Impact of Changjiang-Hanjiang Water Diversion Project on the runoff process in the lower reach of Hanjiang River Authors: Jin Shao, Yuxin Zhang, Jiming Jin, Taohua He, Qian Luo, Yuan Zhou, Yun Kong, Chuan Liang Published: 2026 ABSTRACT Inter-basin water transfer is a commonly adopted measure for improving hydrological and ecological conditions within watersheds. However, systematically quantifying the actual improvement effects of water diversion projects on watershed runoff conditions remains a weak link in current hydraulic engineering evaluation. This study conducts complex modeling of the Changjiang-Hanjiang Water Diversion Project (CHWD), the Middle Route of South-to-North Water Diversion Project (MR-SNWD), the Dongjing River diversion, and the cascade reservoirs in the middle and lower reaches of the Hanjiang River with Soil and Water Assessment Tool (SWAT). Based on this modeling framework, the daily runoff process is simulated in the lower Hanjiang River under a scenario with no CHWD, and the impact of the CHWD on the runoff process is investigated in the lower Hanjiang River. The results indicate that: (1) the water supplementation from the CHWD is insufficient to balance the reduction impact on downstream Hanjiang River runoff caused by water withdrawal from the MR-SNWD; (2) the water supplementation from the CHWD is primarily utilized to supply the Dongjing River diversion, with limited effect on increasing runoff in the Hanjiang River section downstream of Dongjing River; (3) the water supplementation from the CHWD contributes to enhancing the stability of the runoff process in the lower Hanjiang River. This study yields a significant insight that gravity-flow water conveyance is suboptimal for inter-basin water transfer in flat terrain regions. These findings offer valuable implications for the planning and design of such projects globally. Introduction Water resources are essential natural resources for human survival and development1–3. In view of climate change and increasing water demand4–7, large-scale and long-distance inter-basin water transfer has become a common measure to alleviate regional water scarcity8–10. On one hand, inter-basin water transfer has improved water resource utilization efficiency, promoted spatial equilibrium in water resource distribution, and effectively mitigated contradictions between water supply and demand11–13. On the other hand, inter-basin water transfer has caused varying degrees of impact on the hydrological and ecological environments of both water-exporting and water-receiving basins14–17, resulting in notable changes in runoff processes, ecological environments, and social services18–20. Consequently, inter-basin water transfer has become a key factor constraining regional economic and social development, urgently requiring broader social attention21. The Hanjiang Basin represents a typical case involving multiple water diversion projects, including the Changjiang-Hanjiang Water Diversion Project (CHWD), the Middle Route of South-to-North Water Diversion Project (MR-SNWD), the Hanjiang-Weihe Water Diversion (HWWD) Project, and the under-construction Changjiang-Hanjiang Water Supplement Project (CHWS). The MR-SNWD has effectively alleviated water resource shortages in parts of northern China and improved the ecological environment in these regions to a certain extent13,22. However, due to the holistic, regional, and interconnected characteristics of inter-basin water transfer23,24, it has also placed considerable pressure on water demand in the Hanjiang Basin. Moreover, rapid economic and social development within the Hanjiang Basin has led to surging water demand and urgent pollution prevention needs, further intensifying the contradiction between water supply and demand in the basin25. A comprehensive and in-depth analysis of the impact of the CHWD on the runoff process in the lower Hanjiang River not only helps improve the comprehensive utilization efficiency of water resources in the Hanjiang Basin and alleviate regional water supply-demand contradictions26,27, but also provides important reference for exploring solutions to global water scarcity problems. The impact of inter-basin water transfer on ecosystem service supply and demand has long been a research focus in the field of ecological environment28,29, but studies have primarily concentrated on economic optimization and ecological compensation of inter-regional water resource allocation30,31. Research on changes in watershed runoff processes under the combined effects of water export, water import, and joint operation of cascade reservoirs remains limited. The construction of hydropower projects on the Periyar and Muvattupuzha Rivers in India has induced remarkable alterations in downstream flow regimes both pre- and post-implementation32. Additionally, the operation of storage and discharge at the Umiam Reservoir, as well as its inter-basin water transfer to the Umtru River, has exerted a distinct influence on the runoff processes of the Umiam River in India33. Furthermore, the runoff volumes of the Snowy River (Australia) and Churchill