Attribution analysis of runoff and sediment changes in the Yellow River Basin based on the Budyko framework

As a base for food and energy production, the Yellow River Basin (YRB) produces about 30% of the country's food and 70% of the country's coal with 2.5% of the country's water resources (Wang & Xu 2022). In recent decades, about 31% of the world's 145 major rivers have had statistically significant upward (9%) or downward (22%) trends in annual runoff as a result of climate change and increased human activity such as grazing, building terraces, and dams, converting farmland to forest and so on (Walling & Fang 2003).

Runoff from many of China's major rivers is on a decreasing trend, especially in the YRB (Wang et al. 2010; Feng et al. 2016a); at the same time, sediment in the YRB has experienced a significant decline. Runoff and sediment, as integral parts of the river system, are susceptible to anthropogenic and climatic changes. It has been an interesting topic to delineate human and climatic influences on runoff and sediment change, which is important for regional water resources planning and management (Fu et al. 2007).

A series of methods are currently being practiced to quantitatively distinguish the influences of climatic and anthropogenic disturbances on water and sediment variability, broadly categorized into empirical statistics and hydrological models (Chang et al. 2015; Yang et al. 2015). Different from the detailed modeling process, some hydrologists try to describe the hydrological process as a whole, and the Budyko hypothesis is a typical example. The Budyko equation involving surface parameters has good explanatory performance and is widely used in attribution analysis studies of runoff changes (Ji et al. 2022). Integration of fractal theory and the Budyko framework provides a new path for contribution calculation of sediment changes (Zhang et al. 2017, 2019).

The Yellow River is the ‘mother river’ of China, and the realization of the continued utilization of water resources in the YRB is of strategic significance to the ecological environment and social-economic development (Zhang et al. 2018). In past decades, the YRB has been subjected to intense human activities, a series of large reservoirs have been put into operation to meet the increased demand for water. In order to alleviate the desertification of the land in the YRB and to curb soil erosion, the government has taken a basket of ecological engineerings, such as Three-North Shelterbelt Programme (Zhang et al. 2021), Grain for Green Program (Cao et al. 2009), which have effectively increased the vegetation coverage in the YRB, and improved the ecological environment in the area (Feng et al. 2016b; Li et al. 2020). However, although many studies have been carried out to isolate the changes in water and sediment in the YRB due to climate change and human activities, previous studies have tended to study the catchment of individual hydrological stations on the mainstream of the Yellow River, which makes it difficult to reflect the differences between different sections of the mainstream (Yang et al. 2020; Li et al. 2021; Xie et al. 2021; Hou et al. 2022; Dai et al. 2023); besides, there are still relatively few studies on sediment attribution analysis in the YRB, so in order to further fill this gap, six hydrological stations on the Yellow River were selected to explore the contributions of climate change and human activities to runoff, then, runoff–sediment relationships, i.e., relationships between the sediment transport modulus (ratio of sediment to catchment area) and runoff across six catchments, were combined with the Budyko framework to decompose the contribution to sediment change.

Daily meteorological data of 129 meteorological stations in and around the YRB were obtained from the China Meteorological Administration (http://www.cma.gov.cn), including average temperature, maximum and minimum temperature, precipitation, average wind speed, sunshine hours, and relative humidity for the period of 1967–2015. Kriging is a linear unbiased prediction method and widely practiced approach to interpolate spatial data (Kitanidis 1997; Abera et al. 2017; Wu et al. 2023), which was performed to obtain surface precipitation (P) and potential evaporation (ET₀). The records of runoff and sediment at six hydrological stations on the main stem of the Yellow River were obtained from the Yellow River Conservancy Commission (http://www.yrcc.gov.cn/), with the time range of 1967–2015; the normalized difference vegetation index (NDVI) raster images from 1981 to 2015 were obtained from the third generation Advanced Very High-Resolution Radiometer Global Inventory Modeling and Mapping Studies NDVI dataset reconstructed spatially and temporally by the Remote Sensing Team of Land Use and Global Change at the Chinese Academy of Sciences (Yang et al. 2019), with a spatial resolution of 8 km and a temporal resolution of 15 days. A maximum synthetic method was adopted to produce annual NDVI images.

The slope is used to describe the trend of the series, and the Mann–Kendall test is recommended to identify the significance of the trend (Burn & Hag Elnur 2002) and the P-values labeled in this paper are all related to the significance.

The research period was divided into two sub-periods, i.e., the period before the breakpoint listed in Table 2 was the base period (T₁) and the period after that was the change period (T₂), and the characteristic values of the catchment of each hydrological station in the YRB during the T₁ and the T₂ were listed in Table 3, and the elasticity coefficients of the runoff to P, ET₀ and n were in the ranges of 1.74–3.08, −2.08 to 0.74, and −2.70 to 1.22, respectively. The runoff of each catchment area was positively correlated with P, and negatively correlated with ET₀ and n. The runoff was most sensitive to P, followed by the n and ET₀. From upstream to downstream, the sensitivity of runoff to each element gradually increased. Compared with the base period, the sensitivity of runoff to changes in P, ET₀, and n increased in the change period in all catchments.

