experienced higher drought severity in recent years. To evaluate the impact of the trend of drought, this study aims to develop an improved framework for streamflow drought analysis and simulation of future conditions. For our proposed framework based on Copula and Poisson process, this study firstly analyzes historical inflow data to estimate statistical parameters of drought. Secondly, I defined three drought indices: drought duration, drought deficit, and the occurrence time of a drought event. It can be found that these three indices are found to be correlated. Afterwards, the marginal distributions of these indices are estimated. Accordingly, this study constructs a stochastic model using a three-dimensional Gaussian copula to simulate the occurrence of drought events. In addition, to more appropriately simulate drought occurrences, Poisson process is applied to describe recurrence year which is the integral number of years between two drought events. Until now the simulation of drought events is under the stationary assumption in current stage. Therefore, a consideration of the impact of climate change on drought analysis is included in this study as well. The randomly synthetic droughts with several possibilities are generated by the proposed robust non-stationary drought model. The results of this work are valuable and can be utilized to support the decision or policy making processes in drought management. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/7236 |
DOI: | 10.6342/NTU201902777 |
全文授權: | 同意授權(全球公開) |
電子全文公開日期: | 2024-08-15 |
顯示於系所單位: | 土木工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-108-1.pdf 此日期後於網路公開 2024-08-15 | 3.15 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。