Quantifying the ratio of transpiration (T) to evapotranspiration (ET), T/ET, is crucial for understanding and predicting the water cycle and energy balance between the land and atmosphere. Here, we used three well-validated ET partitioning models to estimate T/ET at global fluxnet sites. The models are the Shuttleworth-Wallace-Hu (SWH) model, the Priestley-Taylor Jet Propulsion Lab (PT-JPL) model, and the underlying water use efficiency (uWUE) method. SWH illustrated the most reliable estimate in both the magnitude and the across-site variability of T/ET among three models. Mean annual T/ET derived from SWH, PT-JPL, and uWUE were 0.61±0.14, 0.52±0.12 and 0.59±0.07, respectively. Leaf area index (LAI) was the key driver of spatial variations in T/ET across sites, as well as seasonal variations in T/ET in ecosystems in most climatic zones except for the tropical and arid regions. However, there were discrepancies in factors controlling inter-annual variations in T/ET among the models. SWH and PT-JPL showed that the inter-annual variation in T/ET was more related to LAI than climatic factors but an opposite result was found by uWUE due to the simple structure and less forcing data of uWUE. The findings of our research highlight the importance of capturing the controls of LAI on ET partitioning for predicting future water cycle with land models. We appeal for direct measurements of ET components at the flux tower sites for validating the models and reduce model uncertainties.

量化蒸腾 ( T ) 与蒸散 (ET)的比率T /ET 对于理解和预测陆地和大气之间的水循环和能量平衡至关重要。在这里，我们使用三个经过充分验证的 ET 分区模型来估计全球通量网站点的T /ET。这些模型是 Shuttleworth-Wallace-Hu (SWH) 模型、Priestley-Taylor 喷气推进实验室 (PT-JPL) 模型和基础用水效率 (uWUE) 方法。SWH 说明了三个模型中T /ET的幅度和跨站点变异性的最可靠估计。平均年T源自 SWH、PT-JPL 和 uWUE 的 /ET 分别为 0.61±0.14、0.52±0.12 和 0.59±0.07。叶面积指数 (LAI) 是不同地点T /ET空间变化的关键驱动因素，也是除热带和干旱地区以外的大多数气候区生态系统中T /ET季节性变化的关键驱动因素。然而，模型之间控制T / ET年际变化的因素存在差异。SWH 和 PT-JPL 表明T的年际变化/ET 与 LAI 的相关性高于气候因素，但由于 uWUE 的结构简单且强迫数据较少，uWUE 发现了相反的结果。我们的研究结果强调了捕捉 LAI 对 ET 分区的控制对于使用土地模型预测未来水循环的重要性。我们呼吁在通量塔站点直接测量 ET 组件，以验证模型并减少模型的不确定性。