造成聖嬰現象相位轉換多變化性的機制

Abstract

本研究致力於探索控制聖嬰現象相位轉換以及衰退速率的機制，以進一步了解造成聖嬰現象持續期間長、或是短的根本原因。首先深入檢視「單年型」與「多年型」的聖嬰事件演變的完整過程，鉅細靡遺的比較兩者之間具代表性的對比特徵。分析的兩種資料為 (1)歷史觀測與 (2)大氣─海洋耦合全球環流模式的上千年模擬。從分析這些資料的特徵比較中接著歸納出可能造成聖嬰事件不同持續性的關鍵機制，並在一系列不同複雜程度的數值模式中進行實驗，加以驗證這些特徵造成氣候變異的過程之因果關係。實驗先分別只用海洋和大氣模式模擬各自的反應，拆分海洋與大氣的耦合過程以便闡明因果關聯，最後再加入完整耦合的模式實驗結果統合探討。 單年型與多年型的聖嬰現象之間明顯的差異包括：事件肇始早晚、發展中與巔峰階段的強度、巔峰階段暖海溫的東西位置分佈、以及衰退期間的海洋湧升凱爾文波。其中衰退期間的凱爾文波具有最強烈的對比，湧升的凱爾文波在單年型的聖嬰現象快速地終止事件，但是幾乎不存在於多年型事件。其所伴隨的地轉流距平，使赤道上的緯向海流距平在單年型聖嬰衰退時由往東轉往西，與氣候背景場的海溫梯度造成冷平流，主導此時海洋混合層的海溫冷卻趨勢。然而往東的海流距平在多年型聖嬰中得以持續造成暖平流，與海洋次表層熱容量共同維持多年型事件的暖相位。 聖嬰衰退期間的湧升凱爾文波的強度受聖嬰本身在發展至巔峰階段的風應力強迫影響，發展期氣旋式的風應力產生湧升羅士比波到西邊界反射回赤道上，而巔峰期西太平洋赤道上的東風應力直接驅動湧升凱爾文波。多年型聖嬰較晚發展，因而缺乏能造成湧升羅士比波的強氣旋式風應力；巔峰期的西太平洋東風距平則是被西北太平洋與印度洋的弱反應所限制，因為多年型弱而晚形成的聖嬰強迫不足以造成明顯的遙端氣候變異與回饋。 透過將海洋波動區分成風直接驅動、與邊界反射的效果，海洋模式實驗幫助澄清與量化聖嬰在不同期間的風應力強迫的作用。海洋的動力反應透過 (1)湧升凱爾文波突然阻斷正回饋作用以及 (2)反射波動造成的延滯負回饋在春季快速的翻轉單年型聖嬰現象的相位。而此二機制在多年型聖嬰皆較弱，結果顯示凱爾文波動的影響大於羅士比波，又風直接驅動的凱爾文波動比西邊界反彈的角色強烈，因此冬春二季在熱帶西太平洋的緯向風變異相當重要。 接著大氣模式釐清聖嬰現象發展時的強度與西太平洋緯向風在冬季突然反轉之間的關聯：較晚發展的聖嬰現象無法及早引起明顯的西北太平洋和印度洋的海氣通量回饋來驅動赤道西太平洋上的東風。最後完全耦合模式中的一組實驗，再次突顯西太平洋的耦合反應對聖嬰現象的回饋效果：原本由聖嬰快速衰退翻轉至反聖嬰相位的事件，在北太平洋海溫交互作用受抑制的實驗下，聖嬰衰退減慢且後續反聖嬰發展明顯消失。

This study delves into the mechanisms modulating the phase transition and decay paces of El Niño to advance our knowledge on the root causes determining the shorter or longer lifetime of an El Niño event. To begin with, the evolutions of “single-year” and “multi-year” El Niño events are examined in details and their contrasting features are comprehensively compared, by utilizing both observation and long-term, unforced coupled general circulation model (CGCM) simulation data. Then, the possible crucial mechanisms are deduced from these identified distinct characteristics. Finally, we validate the roles of these mechanisms in a series of numerical experiments, by separate simulations with the oceanic or atmospheric model only in order to first dissect the atmosphere-ocean coupled feedbacks, and then by a fully coupled simulation as well. The distinct features that contrast the single-year and multi-year El Niño episodes include: onset timing, the strength during the growth-mature stage, the center of warm sea surface temperature (SST) in the mature stage [i.e., central Pacific (CP) or eastern Pacific (EP) types], and oceanic upwelling Kelvin waves during the decay stage. Among them, the most striking feature is the upwelling Kelvin waves which rapidly terminate the single-year El Niño but are almost absent in the multi-year events. The upwelling Kelvin waves reverses the equatorial zonal current anomalies from eastward to westward with its geostrophic component, and cause cold advection with the climatological zonal SST gradient, which dominates the mixed-layer temperature cooling tendency during the decay stage. Conversely, the eastward equatorial zonal current anomalies continue, and maintain the warm zonal advection in the multi-year events together with the lasting oceanic subsurface heat content. The strength of the upwelling Kelvin waves is influenced by El Niño’s winds in the previous growth and mature seasons, as the cyclonic wind stress curls during the developing stage force boundary-reflected (BR) upwelling Rossby waves (RW), and the equatorial western Pacific (WP) easterly wind stress during the mature stage forces direct wind-driven (DWD) upwelling Kelvin waves (KW). The later developing cyclonic wind stress curls drives less prominent BR upwelling waves in the multi-year El Niño, and it is also less capable to induce the remote western North Pacific (WNP) and Indian Ocean (IO) responses to drive the WP easterly in the mature winter. The oceanic model experiments identify the impacts of wind forcing from the different stages of El Niño by discerning the DWD and BR waves. The phase transition of the single-year El Niño results from the sudden shutdown of positive feedbacks due to emerging DWD upwelling KW and the maximum damping due to BR waves; both are weaker in the multiyear El Niño. In comparison of the two El Niño types, the KW dominates the zonal current reversal over the RW, and the DWD component dominates the KW over the BR component, highlighting the role of equatorial WP zonal winds. Next, the effect of later or earlier El Niño onset is examined by atmospheric GCM-slab ocean model (AGCM-SOM) experiments. The late developing El Niño is shown to barely induce the remote WNP and IO responses to drive the WP easterly winds during the mature winter compared to the early developing El Niño. Finally, the fully coupled experiments further emphasize the role of WNP feedbacks, as the tropical WP easterly winds vanish and the El Niño-to-La Niña phase transition is aborted when the air-sea interaction is artificially suppressed over the North Pacific in a case study.