非定常性水文頻率分析方法之比較探討：以台灣地區年最大日降雨為例

臺灣近年來極端降雨事件日益頻繁，降雨分布有更加集中和強度增大的趨勢，以往傳統水文頻率分析多基於年最大水文序列一致性與獨立性的先驗假設，但全球暖化與土地利用等人為因素所造成氣候變遷的影響，使得此基本假設的適用性有所質疑，基於上述理由，以往傳統頻率工具所設計的結果可靠度需進一步的檢視，應提出非定常性之頻率分析架構以供因應未來之規劃所需。有別於國內研究大多僅分析第一階動差之非定常性，本研究以分配與趨勢鑑定（Identification of distribution and trend）原則分析臺灣地區九組具代表性的氣象測站（基隆、臺北、新竹、臺中、臺南、高雄、恆春、花蓮以及臺東）日雨量之年極大值，並以Akaike information criteria為依據，決定最佳模型。另外導入其它趨勢估計方法比較結果差異，除了原本的加權最小平方估計，再進行討論離散小波轉換與總和經驗模態分解等共三種理論基礎相異方法。研究結果發現若僅以線性模型假設趨勢函數，則三種估計方法所得到之結果相同，皆認定基隆、臺東和花蓮三站存在非定常性，此外根據配適的最佳模型，估算各站未來極端雨量之10、20、30和40年的重現期距改變量，用以檢討非定常性對於傳統回歸週期觀念的影響，考慮時間風險增量的概念，檢視改變量可以發現在此案例中，回歸週期與平均等待時間之衰退量相當，但是後者定義更適合應用在實務上，另外也顯示加權最小平方估計所得到之回歸週期相較於另外兩種估計法更為保守。

關鍵字

非定常性條件；水文頻率分析；回歸週期

並列摘要

Due to the climate changing, the hydrological stationarity, a fundamental component of engineering design and practice involves predicting or characterizing future conditions based on previous observation or record, could be inappropriate. We have been experiencing more intense and more frequent extreme hydrological events in recent. Under current climate changing condition, the stationary assumption and corresponding assessment approach need to be re-evaluated carefully. This study investigates the nonstationarity of annual maximum daily precipitation in Taiwan. Based on the concept of IDT (identification of distribution and trend), three different schemes are applied to analyze the precipitation data from nine major cities in Taiwan. These studies adopts, Weighted Least Square Method, Discrete Wavelet Transform Method, and Empirical Mode Decomposition, to explore the time variation of first and second statistical moments of annual maximum precipitation. From the analysis, we find that all the three schemes demonstrate clear nonstationarity in Keelung, Taitung and Hwalian. According the result, this study further discusses the change of exceedance probability and return period in the near future. As results, we can determine the hydrological risks, review the current management policies and engineering standards, and have a better long term planning in engineering.

並列關鍵字

Nonstationarity ； Hydrologic frequency analysis ； Return period

參考文獻

1. Bier, V. M., et al. (1999). "A survey of approaches for assessing and managing the risk of extremes." Risk analysis 19(1): 83-94.

2. Cave, B. M. and K. Pearson (1914). "Numerical illustrations of the variate difference correlation method." Biometrika 10(2/3): 340-355.

3. Coles, S., et al. (2003). "A fully probabilistic approach to extreme rainfall modeling." Journal of Hydrology 273(1): 35-50.

4. Cryer, J. D. and K.-S. Chan (2008). Time series analysis with applications in R, Springer.

5. Cunderlik, J. M. and D. H. Burn (2003). "Non-stationary pooled flood frequency analysis." Journal of Hydrology 276(1): 210-223.

國際替代計量

非定常性水文頻率分析方法之比較探討：以台灣地區年最大日降雨為例

全文下載

主題瀏覽