上層海洋熱力結構對颱風強度變化之影響─海氣耦合模式實驗研究

Abstract

儘管已有許多研究定性解釋了海表面冷卻、海洋暖渦及整個上層海洋熱力結構對於颱風強度的影響，但對於其影響程度目前仍缺乏明確解答。為了定量瞭解上層海洋熱力結構於實際個案中對於颱風強度的影響程度，本研究利用含有完整物理過程的海氣耦合模式，模擬具有明顯海洋特徵的辛樂克颱風（2008）。辛樂克颱風為T-PARC實驗的關鍵個案，共使用四架飛機進行聯合觀測，具有大量的投落送觀測資料。正式使用海氣耦合模式進行模擬前，我們特別使用新的EnKF資料同化方法進行颱風初始化，以建構出更接近觀測的初始颱風強度與結構。得到更合理的大氣初始場後，我們進行了不同初始海洋混合層厚度及不同初始海溫的高解析度敏感性實驗，以釐清耦合模式中混合層厚度及海溫對於颱風強度變化扮演的角色。本研究亦進行了模式中加入冷渦的實驗，以定量評估冷渦對於辛樂克強度變化的影響。 我們藉由海表面溫度回饋因子及混合層厚度回饋因子定量評估海表面冷卻的效應及不同暖水層厚度的效應，並利用修改後的渦流回饋因子評估冷渦對於海表面冷卻負回饋作用及颱風強度的影響程度。此外，我們新設計了和颱風最大潛在強度（MPI）相關的回饋因子，以評估上層海洋熱力結構對於颱風無法達到MPI的貢獻程度。 控制實驗結果顯示，考慮了海表面冷卻的實驗模擬出的辛樂克強度變化趨勢較接近觀測，其強度比未考慮海表面冷卻的實驗弱了約40％。多組敏感性實驗結果顯示，初始混合層厚度最厚的OC (ML80)實驗比其它幾組混合層厚度較薄的實驗有明顯較弱的海表面冷卻負回饋作用，模擬出的辛樂克強度較強。OC (ML80)實驗之海洋對於颱風無法達到MPI的貢獻程度較其它幾組混合層厚度較薄的實驗小。OC (ML80CE)實驗中的辛樂克於通過冷渦期間強度減弱了約15％，這段通過冷渦期間的強度變化量中，約有80％是由於模式中加入冷渦造成。OC (ML60CE)實驗中的辛樂克於通過冷渦期間強度減弱了約7％，但是在沒有冷渦的OC (ML60)實驗中，同時段之辛樂克強度卻繼續增強。 根據本研究的實驗結果，我們認為，海氣耦合模式結合以EnKF資料同化方法為基礎的颱風初始化有助於在實際個案中定量評估上層海洋熱力結構對於颱風強度的影響程度。藉由針對辛樂克颱風的研究基礎，後續研究亦可採用類似的方法模擬並深入探討其它颱風個案，以了解在不同的大氣環境之下，上層海洋熱力結構對颱風強度的影響程度是否有所差異。本研究的成果將有機會整合到2010年夏季的大型海洋觀測實驗ITOP，為ITOP建置良好的測試平台，後續研究可望利用數值模擬結果和ITOP的實際觀測資料進行比對與驗證。

Even though the roles of SST cooling, warm ocean eddies, and upper-ocean thermal structure (UOTS) on TC intensity have been qualitatively demonstrated in extant studies, impact of UOTS on TC intensity remains to be quantified. To extend our understanding more thoroughly and quantitatively, we use a comprehensive full-physics coupled atmosphere-ocean model to simulate Typhoon Sinlaku (2008), the case with special and significant characteristics of ocean. Besides, in the case of Sinlaku, unprecedented dropwindsonde data were obtained from four airplanes during T-PARC. In order to have a reasonable initial storm structure and intensity, a new method of TC initialization based on ensemble Kalman filter (EnKF) is applied before conducting the coupled model simulation. The high-resolution sensitivity experiments with different initial ocean mixed layer depth (ML) and SST are performed to identify the roles of these two ocean variables on TC intensity in the coupled model. The experiments with cold eddies are also conducted to quantitatively evaluate the impact of cold eddies on the intensity change of Sinlaku. The SST feedback factor and the ML feedback factor are calculated to quantitatively assess the effect of ocean cooling and different ML. The modified eddy feedback factors are also calculated to investigate the impact of cold eddies on SST feedback and TC intensity. Besides, a new MPI-related factor is defined to estimate the contribution of UOTS to prevent the typhoon from reaching MPI. It is found that the intensity of Sinlaku in the coupled run is closer to JTWC best track data and weaker than that in the uncoupled run by as much as 40%. In OC (ML80) run, the experiment with the deepest initial ML, the intensity of Sinlaku is stronger, and the SST feedback is weaker than that in the other experiments with shallower initial ML. It is also shown that the contribution of ocean which would prevent the typhoon from reaching MPI in OC (ML80) run is less than that in the other ML runs. In OC (ML80CE) run, the storm intensity decreases by about 15% when Sinlaku passes a cold eddy, and 80% of the intensity change in this period is likely due to the presence of cold eddy. In OC (ML60CE) run, the storm intensity decreases by about 7% when Sinlaku passes a cold eddy, while Sinlaku intensifies in the same period in the simulation without cold eddy. From the results of this study, we believe that synergy of TC initialization based on EnKF data assimilation and a coupled atmosphere-ocean model can help us quantitaively evaluate the impact of UOTS on TC intensity. More typhoon cases could be simulated and analyzed with the method used in this study, and the impact of UOTS in different atmospheric environments could be investigated in follow-up research. In addition, results from this study would set up a solid basis for the follow-up researches with the field program of ITOP in 2010. It would be a good chance to use in-situ ocean data to validate the results of coupled model simulation.