評估氣候變遷下有效雨量對農業用水管理之衝擊

Abstract

臺灣雖有高年雨量的水文條件卻同時也是具高度缺水風險的國家，水資源約有70 %用於農業用途，其中以灌溉事業為主，臺灣早期灌溉輸配水設施不發達的年代均以降雨來補充作物生長所需；因此雨量如何有效利用及有效雨量的估算成為灌溉規劃及灌溉管理之重要因素。由灌溉的定義：在於補充雨量雨量之不足，即說明雨量用於灌溉之重要性。隨著輸水系統的發達，導致國內水利會現行灌溉水量擬定時多不考量有效雨量，天然降雨已成為次要灌溉用水來源或已不列為灌溉水源。惟受氣候變遷強化的效應下，河川逕流及降雨現象均受改變，水庫配水已面臨調配困難現象，河川亦有乾旱及災害發生的潛勢。臺灣過去的水文規劃及水資源計畫均未將氣候變遷因素納入考量，以河川水源為主要用水來源的農業灌溉事業亦面臨高頻率的缺水現象。有效雨量的評估及高度應用於灌溉亦為解決河川缺水風險下的方法之一。 臺灣農田水利會灌區對於有效雨量之估算及應用，目前以水庫取水灌溉之灌區為主，如石門、桃園及嘉南農田水利會等，皆於年度灌溉計畫擬定時將有效雨量納入估算計畫灌溉用水量並預先扣除，其他非水庫取水灌區則較不重視天然雨量之利用。本文以河川取水型卻潛在缺水的臺中水利會灌區為研究區域，採用灌區內12處雨量站於1961-2010年期間，共計50年的日降雨記錄進行灌區內有效雨量受氣候變遷衝擊下之時空變化分析。 水田有效雨量計算模式的建立以估算稻作及旱作的有效雨量為主，其中有效雨量門檻值的制定為有效雨量計算合理與否之關鍵，本文以田間管理為基礎的水門操作法進行稻作有效雨量的估算，並以2種門檻值進行有效雨量計算之基礎，分別為(1) 灌溉計畫編列之灌溉水深值、(2) 理論田間需水量；其中門檻值(2)可作為驗證使用；旱作有效雨量則以作物蒸發散量為門檻值。作物蒸發散量乃引用臺灣氣象局12處氣象站資料，以Penman Monteith法計算並應用一般克利金法(OK)進行空間內插推估。 有效雨量的時空分佈的趨勢分析，首先針對灌區內12處雨量站資料，本文應用最小平方法、移動平均法、指數平滑法、無母數統計方法(Mann-Kendall)進行各項雨量參數之長期趨勢分析及檢定，連續性的日降雨資料如有缺值時以反距離權重法(IDW)進行空間插補處理，並以交叉驗證法作為IDW參數優化的方法；趨勢分析結果顯示各種降雨參數於山區、平地及沿海灌區之趨勢存在變異但均不顯著。其次選定5項水文因子及8項氣象因子應用動態因子分析法(DFA)探討有效雨量各種潛在影響因子對有效雨量之貢獻度；結果顯示潛在的水文影響因子分別為年累積雨量、最大日雨量、降雨日數、日雨量第3四分位數等4項及平均溫度、風速等2項氣象因子。 有效雨量的空間分佈以克利金法為理論基礎，經分析研究區內12處有效雨量資料，採用一般克利金法(OK)進行空間分佈推估；同時應用指標克利金法(IK)推估空間發生特定有效雨量值之機率分布值，繪製有效雨量受氣候變遷衝擊下不同時期之發生空間及機率分佈圖，其有效雨量時空分佈結果將可作為農業水資源調配及規劃之決策參酌。 氣候變遷對於有效雨量的衝擊影響評估，本文採用GFCM20、HADCM3、INCM3、MPEH5、MRCGCM等5種基期資料與研究區域之雨量及溫度均具高相關性(相關係數>0.8)的GCM模式，進行A2、A1B、B1等3種情境下之短期(2010-2039)、中期(2040-2059)及長期(2060-2089)的衝擊模擬；結果顯示於A2、A1B、B1等3種情境下，有效雨量均呈現豐水期增加而枯水期減少的共同趨勢，趨勢由高海拔山區往低海拔的沿海灌區更為顯著。因應氣候變遷對水田有效雨量的衝擊影響，本文提出3項調適措施進行提高有效雨量利用之策略，分別為(1) 調整耕種時期、(2) 調整田埂高度、(3) 作物轉作等；經重新計算有效雨量後，以曝露度及缺水指標之乘積進行脆弱度分析，以作為有效雨量應用於農業灌溉用水因應氣候變遷影向下之調整策略擬定；結果顯示以調整田埂高度至300 mm的措施為最佳、水稻轉作為和作物栽培次之、調整耕種時期的策略效果則不顯著。