River (Canada) have experienced significant reductions due to inter-basin water export, whereas the base flow and peak discharge of the River Wear (UK) have increased following inter-basin water import34. In the case of Peru, water diversion from the Camaná River to the Quilca-Chili River has been shown to decrease low-flow discharge and increase flood frequency, while reservoirs within the two basins reduce their respective peak discharge and enhance low-flow discharge35. Quantitatively analyzing the impact of inter-basin water transfer on downstream runoff processes has been a persistent challenge36, while numerical simulation represents an effective approach for investigating such problems37–39. The Soil and Water Assessment Tool (SWAT) has been widely applied to simulate watershed hydrological and ecological processes40,41. This model incorporates reservoir regulation simulation modules and can also simulate inter-basin water transfer, making it a semi-distributed hydrological model with comprehensive and powerful functions. This study conducts complex modeling of the CHWD, the MR-SNWD, the Dongjing River diversion, and the cascade reservoirs in the middle and lower reaches of the Hanjiang River with SWAT to simulate the daily runoff process in the lower Hanjiang River under a scenario with no CHWD (No-CHWD). Subsequently, the changes of the daily runoff process are analyzed in the lower Hanjiang River before and after the CHWD, and the simulated values are compared with the observed values of runoff during different periods in the lower Hanjiang River under the No-CHWD scenario. Based on it, the impact of the CHWD on the runoff process is investigated in the lower Hanjiang River. Data and method Study area description The Hanjiang Basin is located between 30°10′-34°20′ N and 106°15′-114°20′ E, covering an area of approximately 159,000 km2 (Fig. 1). The basin terrain is characterized by high elevations in the northwest and low elevations in the southeast, with a typical subtropical monsoon climate. The multi-year average temperature ranges from 12 to 16 °C, and the multi-year average precipitation ranges from 700 to 1,100 mm, primarily concentrated from May to October each year. The Hanjiang River is the longest primary tributary of the Changjiang River, with a main channel length of approximately 1,577 km. The upstream section extends to Danjiangkou, the middle reach from Danjiangkou to Nianpanshan, and the downstream section below Nianpanshan. The Hanjiang River main channel has a high degree of cascade reservoirs development, with seven reservoirs constructed in the Danjiangkou-Xinglong section alone. The water intake point for the MR-SNWD is located at the Danjiangkou Reservoir in the upstream section, while the intake point for the HWWD is located at the Huangjinxia Reservoir on the Hanjiang River mainstream and conveys water through the Sanhekou Reservoir on a tributary of the Hanjiang River (Fig. 1). The CHWD is a gravity-flow water diversion project that diverts water from the Jingjiang section of the Changjiang River to the Xinglong section at Gaoshibei Town on the Hanjiang River, achieving full gravity-flow water conveyance through natural water level differences. Its outlet is located approximately 2.8 km downstream from the Xinglong Dam on the right bank (Fig. 1). Additionally, the outlet of the under-construction CHWS is planned to be located approximately 5 km downstream from the Danjiangkou Dam on the right bank. Fig. 1Geographical diagram of the Hanjiang Basin. Geographical diagram of the Hanjiang Basin. Basic data sources and processing The basic data used in this study primarily include: (1) a Digital Elevation Model (DEM) with a spatial resolution of 90 m; (2) Annual China Land Cover Dataset (CLCD) produced by Wuhan University with a spatial resolution of 30 m, covering the period 1985–2015, with one map selected every five years, where the time periods represented by each map are shown in Table 1; (3) a 1:1,000,000 soil type map produced by the Nanjing Institute of Soil Science, Chinese Academy of Sciences; (4) average daily runoff from three hydrological stations—Huangzhuang (HZ), Xiantao (XT), and Qianjiang (QJ)—during 1979–2020, and from Xinglong hydrological station (XL) during 2014–2020, where QJ monitors the average daily diversion flow of the Dongjing River, and XL monitors the total incoming water volume from the Xinglong Reservoir and the CHWD; (5) average daily runoff of the CHWD main canal and annual water withdrawal of the MR-SNWD during 2014–2020; (6) characteristic parameters and daily management data for four reservoirs: Danjiangkou, Wangfuzhou, Cuijiaying, and Xinglong; (7) data from 23 national basic meteorological stations in the Hanjiang Basin selected from the China Daily Surface Climate Dataset (V3.0) (Fig. 1), with main meteorological elements including precipitation, maximum and minimum temperature, relative humidity, wind speed, and sunshine duration, covering the period 1976–2020. Table 1Correspondence between actual years and representative periods for land use types.Actual years1985199019952000200520102015Representative