透過您的圖書館登入
IP:34.231.247.254
  • 學位論文

利用WRF-LETKF同化系統探討掩星折射率觀測對於強降水事件預報之影響

Impact of Idealized GPS Radio Occultation Refractivity in predicting a Heavy Rainfall Event with the WRF-Local Ensemble Transform Kalman Filter System

指導教授 : 楊舒芝
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


隨著新一代的掩星觀測即將升空,如何利用同化技術有效地提升掩星觀測的使用效率以改善模式初始場乃至降水預報變成至關重要的課題。系集卡曼濾波器的背景場誤差協方差矩陣隨時間與流場而更動,故能比3DVAR更準確地評估背景動力的特性、更有效地使用觀測;且不似4DVAR需建立伴隨模式,因此便於實作。本研究利用WRF-LETKF(Local Ensemble Transform Kalman Filter)同化系統及觀測系統模擬實驗(OSSE)探討新一代掩星觀測對於中尺度同化系統及預報可能造成之影響,及如何利用系集同化技術更有效利用此觀測資料。   在OSSE實驗中,虛擬真實場為一類似SoWMEX 2008 IOP-8(06/14~17)真實個案降水分布之長時間模式積分。實驗設計共有三組,分別為掩星折射率的觀測數量敏感度測試、觀測誤差敏感度測試與多變量相關性測試。實驗結果顯示,雖然未同化任何觀測的系集降水預報無論分布與強度皆與真實場大相逕庭,但經過1至2日的同化後,系集預報便能捕捉與真實場類似的降水狀況,尤其是同化了掩星折射率的實驗隨觀測數量的增加得以更確切地預報出真實場的降水分布與強度,並能延長其與風場、水氣場等的預報能力達6至24小時。實驗結果亦顯示當觀測較精確時,雖然水氣場得以大幅修正,但風場與溫度場的分析與預報反而可能隨折射率同化數量增加而變差。此結果反應出過度密集的觀測資料可能影響分析增量中的變數平衡。透過控制多變量誤差協方差,可限制掩星觀測之貢獻。在本研究中,如掩星觀測僅調整溫,濕度場而不進行風場調整,結果顯示當風場在中高層(2.5km以上)得以優於使用完整多變量誤差協方差之分析修正。此顯示在中高層,掩星觀測之高準確度對於動力場的修正較無正面幫助。   從實驗結果來看,折射率的同化約需1至2天才能在模式中發揮效果(spin-up)。當觀測較精確時,大量的觀測反而可能透過實驗初期尚不成熟的多變量關係錯誤地修正背景場,但是經過適當地調整多變量關係的同化策略後此種窘境得以獲得改善。

並列摘要


In this study, we perform OSSE experiment to evaluate the impact of next-generation satellites GPS radio occultation for regional data assimilation and prediction, particularly focusing on an event with heavy rainfall in Taiwan. The evaluation is based on the accuracy of the analyses derived from the WRF-LETKF system and the following prediction.   For the OSSE experiment, the natural run is a 3-day simulation with a rainfall distribution similar to a real case (SoWMEX 2008 IOP-8) from the strong convection of the Mei-Yu front. There are three sets of experiments, including the sensitivity for observation density and observation accuracy, and the impact of using a multivariate covariance. Results show that the ensemble forecast can well capture the rainfall distribution similar to truth after spin-up time of LETKF. With the RO refractivity (REF), the wind and water vapor forecasts can be improved with a leadtime from 6 to 24 hours. With more accurate REFs, the advantage of REF data is clearly identified with even a moderate observation density with a resolution of 300km. However, we notice that even though the water vapor can be improved with the accurate observations, the quality of the analysis and forecast for wind and temperature is degraded because of the unbalance between variables. When restricting the impact of REF data to only the water vapor and temperature fields, the wind analysis becomes more accurate at mid to upper troposphere than the one using the full multivariate correlations. But, the effect on improving the heavy rainfall prediction is less clear.   According to this study, the WRF-LETKF assimilation system needs 1 to 2 days to spin up the impact of the REF data. The localization of the variables for the multivariate covariance can be a useful strategy to accelerate the impact of GPS RO data.

並列關鍵字

NWP data assimilation EnKF

參考文獻


Anderson, J. L., 2003: A local least squares framework for ensemble filtering. Mon. Wea. Rev., 131, 634–642.
Anthes, R. A., C. Rocken, and Y.-H. Kuo, 2000: Applications of COSMIC to meteorology and climate. Terr. Atmos. Oceanic Sci., 11, 115-156.
Barker, D. M., W. Huang, Y.-R. Guo, A. J. Bourgeois, and Q. N. Xiao, 2004: A Three-Dimensional Variational Data Assimilation System for MM5: Implementation and Initial Results. Mon. Wea. Rev., 132, 897–914.
Bishop, C. H., B. Etherton, and S. J. Majumdar, 2001: Adaptive sampling with the ensemble transform Kalman filter. Part I: Theoretical aspects. Mon. Wea. Rev., 129, 420–436.
Chen, C.-S., Y.-L. Chen, 2003: The Rainfall Characteristics of Taiwan. Mon. Wea. Rev., 131, 1323–1341.

被引用紀錄


Lin, K. J. (2012). 改善區域系集卡爾曼濾波器在颱風同化及預報中的spin-up問題-2008年颱風辛樂克個案研究 [master's thesis, National Central University]. Airiti Library. https://www.airitilibrary.com/Article/Detail?DocID=U0031-1903201314454828
邵彥銘(2015)。利用局地系集轉換卡爾曼濾波器雷達資料同化系統改善短期定量降雨預報: SoWMEX IOP8 個案分析〔碩士論文,國立中央大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0031-0412201512084876

延伸閱讀