中国科学院水利部水土保持研究所机构知识库

近半个世纪以来，在气候变化和人类活动的共同作用下，黄河径流呈现出下降趋势。研究河流径流变化及其归因分析对流域治理以及水土保持工作意义重大。为此，本研究采用Mann-Kandall趋势检验、Pettitt突变检验法、双累积曲线等方法研究了黄河河龙区间的三个典型流域（皇甫川流域、无定河流域和延河流域）的径流变化特征，并基于Budyko假设定量评估了气候和下垫面变化对径流量变化的影响程度。取得了如下主要结论：

（1）皇甫川、无定河和延河流域的降水量和潜在蒸发量年际变化趋势均不显著，而径流量均呈显著的下降趋势且具有明显的阶段性和突变性特征。3个流域降水量和潜在蒸发量的MK趋势检验统计量均未达到显著性水平，而径流量均达显著的下降水平。皇甫川径流的突变点为1996年；无定河径流存在1971年和1996年2个突变点；延河径流的突变点为1996年。

（2）流域的径流深对下垫面参数的敏感性大于气候因子，且互补法计算的径流深精度最高。三个流域径流深R对ω、P和ET₀的敏感性系数均值分别为-3.52，2.61和-1.62。对3个流域径流变化进行归因分析表明，不同流域变化期下垫面变化均为径流变化的主导因素；皇甫川流域下垫面变化对径流减小的影响为88%，无定河流域1997-2016年间下垫面变化对径流的贡献为61%，延河流域下垫面变化对径流的影响为37%；采用“两段平均”法计算全微分法中的偏导，得到3个流域的径流变化与实际观测值之间的相对误差分别为13.52%、1.74%和0.44%；对比分析四种归因方法：全微分法、互补法、分解法以及外推法的特点可知互补法是计算归因分析最佳方法。

（3）下垫面参数对气候季节性指数敏感性小于对植被覆盖度的季节性指数。3个流域气候季节性指数无显著的年际变化趋势，而植被覆盖度呈显著的上升趋势；各流域年干旱指数与蒸散率的所有点都集中在Budyko曲线周围，参数ω在皇甫川、无定河以及延河流域内对气候性季节指数S的敏感性系数分别为-0.04、-0.05、-0.03，而对植被覆盖度的敏感性系数分别达到0.24、0.29、0.31。表明，在影响无定河流域和皇甫川流域径流变化的因素中，人类活动、土壤状况以及地形坡度等因素的贡献量较大。除气候、下垫面变化以外，流域蓄水量变化也对径流变化有一定的贡献。近年来退耕还林以及水土保持措施对径流的减小起到了很大的作用。

For nearly half a century, dramatic changes in climatic factors and underlying conditions have had a major impact on the catchments' hydrological cycle. With the improvement of the ecological system in the Loess Plateau, the annual runoff and sediment volume of the Yellow River has decreased sharply. Studying river runoff changes and its attribution analysis is of great significance for watershed management and soil and water conservation. To this end, this study used Mann-Kandall trend test, Pettitt mutation test, double cumulative curve and other methods to study the runoff variation of three typical basins in the Yellow River Helong section (Huangfuchuan Basin, Wuding River Basin and Yanhe River Basin). Based on the Budyko hypothesis, quantitatively assess the extent to which climate and underlying surface changes affect runoff changes. The main conclusions are as follows:

1. The interannual variation trend of precipitation and potential evaporation in basins was not significant, while the runoff showed a significant downward trend with obvious stage and abrupt characteristics. The MK trend test statistic of precipitation and potential evaporation in the three basins did not reach a significant level, while the runoff reached a significant decline. we can see that there is a catastrophe point in the Huangfuchuan Basin, which is 1996; there are two mutation points in the Wuding River Basin, in 1971. And in 1996, the Yanhe River Basin had a mutation point for 1996.
2. The runoff depth of basins is more sensitive to the underlying surface parameters than the climatic factors, and the complement method calculates the runoff depth with the highest accuracy. The mean values of the sensitivity of the three runoff depths R to ω, P and ET₀ were -3.52, 2.61 and -1.62, respectively. Attribution analysis of runoff variation in three basins by means of total differential method and complementary method showed that underlying surface changes in different basins were the dominant factors of runoff variation; the influence of underlying surface changes on runoff reduction in Huangfuchuan Basin was 88%, the contribution of the underlying surface change to runoff during the period from 1997 to 2016 in the Wuding River Basin was 61%, and the impact of the underlying surface change on the runoff in the Yanhe River Basin was 37%; the “two-stage average” method was used to calculate the total differential method. The relative error between the runoff variation and the actual observation values of the three basins is 13.52%, 1.74% and 0.44%, respectively. The characteristics of the total differential method, the complementary method, the decomposition method and the extrapolation method are complementary. The method is the best way to calculate attribution analysis.
3. The underlying surface parameters of the three watersheds are less sensitive to the seasonal climate index than that of vegetation coverage. The seasonal climate index of the three basins show no significant interannual variability, and the vegetation coverage showed a significant upward trend. All the annual dryness index and evapotranspiration rate of each watershed were concentrated around the Budyko curve, and the parameter ω was in . The sensitivity coefficients of the parameter ω to the climatic seasonal index S in the Huangfuchuan, Wuding river and Yanhe river basin are -0.04, -0.05, -0.03, respectively, and the sensitivity coefficients to vegetation coverage are 0.24, 0.29, and 0.31, respectively. It is indicated that among the factors affecting the runoff variation in the Wuding River Basin and the Huangfuchuan Basin, factors such as human activities, soil conditions and terrain slope have contributed a lot. In addition to changes in climate and underlying surface, changes in water storage in the basin also contribute to changes in runoff. In recent years, returning farmland to forests and soil and water conservation measures have played a significant role in reducing runoff.