模式初始分析場資料與實際情況的差異，為造成模式預報誤差的原因之一。目前已經有很多觀測資料可供模式同化，例如衛星資料等，使模式初始場更接近真實大氣，但改善的程度不及加入虛擬渦旋的作用大。虛擬渦旋可以由經四維變分資料同化(4DVAR)來調整初始場，使此初始場包含受到模式動力條件限制的颱風渦旋，即颱風渦旋的虛擬資料同化(BDA)。本研究提出一個新的BDA方法，使用MM5 4DVAR經過BDA調整後的初始場，從中取得模式產生的渦旋的變數資訊，經三維變分資料同化(3DVAR)到WRF 模式裡，探討對侵台颱風路徑及強度預報的影響。這也使WRF能導入一個較傳統簡單的渦旋更為理想的平衡渦旋以進行預報作業及改善研究。 本研究目的主要評估此新的同化方法對颱風模擬的影響，針對珊珊颱風(2006)及聖帕颱風(2007)，進行72小時模擬。結果顯示同化4DVAR BDA渦旋的三維風場對颱風模擬大體上有最明顯的改善。此外，3DVAR風場似乎有往質量場調整的趨勢，因此3DAVR同化不合理的BDA溫度場，更容易產生一個不合理的渦旋的結構。本研究模擬顯示此資料同化使颱風的移動速度加快。這可能與經同化後較強的初始渦旋有關。對颱風強度的模擬，此同化顯示出明顯的改善。敏感度測試顯示，若只同化海平面氣壓，對颱風強度模擬僅有較為微弱的改善。加入其他變數場(風場、溫度場及濕度場)，對模擬並未產生一致性的改善。 The discrepancies between model initial analysis and the true state may contribute to the factors for model prediction errors. To adjust the initial fields for reducing the discrepancies, many data, such as satellite observations, are available for assimilation, however still showing less impacts as compared to those from insertion of a bogus vortex. Using a four dimensional variational method (4DVAR), the bogus vortex can be effectively assimilated into the model to adjusts the initial field under constraints of model dynamics, which is the so-called bogus data assimilation (BDA). In this study, we address a new BDA method which adopts the better balanced vortex from MM5 4DVAR than the traditional simple Rankine vortex and then applies the three dimensional variational method (3DVAR) to assimilate this vortex data into WRF to investigate the impacts on track and intensity predictions of typhoons impinging Taiwan. The target of this research is to evaluate the impacts of the new method on typhoon prediction. Two typhoons, Shanshan (2006) and typhoon Sepat (2007), were selected in this study, and they were simulated for 72 h. The results show that assimilation of the 3D wind of the virtual model vortex from 4DVAR in general have the largest improvement on typhoon simulation. Besides, the wind field tends to adjust to the mass field in 3DVAR and hence an unreasonable vortex structure may be produced when an unrealistic temperature from BDA has been assimilated as well by 3DVAR. The results also indicate that this assimilation tends to faster the typhoon movement, possibly due to the intensified 3DVAR vortex after the assimilation of the virtual vortex. For simulation of typhoon intensity, this approach may give significant improvement. Sensitive tests show this improvement, however, becomes much less when sea surface pressure was assimilated instead. Assimilation with the combined data (wind, temperature and moisture) in general does not lead to more consolidated improvement.
