为更准确揭示和认识变化环境下密云水库流域水文过程的非平稳特征，选取该流域1960—2019年降水和径流资料，采用多种时间序列分析方法综合诊断其趋势、突变点和周期等非平稳特征并进行成因分析。结果显示：该流域降水变化主要表现出较明显的随机特性，径流相比降水的下降趋势更显著，且在1979年发生了向下跳跃的强变异；相比气候变异影响，人类活动是密云水库流域径流减少的主导因素；1980—1998年流域水土保持措施与水利工程兴建对径流变化的贡献率为?111.40%，抵消了气候变异11.4%的增水效应；1999—2019年，塘坝建造和用地类型转变对径流变化的贡献率为?66.60%，同时叠加气候变异?33.40%的减水效应，导致近20年来该流域径流呈现显著减少的态势。研究结果可为密云水库安全运行、洪水调度以及流域水资源管理决策提供科学支撑。

Under the influences of climate change and human activities, the hydrological process formation conditions in many basins and regions have changed significantly, and observed hydrological time series show growingly non-stationary characteristics. The uneven spatial and temporal distribution of water resources has been aggravated, resulting in new challenges for regional water resources management. It has been a hot topic to detect and attribute the changes in precipitation-runoff processes in changing environments. As the largest surface drinking water source in Beijing and the largest reservoir in north China, the Miyun Reservoir is responsible for water supply including urban, industries, and agriculture in the region. Precipitation-runoff processes upstream of the Miyun Reservoir also showed obvious non-stationary characteristics and extreme flooding events occurred in 2021 and 2023.For accurately identifying the non-stationary characteristics of hydrological processes in the Miyun Reservoir basin, and further clarifying its physical causes by considering the influences of both climate change and human activities, the observed annual data of precipitation and runoff from 1960 to 2019 were used. Multiple methods for time series analysis, including the Mann-Kendall test, hydrological alteration diagnosis system, and discrete wavelet spectrum for period identification were applied to detect the non-stationary characteristics (including trend, breakpoint, and periodicities) of precipitation and runoff. Based on the detection results, the Budyko framework-based water balance method was used to attribute the changes in runoff.Precipitation displayed random changes across the whole time, with no noteworthy trend, breakpoint, or period observed. However, runoff's non-stationary characteristics differed from those of precipitation. The decreasing trend of runoff was more significant compared with precipitation, and a significant breakpoint in 1979 was identified in the annual runoff time series. Based on that, from 1960 to 1979 was chosen as the baseline period, and the changes in runoff from 1980 to 1998 and from 1999 to 2019 were attributed to using the Budyko framework-based water balance method. The results indicated that human activities dominated the decrease in runoff in the Miyun Reservoir basin, compared to the influences of climate change. From 1980 to 1998, the soil and water conservation measures and constructions of hydraulic projects contributed ?111.40% of runoff compared to the baseline period, offsetting the runoff increase of 11.4% caused by climate change. From 1999 to 2019 change caused by dam construction and land use change was ?66.60% of the total runoff change, and climate change further aggravated ?33.40% of the change In recent years, the Miyun Reservoir basin has been in rainy periods, with higher precipitation compared to an average level. Coupled with the impacts of human activities on the runoff, the hydrological process and runoff present more complex variability in the Miyun Reservoir basin. Detection and attribution of the non-stationary characteristics of precipitation-runoff processes could be a useful basis for guiding the safe operation and water management of the Miyun Reservoir.