濁水溪沖積扇水稻生產乾旱風險評估

Abstract

乾旱對台灣農業影響重大，特別是水稻種植區濁水溪流域，若發生乾旱，將威脅台灣的糧食供應。彰化縣因地理位置和地質特性，農業高度依賴地下水資源，但乾旱期間的過度抽取可能引發地層下陷，帶來其他風險。為探討濁水溪沖積平原乾旱與水稻產量之間的關係，本研究以氣象資料完備的彰化縣作為研究區域，運用聯合國糧食及農業組織開發的水分趨動作物模型 (AquaCrop)，模擬不同氣候情境下的水稻產量變化，並結合第六次氣候變遷評估報告 (AR6) 地球系統模式資料，計算乾旱指標評估未來乾旱事件的發生頻率、延時、嚴重度及強度，瞭解乾旱對水稻產量的影響，並探討可能的調適措施。

研究結果顯示，校正後之AquaCrop模型均方根誤差為0.71 t/ha；平均絕對誤差百分比為7%，能準確反映不同氣候條件下的水稻生長和產量。敏感度分析結果顯示溫度改變與產量呈負向趨勢，增溫將造成減產；降雨量與水稻產量呈正向線性高度相關 (R2=0.97)；潛在蒸發散量相較氣溫及降雨量兩變數而言對產量的影響更加劇烈，為水稻生產的高度敏感因子。此外，收割時間與產量呈線性高度正相關 (R2=0.99)，而於乾旱年適度增加灌溉次數能有效地增加產量。

於未來不同氣候變遷模式情境下乾旱特徵及水稻減產率在不同時期並無一致的變化趨勢，模擬結果顯示ACCESS-ESM1-5模式下，乾旱造成之水稻減產機率較高，特別是SSP370和SSP585兩種情境，減產機率接近或達到100%。相比之下，BMME模式對乾旱影響的模擬結果較為保守，減產機率較低。若乾旱發生於水稻需水期，如孕穗期間，抽取地下水灌溉可緩解乾旱對水稻產量的影響，但需加強配合水資源的永續管理策略。綜上所述，本研究建立之模擬方法，可提供氣候變遷下水稻產量所面臨乾旱風險之科學依據，有助於制定更有效的水資源管理與農業適應策略，以應對未來可能面臨的挑戰，保障糧食安全。

Drought significantly affects agriculture in Taiwan, particularly in the Jhuoshuei River Alluvial Plain, a key area for rice cultivation. Drought events in this region pose a serious threat to Taiwan's food supply. Changhua County, due to its geographical location and geological characteristics, heavily relies on groundwater resources for agriculture. However, excessive groundwater pumping during drought periods can lead to land subsidence and other associated risks. To investigate the relationship between drought and rice yield in the Jhuoshuei River Alluvial Plain, this study selected Changhua County as the study region due to its comprehensive meteorological data, and employed the AquaCrop model, developed by the Food and Agriculture Organization (FAO), to simulate rice yield variations under different climatic scenarios. Additionally, data from the Sixth Assessment Report (AR6) of the Intergovernmental Panel on Climate Change (IPCC) were used to calculate drought indices, and assess the frequency, duration, severity, and intensity of future drought events to understand the impact of drought on rice yields and explore potential adaptation measures.

The results showed that the calibrated AquaCrop model can accurately simulate rice growth and yield under various climatic conditions, with a root mean square error of 0.71 t/ha, and a mean absolute error percentage of 7%. Sensitivity analysis results revealed a negative correlation between temperature change and yield, indicating that increased temperatures would lead to reduced yields. Moreover, a strong positive linear correlation was found between precipitation and rice yield (R²=0.97). Compared to temperature and precipitation, potential evapotranspiration showed a more significant impact on rice yield, making it a highly sensitive factor for rice production. Additionally, harvest time appeared a strong positive linear correlation to yield (R²=0.99). During drought years, research found that increasing irrigation frequency can enhance yields effectively.

Under various climate change scenarios, the characteristics of drought and the rate of rice yield reduction do not consistently exhibit trends across different future periods. The ACCESS-ESM1-5 model predicts a higher probability of rice yield reduction under drought conditions, particularly under SSP370 and SSP585 scenarios, where the reduction probability approaches or reaches 100%. In contrast, the BMME model offers more conservative predictions of drought impacts, indicating lower reduction probabilities. Groundwater extraction for irrigation during critical water-demand periods for rice, such as the booting stage, can mitigate the impact of drought on rice yields. However, sustainable water resource management strategies must be strengthened.

In summary, the modeling approach employed in this study provides scientific evidence of the drought risks faced by rice yields under climate change. This approach supports the development of more effective water resource management and agricultural adaptation strategies to address potential future challenges and ensure food security.