| 摘要: |
| 在气候变化与冻土退化背景下，黄河源区径流组成发生变化，基流识别结果对水资源调控具有重要影响。为揭示黄河源区不同基流分割方法的适用性及季节特征，基于唐乃亥水文站1960–2019年逐日径流数据，采用数字滤波、HYSEP等8种方法进行基流分割。从基流指数统计特征、过程线平滑性和洪峰响应等方面比较其适用性，并分析其年际变化趋势与季节性变化。结果表明:(1)不同方法估算结果具有明显的方法依赖性，Lyne-Hollick和HYSEP类方法估算结果整体偏高，所得多年平均BFI为0.77～0.84，Chapman、Chapman-Maxwell及Eckhardt滤波法BFI为0.48～0.51;(2)Chapman、Chapman-Maxwell和Eckhardt滤波法年际波动较小，基流过程线较平滑，且对洪水过程响应适中，表现出相对较好的适用性;(3)HYSEP方法在汛期对洪水过程响应偏强，可能将部分尚未消退的快速径流或壤中流计入基流，导致基流贡献估算偏高;(4)各方法虽在BFI值上存在差异，均表现出一致的冬高夏低季节性规律，但BFI与估算基流量在季节尺度上并不同步，表现为冬季高占比、低水量和夏季低占比、高水量。基于方法在不同水文阶段的差异，提出分季节基流分割方法优选猜想，但效果仍需结合实测数据进一步验证。研究结果可为黄河源区及类似流域基流分割方法选择和过程分析提供参考。 |
| 关键词: 基流 基流分割 黄河源区 数字滤波法 季节分异 基流指数 |
| DOI： |
| 分类号:P333 ;TV11 |
| 基金项目:国家重点研发计划项目（2023YFC3206305）；水利部重大科技项目(SKS-2025094)；中央级公益性科研院所基本科研业务费项目(Yj526019、Yj526020) |
|
| Adaptability evaluation of baseflow separation methods and seasonal differentiation characteristics in the source region of the Yellow River |
|
LI Yuhang,ZHANG Xianzhi,WU Nan,CHEN Liangang,GUO Xiaojing
|
|
1.State Key Laboratory of Hydraulic Disaster Prevention,Nanjing Hydraulic Research Institute;2.Bureau of Hydrology,Yellow River Conservancy Commission
|
| Abstract: |
| Under the combined effects of climate change and permafrost degradation, hydrological processes and runoff composition in the source region of the Yellow River have changed substantially. Variations in precipitation, temperature, snowmelt timing, frozen-ground conditions, and subsurface water storage can alter the generation and seasonal distribution of runoff components. Reliable identification of baseflow is therefore important for understanding runoff-generation mechanisms and supporting water resources management. However, baseflow is difficult to measure directly at the basin scale and is generally estimated using hydrograph separation methods. Because different methods rely on distinct assumptions, algorithms, and parameter settings, their estimates may vary considerably, especially in alpine cold-region basins with snowmelt, seasonally frozen soil, permafrost degradation, and marked seasonal runoff variability. To evaluate the applicability of different baseflow separation methods and clarify the interannual and seasonal characteristics of baseflow in the Yellow River source region, eight methods, including digital filtering and HYSEP approaches, were applied to daily discharge records from the Tangnaihai Hydrological Station for 1960–2019. Their performance was compared in terms of BFI statistics, interannual variability, hydrograph smoothness, and responses to flood peaks. Seasonal variations in BFI and estimated baseflow volume were also analyzed. The results showed that: (1) Baseflow estimates exhibited pronounced method dependence. The Lyne–Hollick and HYSEP methods generally produced high estimates, with long-term mean BFI values of 0.77–0.84, whereas the Chapman, Chapman–Maxwell, and Eckhardt filters yielded lower values of 0.48–0.51. These differences indicate that method selection strongly affects the estimated contribution of baseflow to total runoff and may influence interpretations of runoff composition and hydrological processes. Therefore, BFI values obtained using different methods should be compared cautiously, considering their mathematical structures and hydrological assumptions. (2) The Chapman, Chapman–Maxwell, and Eckhardt filters showed small interannual fluctuations and generated smoother baseflow hydrographs. Their responses to flood events were generally moderate, avoiding both excessive suppression of hydrological variability and unrealistic tracking of short-term discharge peaks. These characteristics indicate comparatively better applicability in the Tangnaihai catchment. Nevertheless, hydrograph smoothness alone cannot be regarded as evidence of accuracy because it is partly controlled by each method’s filtering characteristics. Method suitability should therefore be assessed by jointly considering temporal stability, hydrograph morphology, and flood-response behavior. (3) The HYSEP methods exhibited strong responses to flood processes, particularly during the flood season. In some events, separated baseflow remained high during or immediately after the rising and recession stages of the flood hydrograph. This behavior suggests that part of the quickflow that had not fully receded, or part of the relatively slow interflow generated by rainfall and meltwater, may have been classified as baseflow. Such classification can overestimate the baseflow contribution and may partly explain the high long-term mean BFI values obtained using HYSEP methods. The Lyne–Hollick filter also generated high estimates, although its temporal response differed from HYSEP because of differences in algorithmic structure. (4) Although absolute BFI values differed markedly among the eight methods, all methods showed a consistent seasonal pattern of higher BFI in winter and lower BFI in summer. During winter, total discharge was relatively low and streamflow was mainly sustained by slowly varying subsurface contributions, resulting in a high baseflow proportion. During summer, rainfall, snowmelt, and glacier melt increased total runoff, while rapid runoff components became more prominent, resulting in a lower BFI. However, seasonal variation in BFI was not synchronized with that in estimated baseflow volume. Winter was characterized by a high baseflow proportion but a low absolute baseflow volume, whereas summer was characterized by a lower baseflow proportion but a substantially higher baseflow volume. This finding demonstrates that BFI represents the relative contribution of baseflow rather than its absolute magnitude. Seasonal interpretation should therefore consider both BFI and baseflow volume. Based on the contrasting performance of the methods across hydrological periods, a working hypothesis of season-specific selection or combination of baseflow separation methods is proposed. Methods could be selected according to their performance during frozen periods, snowmelt-transition periods, flood seasons, and post-flood recession periods. However, this hypothesis remains exploratory and requires further validation using independent observations, such as hydrochemical tracers, stable isotopes, groundwater-level records, or hydrological simulations. These findings provide a methodological reference for selecting baseflow separation methods and interpreting baseflow processes in the Yellow River source region and other cold-region basins with similar hydroclimatic and cryospheric conditions. |
| Key words: Baseflow Baseflow separation Source region of the Yellow River Digital filtering method Seasonal variation Baseflow index |
