植被與氣候驅動因子對尼羅河流域水儲量動態之時空影響：2003–2022年

Abstract

水儲量變化是評估水文變遷的重要指標，因為其反映降雨、蒸發散、土壤水分、地下水、地表水、植被活動及人類用水等多種過程的綜合影響。本研究探討2003年至2022年間，植被與氣候驅動因子對尼羅河流域水儲量動態之影響。研究分析三項水儲量目標變數，包括GRACE/GRACE-FO陸地水儲量異常（TWSA）、GLDAS土壤水分異常（SMA）及GLDAS地下水儲量異常（GWSA）。本研究選取四項解釋變數，分別為降水量、降水吉尼指數（GI）、校正後正向蒸發需求（PETd）及MODIS常態化差異植生指數（NDVI）。 所有資料均處理至共同的每月0.25°尼羅河流域網格，並轉換為去季節化異常值。分析以網格尺度進行，並進一步彙整至九個HydroBASINS Level-04子流域。首先，本研究採用0至3個月最佳滯後相關分析，以考量水文系統可能存在的延遲反應。接著，使用逐步多元線性迴歸辨識具統計意義的解釋變數，並利用標準化係數判讀水儲量與驅動因子之間的關係方向。最後，透過優勢分析量化各解釋變數的相對重要性，並採用Theil–Sen斜率與Kendall顯著性檢定進行水儲量趨勢熱點分析。 研究結果顯示，尼羅河流域水儲量動態具有明顯的空間異質性，且不同水儲量組成之間呈現不同反應。GRACE TWSA在流域多數區域呈現整體增加趨勢，而GLDAS土壤水分與地下水儲量異常則呈現較為區域化且混合的變化型態。降水量與土壤水分及地下水相關儲量多呈正相關，顯示其作為主要水分輸入因子的作用。PETd與水儲量之間則呈現一致的負向關係，尤其在SMA與GWSA中更為明顯，表示大氣蒸發需求對水儲量消耗具有重要影響。NDVI是解釋GRACE TWSA變化最主要的因子，顯示總水儲量與植被綠度之間存在緊密耦合關係。相較之下，SMA與GWSA受到降水量、PETd與NDVI較為平衡的共同控制。GI則較多扮演調節降雨有效性的角色，而非主要主導因子。 整體而言，本研究指出尼羅河流域水儲量變化無法僅由降水量解釋，而是受到降雨供給、降雨分布、蒸發需求、植被動態及區域水文條件交互作用的共同影響。

Water storage variability is an important indicator of hydrological change because it reflects the combined effects of rainfall, evapotranspiration, soil moisture, groundwater, surface water, vegetation activity, and human water use. This study examines how vegetation and climate drivers influence water storage dynamics across the Nile Basin from 2003 to 2022. The analysis focuses on three storage targets: GRACE/GRACE-FO terrestrial water storage anomaly (TWSA), GLDAS soil moisture anomaly (SMA), and GLDAS groundwater storage anomaly (GWSA). Four explanatory drivers were considered: precipitation, precipitation Gini Index (GI), corrected positive evaporative demand (PETd), and MODIS NDVI. All datasets were processed to a common monthly 0.25° Nile Basin grid and converted into deseasonalized anomalies. The analysis was conducted at the grid-cell scale and summarized across nine HydroBASINS Level-04 subbasins. A 0–3 month best-lag correlation analysis was used to account for delayed hydrological response. Stepwise multiple linear regression was then applied to identify statistically selected predictors. Standardized coefficients were used to interpret the direction of storage–driver relationships, dominance analysis was used to quantify relative predictor importance, and Theil–Sen/Kendall hotspot analysis was used to detect significant storage trends. The results show that Nile Basin storage dynamics are spatially heterogeneous and differ among storage components. GRACE TWSA shows broad increasing tendencies across much of the basin, while GLDAS soil moisture and groundwater storage anomalies show more regionally mixed patterns. Precipitation is strongly and positively associated with soil moisture and groundwater-related storage, confirming its role as a primary water-input driver. PETd shows a consistent negative relationship with storage, especially for SMA and GWSA, indicating the importance of atmospheric drying pressure. NDVI is the strongest explanatory factor for GRACE TWSA, suggesting close coupling between total storage and vegetation greenness. In contrast, SMA and GWSA are controlled by a more balanced combination of precipitation, PETd, and NDVI. GI contributes more as a modifier of rainfall effectiveness than as a dominant driver. Overall, this study shows that Nile Basin water storage cannot be explained by rainfall amount alone. Storage variability reflects the interaction of rainfall supply, rainfall distribution, evaporative demand, vegetation dynamics, and regional hydrological conditions.