periods1976–19871988–19921993–19971998–20022003–20072008–20132014–2020 Table 1Correspondence between actual years and representative periods for land use types.Actual years1985199019952000200520102015Representative periods1976–19871988–19921993–19971998–20022003–20072008–20132014–2020 Correspondence between actual years and representative periods for land use types. Since the SWAT requires radiation input, the daily sunshine duration must be converted into daily radiation at the top of the atmosphere (H0) using the following formulas42:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${H}_{0}=\frac{24}{\pi\:}{I}_{sc}{E}_{0}\left({W}_{s}\mathrm{sin}\delta\:\mathrm{sin}\phi\:+\mathrm{sin}{W}_{s}\mathrm{cos}\delta\:\mathrm{cos}\phi\:\right)$$\end{document} where Isc is the solar constant, taken as 4.9212 MJ m− 2 h− 1, E0 is the Earth’s orbital eccentricity correction factor, Ws is the hour angle (rad), δ is the solar declination, and ϕ is the geographic latitude. The calculation formula for E0 is:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{*{20}c} {E_{0} = \left( {\frac{{r_{0} }}{r}} \right)^{2} = 1 + 0.033{\mathrm{cos}}\left( {\frac{{2\pi \:d_{n} }}{{D_{Y} }}} \right)} \\ \end{array}$$\end{document} where r0 is the average Earth-Sun distance, r is the Earth-Sun distance on a given day, dn is the day order in the year, and DY is the total number of days in the year. The calculation formula for the hour angle (Ws) is:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{*{20}c} {W_{s} = cos^{{ - 1}} \left( { - {\mathrm{tan}}\delta \:{\mathrm{tan}}\phi \:} \right)} \\ \end{array}$$\end{document} The calculation formula for the solar declination (δ) is:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{*{20}c} {\delta \: = {\mathrm{sin}}^{{ - 1}} \left\{ {0.4{\mathrm{sin}}\left[ {\frac{{2\pi \:}}{{D_{Y} }}\left( {d_{n} - 82} \right)} \right]} \right\}} \\ \end{array}$$\end{document} Under clear sky conditions, the daily radiation (HL) at the ground surface is calculated as:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{*{20}c} {H_{L} = \beta \:H_{0} } \\ \end{array}$$\end{document} where β is the transparency coefficient of the atmosphere to solar radiation under clear sky conditions, usually taken as around 0.8. The time interval between sunrise and sunset is defined as the day length (TL), assuming the solar elevation angle is 0 at sunrise and sunset, and its calculation formulas are:6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T_{L} = \frac{{2W_{S} }}{{\omega \:}}}$$\end{document}7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}\omega\:=\frac{\pi\:}{12}\end{array}$$\end{document} where ω is the angular velocity of Earth’s rotation (rad h− 1). Generally, the formula for calculating daily radiation (H) at the ground surface can be expressed as:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${H = H_{L} \left( {a + b\frac{T}{{T_{L} }}} \right)}$$\end{document} where T is the daily sunshine duration, a and b are empirical constants, usually taken as 0.248 and 0.752, respectively, in China. Research method The SWAT was used to simulate the daily runoff process in the lower Hanjiang River from 1979 to 2020, which was compared with observed daily runoff processes to evaluate the applicability and accuracy of the SWAT for runoff simulation in the Hanjiang Basin. Since the MR-SNWD and CHWD commenced operation in 2014, the entire simulation period was divided into two periods: 1979–2013 and 2014–2020, representing periods before and after the CHWD, with 1976–1978 set as the model warm-up period. Simultaneously, the operational period of the CHWD from 2014 to 2020 was subdivided into daily and seasonal time scales for investigation. The accuracy evaluation indicators adopted by the model are the Nash-Sutcliffe Efficiency coefficient (NSE) and the coefficient of determination (R2), with calculation formulas as follows:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${NSE = 1 - \frac{{\sum {\:_{{i = 1}}^{n} } \left( {x_{i} - y_{i} } \right)^{2} }}{{\sum {\:_{{i = 1}}^{n} } \left( {y_{i} - \mathop y\limits^{ - } } \right)^{2} }}}$$\end{document}10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R^{2} = \frac{{[\sum {\:_{{i = 1}}^{n} } (x_{i} - \mathop x\limits^{ - } )(y_{i} - \mathop y\limits^{ - } )]^{2} }}{{\sum {\:_{{i = 1}}^{n} } \left( {x_{i} - \mathop x\limits^{ - } } \right)^{2} \sum {\:_{{i = 1}}^{n} } \left( {y_{i} - \mathop y\limits^{ - } } \right)^{2} }}}$$\end{document} where xi and yi are the i-th elements of the simulated series X and observed series Y, respectively; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\bar{x}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\bar{y}$$\end{document} are the means of the two series, and n is the number of elements in the series. This study is based on extensive observed flow data from the CHWD, the MR-SNWD, and the Dongjing River, as well as characteristic and management information data from the cascade reservoirs in the middle and lower reaches of the Hanjiang River. After extensive parameter calibration work through SWAT-CUP