Abstract:

Analyzing the spatiotemporal variations of potential evaporation and runoff is crucial for accurately understanding the actual evaporation and water balance changes in climate-sensitive areas. However, due to the inter-annual and seasonal variations of the global climate, predicting and evaluating potential evaporation and runoff at the watershed scale are challenging. In this study, we selected the source area of the Yellow River, a climate-sensitive region, as the research target. We utilized the Comprehensive Differential Sample Method (CDSST) to classify the dry and wet states of the watershed into wet years, dry years, and mixed years. Then, we constructed monthly-scale abcd hydrological models for each state. We also investigated how the uncertainty in four potential evaporation algorithms (Haregreaves, Makkink, Penman-Monteith, and Jensen-Haise) affects the prediction uncertainty of potential evaporation and runoff in the watershed. Subsequently, by employing hydrological indicators such as the unevenness coefficient, concentration degree, and relative variation range, we revealed the influence of these four different potential evaporation algorithms on watershed water resource prediction. Results show that, compared with the period before 1990, the number of years when the watershed was in a dry state increased (from 4 years to 10 years). The corresponding proportions of wet years, dry years, and mixed years were 25.86%, 24.14%, and 50%, respectively. Meanwhile, the uncertainty of potential evaporation algorithms alters the spatiotemporal distribution characteristics of watershed evaporation and runoff. These findings are essential for scientifically grasping the dynamic changes of watershed hydrological processes under changing environments, as well as for regional water resource management and ecological restoration.

Key words: potential evaporation, runoff uncertainty, water resources, source region of the Yellow River