Abstract Estimating continental water availability is of great importance for water resources management, in terms of maintaining ecosystem integrity and sustaining society development. To more accurately quantify the predictability of water availability, on the basis of univariate probabilistic Budyko framework, a bivariate probabilistic Budyko approach was developed using copula-based joint distribution model for considering the dependence between parameter ω of Wang-Tang’s equation and the Normalized Difference Vegetation Index (NDVI), and was applied globally. The results indicate the predictive performance in global water availability is conditional on the climatic condition. In comparison with simple univariate distribution, the bivariate one produces the lower interquartile range under the same global dataset, especially in the regions with higher NDVI values, highlighting the importance of developing the joint distribution by taking into account the dependence structure of parameter ω and NDVI, which can provide more accurate probabilistic evaluation of water availability.

中文翻译：

---

水资源可用性可预测性的全球评估：双变量概率 Budyko 分析

摘要 就维持生态系统完整性和维持社会发展而言，估算大陆可用水量对于水资源管理非常重要。为了更准确地量化可用水量的可预测性，在单变量概率 Budyko 框架的基础上，使用基于 copula 的联合分布模型开发了一种双变量概率 Budyko 方法，以考虑 Wang-Tang 方程的参数 ω 与归一化差分植被之间的依赖性指数（NDVI），并在全球范围内应用。结果表明，全球可用水量的预测性能取决于气候条件。与简单的单变量分布相比，双变量分布在同一全局数据集下产生较低的四分位距，