to achieve high simulation accuracy, the water supplementation flow from the CHWD was set to zero and reintroduced into the SWAT for re-simulation to obtain the runoff process in the lower Hanjiang River under the No-CHWD scenario. Accordingly, the simulated values are compared with the observed values of runoff during different periods at XL and XT under this scenario to investigate the direct impact of the CHWD on the runoff process in the lower Hanjiang River. Additionally, considering that in reality the water supplementation flow from the CHWD is typically coordinated with the discharge flow from its upstream cascade reservoirs, the daily runoff processes at HZ and XT before and after the CHWD are compared to investigate the indirect and combined impacts of the CHWD on the runoff process in the lower Hanjiang River. This study primarily investigates the impact of the CHWD on the runoff process in the lower Hanjiang River from three aspects: mean level, volatility, and trend. The evaluation indicators adopted are mean (ε), standard deviation (σ), and trend change rate with year(yr) scale (v), respectively. The specific calculation formulas are:11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\epsilon \: = \bar{x} = \frac{1}{n}\sum \: _{{i = 1}}^{n} x}$$\end{document}12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{*{20}c} {\sigma \: = \sqrt {\frac{1}{{n - 1}}\sum \: _{{i = 1}}^{n} \left( {x_{i} - \bar{x}} \right)^{2} } } \\ \end{array}$$\end{document}13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${v = 365k}$$\end{document} where k is the slope of the trend line for series X. If standard deviation σ is defined as an evaluation indicator of absolute volatility in the runoff process, then the coefficient of variation µ can be used as an evaluation indicator of relative volatility in the runoff process, expressed as:14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mu \: = \frac{{\sigma \:}}{{\epsilon \:}}}$$\end{document} Model calibration and evaluation During the period 1979–2013, before the CHWD, the simulated daily runoff processes at Huangzhuang and Xiantao stations generally fit well with the observed daily runoff processes, with NSE reaching 0.80 and 0.72, respectively (Fig. 2). Simultaneously, R2 for runoff simulation at both stations exceeded 0.70, indicating that the simulated daily runoff processes are relatively consistent with the observed daily runoff processes. Except for significant differences between simulated and observed values at a few peak flow points, the errors between simulated and observed daily runoff values at both stations are relatively small (Fig. 2). This is primarily because the actual runoff at HZ increases sharply due to discharge from the cascade reservoirs in the middle and upper reaches of the Hanjiang River and then decreases sharply during recession. Current watershed hydrological models, including the SWAT, generally have difficulty accurately depicting such processes of sharp changes in channel runoff caused by highly subjective human regulation activities, especially in rivers like the Hanjiang River where the regulatory effects of main channel cascade reservoirs are pronounced. However, this does not affect the overall simulation performance (Fig. 2). Therefore, the SWAT is applicable to runoff simulation in the Hanjiang Basin with relatively high simulation accuracy. Fig. 2Observed and simulated daily runoff processes in the lower Hanjiang River before the CHWD. NSE and R2 represent the Nash-Sutcliffe Efficiency coefficient and coefficient of determination, respectively. Observed and simulated daily runoff processes in the lower Hanjiang River before the CHWD. NSE and R2 represent the Nash-Sutcliffe Efficiency coefficient and coefficient of determination, respectively. During the period 2014–2020, after the CHWD, NSE and R2 for runoff simulation at HZ were both 0.77, essentially consistent with the simulation accuracy before the CHWD. The simulated daily runoff process fits well with the observed values overall, with high consistency in their variations (Fig. 3a). During the same period, the simulation accuracy of the runoff process at XL was slightly better than that at HZ, with NSE and R2 of 0.78 and 0.80, respectively (Fig. 3b). Although the runoff simulation accuracy at Xiantao station during 2014–2020 was inferior to that at Huangzhuang and Xinglong stations, its NSE and R2 also reached 0.71 and 0.72, respectively, indicating that the simulated daily runoff process at XT generally fits well with the observed daily runoff process, with relatively consistent variations between the two (Fig. 3c). Fig. 3Observed and simulated daily runoff processes in the lower Hanjiang River after the CHWD. NSE and R2 represent the Nash-Sutcliffe Efficiency coefficient and coefficient of determination, respectively. Observed and simulated daily runoff processes in the lower Hanjiang River after the CHWD. NSE and R2 represent the Nash-Sutcliffe Efficiency coefficient and coefficient of determination, respectively. The simulation results for the daily runoff process in the lower Hanjiang River both