Abstract Freshwater scarcity is a major global concern that is being exacerbated by the increasing population and changing climate. While close estimates of water cycle components and water budget analysis can help with planning for sustainable water resources management, limited hydroclimatic data have made such analysis challenging. The Upper Blue Nile basin in Ethiopia is such a hydroclimatic data–scarce region, with relatively arid and hot weather. Multiple studies have focused on water budget analysis in the UBN basin, but water cycle analyses are missing at high spatiotemporal resolution yield from a fully distributed hydrological model that closes water and energy budgets. In this paper, we describe the simulation of evapotranspiration (ET) and discharge at fine spatiotemporal resolution (500 m and 3 hourly) from 1979 to 2014, using the Coupled Routing and Excess STorage Soil-Vegetation-Atmosphere-Snow (CREST- SVAS) distributed hydrological model, which physically maintains water and energy balance. We also present a comparison of ET and discharge as derived from CREST with those simulated by other existing global models. Our results show that, where the global models perform well for specific hydrological components, ET, and discharge, CREST can simulate ET at 500 m resolution with 0.93 correlation coefficient, 0.81 Nash-Sutcliffe efficiency (NSE), and –4.75% bias (for daily scale) and discharge at different basin scales with NSE values ranging from 0.46 to 0.75. We performed cascade calibration for discharge and tested the optimized parameters on noncalibrated sites. The good model performance at the noncalibrated sites (NSE > 0.6; bias within 15%) indicated the calibrated parameters can be used for ungauged locations. Finally, the terrestrial water storage change (TWSC) from CREST was also consistent with reference data and the Gravity Recovery and Climate Experiment (GRACE). The hyper-resolution accurate hydrological components from CREST will enable effective prediction of water budget, helpful for initiating local and large-scale sustainable development plans that address social and policy requirements, as well as extreme climatic conditions.

中文翻译：

---

上青尼罗河流域超分辨率模型衍生的水循环分量的评估

摘要 淡水稀缺是一个主要的全球问题，随着人口的增加和气候的变化而加剧。虽然对水循环组成部分的密切估计和水预算分析有助于规划可持续水资源管理，但有限的水文气候数据使此类分析具有挑战性。埃塞俄比亚的上青尼罗河流域是这样一个水文气候数据稀缺的地区，天气相对干旱和炎热。多项研究都集中在 UBN 流域的水收支分析上，但在关闭水和能源收支的完全分布式水文模型的高时空分辨率下，水循环分析缺失。在本文中，我们描述了 1979 年至 2014 年间以精细时空分辨率（500 m 和 3 小时）的蒸散 (ET) 和排放模拟，使用耦合路由和过量存储土壤-植被-大气-雪 (CREST-SVAS) 分布式水文模型，该模型在物理上保持水和能量平衡。我们还比较了来自 CREST 的 ET 和排放量与其他现有全球模型模拟的排放量。我们的结果表明，在全球模型对特定水文成分、ET 和流量表现良好的情况下，CREST 可以在 500 m 分辨率下模拟 ET，相关系数为 0.93，纳什-萨特克利夫效率 (NSE) 为 0.81，偏差为 –4.75%（对于日尺度）和不同流域尺度的流量，NSE 值范围为 0.46 至 0.75。我们对放电进行了级联校准，并在未校准的站点上测试了优化的参数。非校准站点的良好模型性能（NSE > 0.6；偏差在 15% 以内）表明校准参数可用于未测量的位置。最后，来自 CREST 的陆地储水量变化 (TWSC) 也与参考数据和重力恢复和气候实验 (GRACE) 一致。CREST 的超分辨率准确水文组件将能够有效预测水资源预算，有助于启动满足社会和政策要求以及极端气候条件的地方和大规模可持续发展计划。