Table 3

Hydrometeorological parameters and runoff elasticity during T₁ and T₂

Catchment .	Period .	P (mm) .	R (mm) .	ET0 (mm) .	.	n .	.	.	.
TNH	T1	501.8	184.4	846.0	1.69	1.03	1.66	−0.66	−1.14
	T2	492.8	149.5	853.4	1.73	1.19	1.81	−0.81	−1.28
LZ	T1	470.7	153.7	871.7	1.85	1.07	1.72	−0.72	−1.27
	T2	460.1	124.5	872.3	1.90	1.22	1.86	−0.86	−1.40
TDG	T1	381.3	69.0	975.1	2.56	1.32	2.06	−1.06	−1.91
	T2	375.5	44.6	973.8	2.59	1.61	2.37	−1.37	−2.24
LM	T1	392.9	61.6	1,002.8	2.55	1.42	2.16	−1.16	−2.01
	T2	381.0	39.3	997.9	2.62	1.69	2.47	−1.47	−2.35
TG	T1	426.3	55.0	997.4	2.34	1.65	2.38	−1.38	−2.11
	T2	412.7	34.4	1,002.0	2.43	1.94	2.71	−1.71	−2.47
LJ	T1	415.7	47.6	1,010.2	2.21	1.88	2.61	−1.61	−2.26
	T2	433.1	20.7	1,002.1	2.31	2.57	3.37	−2.37	−2.98

Catchment .	Period .	P (mm) .	R (mm) .	ET0 (mm) .	.	n .	.	.	.
TNH	T1	501.8	184.4	846.0	1.69	1.03	1.66	−0.66	−1.14
	T2	492.8	149.5	853.4	1.73	1.19	1.81	−0.81	−1.28
LZ	T1	470.7	153.7	871.7	1.85	1.07	1.72	−0.72	−1.27
	T2	460.1	124.5	872.3	1.90	1.22	1.86	−0.86	−1.40
TDG	T1	381.3	69.0	975.1	2.56	1.32	2.06	−1.06	−1.91
	T2	375.5	44.6	973.8	2.59	1.61	2.37	−1.37	−2.24
LM	T1	392.9	61.6	1,002.8	2.55	1.42	2.16	−1.16	−2.01
	T2	381.0	39.3	997.9	2.62	1.69	2.47	−1.47	−2.35
TG	T1	426.3	55.0	997.4	2.34	1.65	2.38	−1.38	−2.11
	T2	412.7	34.4	1,002.0	2.43	1.94	2.71	−1.71	−2.47
LJ	T1	415.7	47.6	1,010.2	2.21	1.88	2.61	−1.61	−2.26
	T2	433.1	20.7	1,002.1	2.31	2.57	3.37	−2.37	−2.98

View Large

Table 4 listed the elasticity coefficients of sediment to each element calculated from the fitted equations.

Table 4

Sediment transport modulus and elasticity of sediment during T₁ and T₂

Catchment .	Period .	Qs (108 t/km2/year) .	.	.	.	.	.
TNH	T1	130.1	1.89	3.05	−1.16	−2.04	1.44
	T2	77.5	1.20	2.09	−0.89	−1.41	1.22
LZ	T1	265.2	0.68	0.93	−0.25	−0.51	1.12
	T2	87.5	−0.12	−0.32	0.20	0.32	0.56
TDG	T1	351.5	1.93	3.99	−2.05	−3.68	1.88
	T2	121.6	2.03	4.79	−2.77	−4.55	1.39
LM	T1	1,470.4	1.35	2.98	−1.62	−2.74	1.07
	T2	353.2	0.63	1.56	−0.94	−1.47	1.21
TG	T1	1,567.6	1.20	2.93	−1.73	−2.65	1.03
	T2	449.3	1.84	5.06	−3.22	−4.70	1.08
LJ	T1	1,237.5	1.25	3.27	−2.02	−2.82	0.69
	T2	336.7	1.32	4.45	−3.13	−3.88	0.76

Catchment .	Period .	Qs (108 t/km2/year) .	.	.	.	.	.
TNH	T1	130.1	1.89	3.05	−1.16	−2.04	1.44
	T2	77.5	1.20	2.09	−0.89	−1.41	1.22
LZ	T1	265.2	0.68	0.93	−0.25	−0.51	1.12
	T2	87.5	−0.12	−0.32	0.20	0.32	0.56
TDG	T1	351.5	1.93	3.99	−2.05	−3.68	1.88
	T2	121.6	2.03	4.79	−2.77	−4.55	1.39
LM	T1	1,470.4	1.35	2.98	−1.62	−2.74	1.07
	T2	353.2	0.63	1.56	−0.94	−1.47	1.21
TG	T1	1,567.6	1.20	2.93	−1.73	−2.65	1.03
	T2	449.3	1.84	5.06	−3.22	−4.70	1.08
LJ	T1	1,237.5	1.25	3.27	−2.02	−2.82	0.69
	T2	336.7	1.32	4.45	−3.13	−3.88	0.76