before and after the CHWD are relatively good, indicating that the SWAT achieves required accuracy when applied to simulate the daily runoff process in the lower Hanjiang River under the complex scenario of joint water resource allocation by multiple hydraulic projects. Therefore, simulating the daily runoff process in the lower Hanjiang River under the No-CHWD scenario by setting the water supplementation flow from the CHWD to zero is feasible based on this foundation. Results Runoff changes in the lower Hanjiang River before and after the CHWD Before the CHWD during 1979–2013, the daily runoff at HZ was relatively large, with εHZ1 of 1,442 m3 s− 1 (Fig. 4). Simultaneously, the volatility of daily runoff was also considerable, with fluctuation amplitude comparable to its mean level, and the corresponding σHZ1 reaching 1,483 m3 s− 1 (Fig. 4). During this period, the daily runoff at HZ showed an overall decreasing trend year by year, with vHZ1 of -11 m3 s− 1 yr− 1 (Fig. 4). On the other hand, after the CHWD during 2014–2020, the daily runoff at HZ decreased significantly by 28.2%, with εHZ2 reduced to 1,035 m3 s− 1 (Fig. 4). The volatility of daily runoff also decreased significantly, with a reduction of up to 34.4%, and the corresponding σHZ2 of 972 m3 s− 1 (Fig. 4). Conversely, during this period, the overall trend of daily runoff at HZ changed to an increasing pattern year by year, with vHZ2 of 57 m3 s− 1 yr− 1 (Fig. 4). Fig. 4Changes in daily runoff at Huangzhuang station (HZ) before and after the CHWD. ε, σ, v represent mean, standard deviation, and trend change rate, respectively. Subscripts 1 and 2 represent before and after the CHWD, respectively. Changes in daily runoff at Huangzhuang station (HZ) before and after the CHWD. ε, σ, v represent mean, standard deviation, and trend change rate, respectively. Subscripts 1 and 2 represent before and after the CHWD, respectively. Before the CHWD, compared with HZ, the daily runoff at XT was relatively smaller, with εXT1 of 1,231 m3 s− 1 (Figs. 4 and 5). Simultaneously, the volatility of daily runoff at XT was also smaller than that at HZ, with fluctuation amplitude less than the mean level of daily runoff, and σXT1 of 1,084 m3 s− 1 (Figs. 4 and 5). During this period, the overall trend of daily runoff at XT was consistent with that at HZ, showing a year-by-year decrease also with vXT1 of -11 m3 s− 1 yr− 1 (Figs. 4 and 5). On the other hand, after the CHWD, the daily runoff at XT also decreased significantly by approximately 25.0%, with εXT2 of 924 m3 s− 1, which was 10.7% lower than εHZ2 at HZ during the same period (Figs. 4 and 5). The volatility of daily runoff at XT also decreased notably by 27.7%, with σXT2 of 783 m3 s− 1, which was 19.4% lower than σHZ2 at HZ during the same period (Figs. 4 and 5). Additionally, during this period, the overall trend of daily runoff at XT also changed to an increasing pattern year by year, with vXT2 of 59 m3 s− 1 yr− 1, essentially consistent with vHZ2 at HZ during the same period (Figs. 4 and 5). Fig. 5Changes in daily runoff at Xiantao station (XT) before and after the CHWD. ε, σ, v represent mean, standard deviation, and trend change rate, respectively. Subscripts 1 and 2 represent before and after the CHWD, respectively. Changes in daily runoff at Xiantao station (XT) before and after the CHWD. ε, σ, v represent mean, standard deviation, and trend change rate, respectively. Subscripts 1 and 2 represent before and after the CHWD, respectively. Runoff process in the lower Hanjiang River under the No-CHWD scenario Under the No-CHWD scenario, the simulated daily runoff at XL during 2014–2020 changed relatively noticeably compared with the observed daily runoff, with the simulated εXL0 of 938 m3 s− 1, representing a 10.2% decrease compared with the observed εXL2 (Fig. 6). Simultaneously, the volatility of simulated daily runoff at XL was less than its observed volatility, with the simulated σXL0 of 762 m3 s− 1, showing a reduction of up to 23.4% compared with the observed σXL2 (Fig. 6). Moreover, under this scenario, the daily runoff at XL showed an overall increasing trend year by year, with the simulated vXL0 of 70 m3 s− 1 yr− 1, representing a 7.9% decrease compared with the observed vXL2 during the same period (Fig. 6). Fig. 6Distribution pattern of daily runoff at Xinglong station (XL) under the No-CHWD scenario. ε, σ, v represent mean, standard deviation, and trend change rate, respectively. Subscript 2 indicates the period after the CHWD, and subscript 0 indicates the No-CHWD scenario. Distribution pattern of daily runoff at Xinglong station (XL) under the No-CHWD scenario. ε, σ, v represent mean, standard deviation, and trend change rate, respectively. Subscript 2 indicates the period after the CHWD, and subscript 0 indicates the No-CHWD scenario. Two typical periods during which the CHWD continuously and relatively concentratedly supplemented runoff in the lower Hanjiang River were from April 1 to July 22, 2017, and from May 9 to September 14, 2019. During the first water diversion period, the simulated average daily runoff at XL under the No-CHWD scenario