在四種不同全球模式後報模擬中對流耦合赤道羅士比波之評估

Abstract

本研究透過拉格朗日觀點(Lagrangian view)來探討赤道羅士比波(equatorial Rossby waves)中的對流耦合過程，利用後報方法(hindcast)來評估四個擁有不同空間解析度與對流呈現方式的全球模式(TaiESM，SPCAM，CWBGFS，MPAS)模擬赤道羅士比波的表現。整組實驗使用ERA-Interim再分析資料以及給定海表面溫度作為模式的初始條件，在約40天的實驗期間內(2017/12/24-2018/02/04)進行每次進行15天的後報模擬。我們對地球長波輻射進行時空間濾波並利用追蹤演算法(tracking algorithm)來診斷實驗期間內出現的九個赤道羅士比波事件。觀測事件與模式模擬的相速度都大約在3-4 m/s，至於模擬的生命週期(7-8天)則是短於觀測中的週期(5-28天)。以觀測中赤道羅士比波事件的對流中心位置進行合成統計，濾波後的地球長波輻射及低層風場呈現赤道羅士比波特有的理論水平特徵結構，且降水與水氣的演進也與地球長波輻射同相位。模式的合成結果中赤道羅士比波的對流活躍區強度略弱於觀測，但在對流抑制區的地球長波輻射/降雨卻有較大的負偏差/正偏差。透過個案分析進行水氣收支計算，顯示事件中水平方向的水氣通量輻合是與降水平衡的主要過程。此外，當模式能夠模擬出足夠強度的水氣通量輻合時，也能產生強降雨事件。反之，在水氣通量輻合偏弱的情境下，模擬中的赤道羅士比波會在3天內減弱並消散。目前的結果顯示追蹤演算法能偵測赤道羅士比波的對流活躍區的位置及強度，也能夠利用此方法比較觀測與模式中強對流與波動耦合的差異。SPCAM與其他模式相比模擬出較多的波動事件，同時從合成場中SPCAM也展現出較強的赤道羅士比波訊號。未來也可以此組後報模擬為基礎對這些波動事件的發展過程進行更多敏感性實驗，系統性探討赤道羅士比波對於SPCAM二維雲模式架構的反應。

This study investigates the convective coupling of equatorial Rossby (ER) waves from a Lagrangian view, and we evaluate the simulated ER waves using the hindcast approach with four global models with different convection representation and spatial resolution, namely, TaiESM, SPCAM, CWBGFS, and MPAS. The hindcasts are all initialized with daily ERA-Interim reanalysis data and prescribed sea surface temperature (SST) over the 40 days (2017/12/24-2018/02/04), consisting of 15-day integration for each initialization. We apply a tracking algorithm to diagnose nine ER wave events in this boreal winter through the space-time filtered outgoing longwave radiation (OLR). The simulated ER waves have a reasonable phase speed (3-4 m/s) compared with observed wave events, while the lifetime of simulated waves (7-8 days) is much shorter than observation (5-28 days). The composite of filtered OLR and low-level wind obtained from the enhanced convection centers in observed wave events demonstrate the theoretically horizontal structure of ER waves, and the precipitation and column water vapor (CWV) is in phase with filtered OLR. The models show the slightly weaker intensity of ER waves in the convectively active regions from the composite, while they produce a larger negative/positive bias of OLR/precipitation in the convectively suppressed phase. From the moisture budget in the case study, it is found that the horizontal MFC is the main contributor to balance with the precipitation in the observed wave event. In addition, the models are able to produce strong rainfall events when a comparable amplitude of horizontal MFC is presented in the simulations, whereas in the weak horizontal MFC scenarios for the models, the ER waves quickly weaken and dissipate within 3 days. The current results show that the tracking method can detect the location and intensity of the convectively active phase in ER waves, and the coupling between deep convection in total fields and the wave dynamics for the observation and models can also be compared using this method. SPCAM produces more ER wave events than the other models, and it also demonstrates a strong filtered OLR signal as the observation in the wave composite. To systematically investigate the behavior of simulated ER waves with different configurations in the embedded CRM in the SPCAM, more sensitivity tests about the evolution of these wave events can be conducted in the future based on this hindcast experiment.