Taiwan is a country of abundant annual rainfall but is also a country under high risk of water deficiency. Up to 70 percentage of water resource are used for agricultural purposes, and the main component is irrigation water. Rainfall was seen as the main source towards irrigation for crops growth in the past, hence the effective rainfall (ER) became a critical factor in crop irrigation due to the underdevelopment of canal systems during the earlier periods in Taiwan. The usage and determination of ER is important in the process towards planning irrigation scheduling and agricultural water management. Through climate change, rainfall patterns and stream runoff have altered drastically causing increased frequencies in some phenomena such as drought and other water related disasters, as well as difficulties in water resource allocation. This has put an increased pressure on irrigation to satisfy the water resource that crop require. Hence a precise evaluation and high application of ER was considered one of the most suitable solutions toward the future. Applications of ER are usually conducted in the process of planning annual irrigation scheduling by irrigation associations where water source intake are specifically from reservoirs (e.g. Shimen, Taoyuan, and Chianan irrigation associations). ER has become an increasingly important factor to other irrigation associations where the irrigation water is drawn from rivers, even if it is not commonly applied by them. The irrigation region managed by the Taichung Irrigation Association (TIA) was selected as the research objective in this study. TIA mainly takes water from rivers and potentially suffers by water shortage. Rainfall data from 12 TIA rainfall stations and 600 sets of meteorological data from the Central Weather Bureau (CWB) during 1961 to 2010 were used to analyze ER variations under impact of climate change. The ER estimation model was eatablished to estimate ER in paddy fields and in upland fields, where the threshold value of ER was the crucial factor for a reasonable ER estimation. Two types of threshold values were conducted: (1) ponding depth from irrigation scheduling and (2) theoretical water requirement. The latter was used for verification purposes. The amount of evapotranspiration based on meteorological data collected from 12 selected meteorological stations of the CWB is adopted as the threshold value in upland fields and it is calculated and interpolated by using the Penman Monteith and the ordinary kriging methods. Moving average (MA), the least square method (LS), exponsential smoothing (ES) and nonparametric statistical methods such as Mann-Kendall (MK) were adopted for analyzing the long-term temporal trend of rainfall parameters at an individual rainfall station. Inverse distance weight (IDW) was incorporated to overcome the issue of missing daily rainfall data in this study whereas cross validation was used to ensure the optimal parameters of IDW. Trend analysis results show that rainfall parameters in mountant areas, plain areas and seashore areas are different but differences between them are not significant. The dynamic factor analysis (DFA) is used to clarify important influence factors on ER among 5 hydrological factors and 8 meteorological factors. The results indicat that annual rainfall (R), maximum daily rainfall (MDR), rainy days (RD), and the 3rd quartile daily rainfall (RQ3) are recognized as important hydrological factors and mean air temperature (AT) and wind speed at 2 m height (WS) are recognized as important meteorological factors. In this study, the ordinary kriging (OK) method is used to estimate spatial rainfall distributions and the indicator kriging (IK) method is used to stimulate probability distributions of a specific ER. They could be used to effectively predict spatial and temporal ER distributions under impact of climate change. Thus, these results are seen as advantageous towards agricultural water resources planning and management. To conduct the impact assessment of ER under the influence of climate change, ten types of GCMs such as Mk3.0, ECHAM5-OM, CM2.0, CM2.1, CM3.0, CM4, MIROC3.2 medres, CGCM2.3.2, CCSM3, HadCM3 are selected. The simulation are adopted in three scenarios of A2, A1B, B1 for the short-term (2010-2039), mid-term (2040-2059), and long-term predictions (2060-2089) using TaiWAP model. The results show that the ER increases in flood seasons but decreases in dry seasons. In addition, variations in mountain areas are greater than those in seashore areas. To reduce impact of climate change, the following three adaptation strategies are proposed: (1) adjustment of cultivation period, (2) adjustment of ridge height, (3) conversion from paddy field to upland field. The exposure and the water shortage index are adopted to evaluate performances of vulnerability for these three strategies. Adjusting ridges to 300 mm height is the optimal strategy and is recommended. The second priority is the conversion from paddy field to upland field. Adjustment of cultivation period is not recommended due to the worse performance of vulnerability.