was 774 m3 s− 1, representing a 17.7% decrease compared with the observed average daily runoff (Fig. 6). Moreover, under this scenario, the simulated maximum and minimum daily runoff at XL during this period were 1,055 and 630 m3 s− 1, respectively, representing decreases of 28.7% and 3.1% compared with the observed maximum and minimum daily runoff, respectively (Fig. 6). During the second water diversion period, the simulated average daily runoff at XL under the No-CHWD scenario was 642 m3 s− 1, representing a 21.0% decrease compared with the observed average daily runoff (Fig. 6). Under this scenario, the simulated maximum and minimum daily runoff at XL during this period were 1,015 and 510 m3 s− 1, respectively, representing decreases of 5.1% and 18.8% compared with the observed maximum and minimum daily runoff, respectively (Fig. 6). Under the No-CHWD scenario, the simulated daily runoff at XT during 2014–2020 did not change significantly compared with the observed daily runoff, with the simulated εXT0 of 878 m3 s− 1, representing less than 5.0% decrease compared with the observed εXT2 (Fig. 7). Simultaneously, the volatility of simulated daily runoff at XT was less than its observed volatility, with the simulated σXT0 of 600 m3 s− 1, also showing a reduction of 23.4% compared with the observed σXT2 (Fig. 7). Moreover, under this scenario, the simulated daily runoff at XT showed an overall increasing trend year by year, with the simulated vXT0 of 66 m3 s− 1 yr− 1, which was significantly higher than the observed vXT2 during the same period, with an increase of up to 11.9% (Fig. 7). Fig. 7Distribution pattern of daily runoff at Xiantao station (XT) under the No-CHWD scenario. ε, σ, v represent mean, standard deviation, and trend change rate, respectively. Subscript 2 indicates the period after the CHWD, and subscript 0 indicates the No-CHWD scenario. Distribution pattern of daily runoff at Xiantao station (XT) under the No-CHWD scenario. ε, σ, v represent mean, standard deviation, and trend change rate, respectively. Subscript 2 indicates the period after the CHWD, and subscript 0 indicates the No-CHWD scenario. During the first water diversion period, the simulated average daily runoff at XT under the No-CHWD scenario was 731 m3 s− 1, representing an 11.9% decrease compared with the observed average daily runoff (Fig. 7). Moreover, under this scenario, the simulated maximum and minimum daily runoff at XT during this period were 975 and 579 m3 s− 1, respectively, representing decreases of 23.9% and 0.6% compared with the observed maximum and minimum daily runoff, respectively (Fig. 7). During the second water diversion period, the simulated average daily runoff at XT under the No-CHWD scenario was 642 m3 s− 1, representing a 13.5% decrease compared with the observed average daily runoff (Fig. 7). Under this scenario, the simulated maximum and minimum daily runoff at XT during this period were 949 and 493 m3 s− 1, respectively, representing decreases of 16.0% and 9.5% compared with the observed maximum and minimum daily runoff, respectively (Fig. 7). Under the No-CHWD scenario, the simulated average and maximum seasonal runoff of spring and summer at XL during 2014–2020 were both lower than those of the observed seasonal runoff, respectively (Table 2). Therein, the simulated average and maximum seasonal runoff of spring decreased by 149 and 463 m3 s− 1 compared with the corresponding observed values, representing reductions of 16.5% and 31.4%, respectively. The simulated average and maximum seasonal runoff of summer decreased by 298 and 641 m3 s− 1 compared with the corresponding observed values, representing reductions of 22.2% and 27.5%, respectively (Table 2). Additionally, under this scenario during this period, the simulated maximum seasonal runoff of autumn and winter at XL decreased by 530 and 53 m3 s− 1 compared with the observed maximum values, representing reductions of 15.0% and 6.2%, respectively (Table 2). However, under the No-CHWD scenario, the simulated minimum seasonal runoff at XL during 2014–2020 were all larger than the observed minimum values, exceeding them by 203, 34, 39, and 18 m3 s− 1, representing increases of 46.1%, 7.5%, 5.8%, and 3.1%, respectively (Table 2). Table 2Comparison of seasonal runoff characteristic values at XL under the No-CHWD scenario (m3 s− 1) Under the No-CHWD scenario, the simulated average and maximum seasonal runoff of spring and summer at XT during 2014–2020 were both lower than their corresponding observed values (Table 3). Therein, the simulated average and maximum seasonal runoff of spring decreased by 108 and 341 m3 s− 1 compared with the corresponding observed values, representing reductions of 13.2% and 27.9%, respectively. The simulated average and maximum seasonal runoff of summer decreased by 223 and 517 m3 s− 1 compared with the corresponding observed values, representing reductions of 18.9% and 25.9%, respectively (Table 3). Furthermore, under this scenario during this period, the simulated maximum seasonal runoff of autumn and winter at XT decreased by 306 and 11 m3 s− 1 compared with