多年份反聖嬰事件對東南亞降雨與農業影響之探討：以芒果為案例

Abstract

隨著氣候變遷加劇，多年份反聖嬰事件的頻率與強度已出現變化，對東南亞地區的水文氣候與農業生產構成潛在威脅。然而，其背後的大氣動力機制及對高經濟價值作物（如芒果）的長期影響，迄今仍缺乏明確的理解。本研究探討多年份反聖嬰事件如何影響東南亞的水文氣候，並量化其對台灣芒果產量的影響。研究中利用 NCAR Community Earth System Model Version 2（CESM2），氣候模式結合大氣再分析資料與觀測降水資料，分析多年份反聖嬰的降水變化以及動力機制。模擬結果顯示，此類事件呈現第一年偏乾、第二年偏濕的「乾–濕」階段轉換特徵，與太平洋海溫配置及菲律賓氣旋異常有密切關聯。在升溫 2°C 的氣候情境下，模擬結果亦顯示暖化可能改變此種乾濕結構，需進一步評估對農業系統的潛在衝擊。另一方面，本研究亦透過經濟計量模型評估氣候異常對台灣芒果產量的影響。結果顯示，「第一年偏乾、第二年偏濕」的氣候條件對產量產生相反的效果，在假設對數線性關係的情況下，二月降雨量每增加 10 毫米，平均每公頃的芒果產量將減少約 0.5%。綜合農業生理知識推論，花期期間若降雨過多，可能導致授粉受阻，進而造成產量下降。最後，本研究分析 CMIP6 模擬資料中不同暖化情境下，台灣二月降水的變化趨勢。結果指出未來降水偏多的異常機率有上升趨勢，顯示芒果花期所面臨的氣候風險可能進一步加劇，需及早規劃調適對策。本研究結合氣候模擬與經濟分析，深入探討多年份反聖嬰事件下東亞地區「乾濕循環」的動態驅動機制，並量化其對芒果產業的衝擊。研究成果有助於提升對農業部門在當前與未來氣候變異下脆弱性的理解，亦可作為制定農業調適策略的重要科學依據。

As climate change intensifies, the frequency and strength of multi-year La Niña events have shifted, increasing potential threats to the climate and agricultural production across Southeast Asia. However, the underlying climate dynamics and their long-term impacts on high-value crops, such as mangoes, remain poorly understood. This study investigates how multi-year La Niña events influence atmospheric processes and rainfall in Southeast Asia and quantifies their effects on Taiwan’s mango yields. The NCAR Community Earth System Model Version 2 (CESM2) climate model combined with atmospheric reanalysis dataset and observed rainfall data are used to explore the dynamic mechanisms behind these events. Results reveal a distinct "dry–wet" phase transition, characterized by drier conditions in the first year and wetter conditions in the second. This precipitation pattern is associated with Pacific sea surface temperature patterns and the development of the Philippine Sea anomalous cyclone (PSCC). Further model simulations under a +2°C warming scenario suggest that such dry–wet structures may intensify under future climate conditions, with important implications for agricultural systems. Additionally, a panel data model is applied to assess the effects of climate anomalies on mango yields in Taiwan. The analysis shows that drier conditions in the first year of a multi-year La Niña tend to increase yields, while wetter conditions in the second year may result in excessive rainfall during flowering, which can disrupt pollination and lower productivity. To quantify this effect, we estimate that assuming a log-linear relationship within the observed range, a 10 mm increase in February precipitation is associated with an average reduction of approximately 0.5% in mango yield per hectare. Finally, projections based on CMIP6 simulations indicate an increased likelihood of positive precipitation anomalies across Taiwan in February under future warming scenarios. This points to rising climate risks during the mango flowering period and highlights the need for proactive adaptation strategies. By integrating climate modeling and economic analysis, this study reveals the dynamic drivers behind the dry–wet cycles associated with multi-year La Niña events in Southeast Asia and quantifies their impacts on Taiwan’s mango industry. The findings provide a scientific foundation for understanding agricultural vulnerability under both current and future climate extremes and support the formulation of targeted adaptation strategies.