View Large

The elasticity coefficients of sediment to runoff and sediment concentration were 0.43–1.99 and 0.73–1.58, respectively, and most of the catchment areas were more sensitive to runoff change. Considering the influencing factors of runoff, the elasticity coefficients of sediment to precipitation, potential evaporation, and surface factors were in the ranges of 0.55–4.48, −2.70 to − 0.12, and −4.21 to −0.26, respectively. The magnitude of sensitivity of sediment to each factor was mainly shown as P > n > ET₀ > C. Compared with the base period, the elasticity coefficients of sediment were predominantly decreasing in the upper reaches, while the middle and lower reaches were more sensitive to changes in runoff and sediment concentration.

The fractal theory establishes the rank curve of sediment concentration and runoff through extensive records, but the process is relatively complicated and the data are not always available. In this paper, we adopted the empirical equations on an annual scale to establish the relationships between sediment transport modulus and runoff as well as sediment concentration, which can effectively decompose the elasticity of sediment to runoff and sediment concentration and it proved to be a feasible approach as there was good correlation between them. The higher human activity contributions to sediment decline resulted when the concentration factor was taken into account. Runoff decreased by 18.9, 8.3, 35.6, 34, 36.1, and 56.6% from upstream to downstream, respectively, while sediment concentration decreased by 21.5, 62.6, 45.2, 61.6, 52.9, and 39.3%, with a greater decrease in concentration than runoff in most of the catchments, especially in the middle reaches, where soil and water conservation measures on the Loess Plateau reduced the capacity to produce sand and resulted in a decrease erosion of sediment, while in the lower reaches with narrow catchment, sediment was probably mainly affected by reduced channel erosion with decreased runoff. The contribution of each factor varies, but there was consistency in the human activity's contribution, suggesting that human activities dominated the decline of sediment in all catchments. More than 90% of the sediment originated from the Loess Plateau (Qiu et al. 2024), and the sediment of Longmen Station decreased from 732 million tons during T₁ to 176 million tons during T₂. Although the contributions of the different river reaches to the reduction of sediment were consistent, the greatest efforts have been made to the reduction of sand in the middle reaches, and ecological projects in this area were acknowledged to be an efficient way of decreasing sediment in the YRB for higher sensitivity of sediment to surface change (Gao et al. 2023; Jiang & Liu 2023), which provides an alternative path to control sediment as there was excessive dam regulation. However, the lasting downward trend in sediment was likely to increase the risk of delta erosion (Yi et al. 2022), and it is a long-term goal to achieve water–sediment balance in the YRB through artificial regulation. In addition, most studies, including our study, treated different human activities such as grazing, building terraces and dams, converting farmland to forest and so on as a single category, that is, ecological restoration or land use/cover change, and further breakdown of the contributions made by these measures in the future could help to better understand the dominant factors of runoff and sediment changes in the YRB.

The potential limitations and uncertainties of this paper are (1) the inaccuracy due to interpolation of meteorological stations distributed unevenly within the watershed; (2) the first-order elasticity-based Budyko framework is non-closed because the climate and human factors are not independent of each other as assumed (Yu et al. 2021; Swain et al. 2023); (3) besides, sediment is closely related to precipitation intensity (Xu et al. 2021; Zhang et al. 2023), and it could not fully reflect the contribution of climate factors to sediment change with precipitation focused only.

This study revealed the climate change in the YRB in the last 50 years and quantified the contributions of climate and human factors to the water–sediment changes in the catchment areas of different hydrological stations based on the Budyko framework. The main conclusions obtained were as follows:

• (1) Precipitation and potential evaporation in the YRB showed a non-significant decreasing trend (P > 0.05), in which the increase of precipitation is mainly concentrated in the upper reaches, while the decrease of potential evaporation is mainly concentrated in the lower reaches.

• (2) The contributions of climate change and human activities to the reduction of runoff in the catchment areas of six hydrological stations in the YRB range from 7.6 to 19.7% and from 80.3 to 92.4%, respectively, and the magnitude of the influence of different elements was n > P > ET₀, and the intensity of the human activities in each river section was manifested as the middle reaches > lower reaches > upper reaches.

• (3) The contributions of climate and human factors to sediment reduction in the catchments of six hydrological stations in the YRB were −0.3 to 9.4% and 90.6–100.3%, respectively, and the influence magnitude of different elements was mainly shown as C > n > P >ET₀. The midstream region makes the greatest contribution to sediment reduction.

This work is supported by the Shaanxi Provincial Science and Technology Department (2024QCY-KXJ-093) and the National Natural Science Foundation of China (U1203182).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.