the observed maximum values, representing reductions of 10.9% and 1.4%, respectively (Table 3). Whereas, under the No-CHWD scenario, the simulated minimum seasonal runoff at XT during 2014–2020 were all higher than the observed minimum values, exceeding them by 236, 94, 96, and 57 m3 s− 1, representing increases of 57.8%, 23.7%, 15.6%, and 10.8%, respectively (Table 3).PeriodAverageMaximumMinimumCHWDNo-CHWDCHWDNo-CHWDCHWDNo-CHWDSpring90475514731010440643Summer1343104523291688455489Autumn1250127035423012678717Winter682697849796578596 Table 2Comparison of seasonal runoff characteristic values at XL under the No-CHWD scenario (m3 s− 1) Under the No-CHWD scenario, the simulated average and maximum seasonal runoff of spring and summer at XT during 2014–2020 were both lower than their corresponding observed values (Table 3). Therein, the simulated average and maximum seasonal runoff of spring decreased by 108 and 341 m3 s− 1 compared with the corresponding observed values, representing reductions of 13.2% and 27.9%, respectively. The simulated average and maximum seasonal runoff of summer decreased by 223 and 517 m3 s− 1 compared with the corresponding observed values, representing reductions of 18.9% and 25.9%, respectively (Table 3). Furthermore, under this scenario during this period, the simulated maximum seasonal runoff of autumn and winter at XT decreased by 306 and 11 m3 s− 1 compared with the observed maximum values, representing reductions of 10.9% and 1.4%, respectively (Table 3). Whereas, under the No-CHWD scenario, the simulated minimum seasonal runoff at XT during 2014–2020 were all higher than the observed minimum values, exceeding them by 236, 94, 96, and 57 m3 s− 1, representing increases of 57.8%, 23.7%, 15.6%, and 10.8%, respectively (Table 3).PeriodAverageMaximumMinimumCHWDNo-CHWDCHWDNo-CHWDCHWDNo-CHWDSpring90475514731010440643Summer1343104523291688455489Autumn1250127035423012678717Winter682697849796578596 Comparison of seasonal runoff characteristic values at XL under the No-CHWD scenario (m3 s− 1) Under the No-CHWD scenario, the simulated average and maximum seasonal runoff of spring and summer at XT during 2014–2020 were both lower than their corresponding observed values (Table 3). Therein, the simulated average and maximum seasonal runoff of spring decreased by 108 and 341 m3 s− 1 compared with the corresponding observed values, representing reductions of 13.2% and 27.9%, respectively. The simulated average and maximum seasonal runoff of summer decreased by 223 and 517 m3 s− 1 compared with the corresponding observed values, representing reductions of 18.9% and 25.9%, respectively (Table 3). Furthermore, under this scenario during this period, the simulated maximum seasonal runoff of autumn and winter at XT decreased by 306 and 11 m3 s− 1 compared with the observed maximum values, representing reductions of 10.9% and 1.4%, respectively (Table 3). Whereas, under the No-CHWD scenario, the simulated minimum seasonal runoff at XT during 2014–2020 were all higher than the observed minimum values, exceeding them by 236, 94, 96, and 57 m3 s− 1, representing increases of 57.8%, 23.7%, 15.6%, and 10.8%, respectively (Table 3). Table 3Comparison of seasonal runoff characteristic values at XT under the No-CHWD scenario (m3 s− 1).PeriodAverageMaximumMinimumCHWDNo-CHWDCHWDNo-CHWDCHWDNo-CHWDSpring8187101221880408644Summer118195819941477396490Autumn1062117628132507617713Winter638687808797528585 Table 3Comparison of seasonal runoff characteristic values at XT under the No-CHWD scenario (m3 s− 1).PeriodAverageMaximumMinimumCHWDNo-CHWDCHWDNo-CHWDCHWDNo-CHWDSpring8187101221880408644Summer118195819941477396490Autumn1062117628132507617713Winter638687808797528585 Comparison of seasonal runoff characteristic values at XT under the No-CHWD scenario (m3 s− 1). Discussion After the CHWD, both the mean level and volatility of daily runoff in the lower Hanjiang River decreased significantly (Figs. 4 and 5), indicating that the water supplementation from the CHWD is insufficient to balance the reduction impact on downstream Hanjiang River runoff caused by water withdrawal from the MR-SNWD. Before and after the CHWD, the µ of daily runoff at HZ were 1.03 and 0.94, respectively, while those at XT were 0.88 and 0.85, respectively (Figs. 4 and 5). This indicates that after the CHWD, the reduction in relative volatility of the runoff process at both HZ and XT was significantly smaller than the reduction in absolute volatility, suggesting that the absolute volatility of the runoff process in the lower Hanjiang River is primarily influenced by the MR-SNWD and the CHWD, while its relative volatility is mainly controlled by climatic conditions in the Hanjiang Basin. Furthermore, since HZ and XT are located upstream and downstream of the CHWD, respectively, with the former only affected by water withdrawal from the MR-SNWD and the latter affected by both water withdrawal from the MR-SNWD and water supplementation from the CHWD, yet the v of daily runoff at HZ and XT after the CHWD are essentially consistent (Figs. (Figs. 4 and 5), this indicates that the water supplementation from the CHWD has limited impact on the trend of runoff process in the lower Hanjiang River. Under the No-CHWD scenario, the simulated daily runoff at XL decreased relatively significantly, while the decrease at XT was relatively insignificant (Figs. 6 and 7), indicating that most of the water supplementation from the CHWD is diverted by the Dongjing River located between XL and XT for agricultural irrigation, with limited effect on increasing runoff in the Hanjiang River section downstream of Dongjing River. Simultaneously, the volatility of simulated daily runoff at both XL and XT decreased noticeably compared with their observed volatility (Figs. 6 and 7), indicating that the water supplementation from the CHWD contributes to enhancing the stability of the runoff process in the lower Hanjiang River. Additionally, under this scenario, the simulated daily runoff at XL showed a relatively obvious decreasing trend, while the simulated daily runoff at XT exhibited an increasing trend (Figs. 6 and 7), further confirming that the water supplementation from the CHWD is primarily utilized to supply the Dongjing River diversion for agricultural irrigation, and that the diversion flow of Dongjing River has a distinct impact on the trend of runoff process in its downstream Hanjiang River section. Under the No-CHWD scenario, comparing with the corresponding observed values, the simulated average seasonal runoff of spring and summer at both XL and XT decreased significantly (Tables 2 and 3), indicating that the water supplementation effect of the CHWD on runoff in the lower Hanjiang River occurs primarily during spring and summer. Simultaneously, comparing with the corresponding observed values, the simulated maximum seasonal runoff at both XL and XT all decreased, while their simulated minimum values all increased, and the simulated average seasonal runoff of autumn and winter at both stations also increased (Tables 2 and 3), suggesting that the water supplementation effect of the CHWD on runoff in the lower Hanjiang River occurs primarily during periods of high river water levels. During periods of low river water levels, the difficulty for the CHWD to provide water supplementation results in more pronounced simulation deviations. Even during the two typical periods when the CHWD supplemented runoff into the lower Hanjiang River, simulation deviations were more evident during periods of low river water levels (Figs. 6 and 7), which is consistent with the gravity-flow water diversion method of the CHWD for supplementing runoff into the lower Hanjiang River. Most inter-basin water diversion projects worldwide are situated in mountainous and hilly regions, where topographic undulations generate substantial water head differences, thereby enabling effective gravity-flow water diversion between basins. Examples include India’s Umiam-Umtru Water Diversion, Australia’s Snowy Mountains Scheme, Canada’s Churchill River Diversion, and the UK’s Kielder Water Scheme32–35. However, the CHWD is located in the hinterland of the Jianghan Plain, where minimal topographic relief results in insufficient hydraulic gradient within the main canal. Hence, the Changjiang River’s runoff cannot be diverted into the Hanjiang River via gravity. Conversely, the Eastern Route of the SNWD, which is also situated in the relatively flat region of eastern China, employs a stepwise pumping system to lift water to Dongping Lake for long-distance conveyance43. This also proves from the side that the replenishing effect of the CHWD on the Hanjiang River is not significant, which is reasonable to some extent. Conclusions After the CHWD, runoff in the lower Hanjiang River decreased significantly, and the water supplementation from the CHWD is insufficient to balance the reduction impact on downstream Hanjiang River runoff caused by water withdrawal from the MR-SNWD. The water supplementation of the CHWD is primarily utilized to supply the Dongjing River diversion, with limited effect on increasing runoff in the Hanjiang River section downstream of Dongjing River, but it contributes to enhancing the stability of the runoff process in the lower Hanjiang River. The gravity-flow supplementation effect of the CHWD on runoff in the lower Hanjiang River occurs primarily during spring and summer when river water levels are relatively high, whereas is not significant during periods of low river water levels. This provides a significant insight that gravity-flow water conveyance is not the optimal method for inter-basin water transfer in flat terrain regions. Therefore, it is necessary to improve the runoff conditions of the downstream Hanjiang River through proactive pumping. Since the CHWD commenced operation in 2014, only several years have elapsed, and the observed runoff series at the CHWD main canal and relevant hydrological stations during the period have limited representativeness, making it difficult to fully reflect the actual water conveyance patterns of the CHWD. Further in-depth research is needed regarding the impact of inter-basin water transfer on the runoff process in the Hanjiang Basin.