Attribution of Hydrologic-Cycle Changes to Climate Change and Human Activities in the Shaying River Basin, China

Abstract

The Shaying River Basin, the largest tributary basin of the Huaihe River in eastern China, is a representative north–south transitional basin with strong mountain–plain contrasts and intensive human disturbance. This study quantified hydrologic-cycle changes and attributed them to climate change and human activities using the Simulation Water-cycle and Allocation Model (SWAM). Hydrologic processes were simulated for a baseline period (1971–1980) and a changed-environment period (2001–2010). Four scenarios were designed to isolate the effects of climate change, land-use change, and water-resource development. The results show clear spatial differences in hydrologic response. At the basin scale, human activities and climate change contributed 59% and 41%, respectively, to hydrologic-cycle changes. In the plain area, human activities were dominant, contributing 67%, whereas in the mountainous area, climate change was dominant, contributing 61%. Water-resource development reduced surface-water, soil-water, and groundwater storage, especially in the plain area. These findings highlight the need for differentiated water-management strategies in transitional basins and provide a representative case for understanding climate–human impacts on hydrologic-cycle changes.

1. Introduction

In 2021, the Intergovernmental Panel on Climate Change (IPCC) released its latest report in Geneva, stating that greenhouse gas emissions from human activities have caused approximately 1.1 °C of global warming above the 1850–1900 baseline and that the global average temperature is expected to reach or exceed 1.5 °C of warming within the next two decades [1]. Climate change leads to rising atmospheric temperatures and shifts in precipitation patterns [2]. Meanwhile, with continued population growth, economic development, and scientific advancement, the scope and intensity of human activities are also increasing [3]. In China, rapid economic development and population concentration have substantially altered the terrestrial hydrologic cycle across various watersheds and regions [4]. As a result, the impacts of climate change and human activities on hydrological processes and the evolution of water resources have long been a central focus of scientific research and public concern [5,6,7].

A variety of attribution approaches have been used to assess the impacts of climate change and human activities on hydrologic processes, including the Budyko framework, climate–runoff elasticity analysis, double-mass curves, hydrological change-point analysis, statistical decomposition methods, and process-based hydrological models. Classical statistical and Budyko-type approaches are useful for diagnosing basin-scale runoff changes, but they usually simplify spatial heterogeneity and have limited capacity to represent groundwater–surface-water interactions, irrigation withdrawals, reservoir and sluice regulation, land-use-dependent evapotranspiration, and the separation of mountainous and plain hydrologic responses. These limitations are particularly important in basins where natural and artificial water-cycle processes are strongly coupled.

Currently, extensive research on hydrologic-cycle theory has been conducted both domestically and internationally [8,9]. Studies examining the impacts of a changing environment on the hydrologic cycle generally fall into two categories: quantitative assessments of how environmental changes affect hydrological processes and water resources [10,11], and attribution analyses of changes in hydrologic-cycle elements under evolving environmental conditions [9]. Since the late 1970s, considerable research has been undertaken abroad to address the effects of climate change on hydrological processes and water resources [12]. Furthermore, precipitation data has indicated a global increase in wetting conditions since 1950 [13]. The IPCC has carried out numerous studies on the attribution of global climate change, its implications for water resources, and potential adaptation strategies while also coordinating the periodic publication of global climate assessment reports [1]. In China, quantitative research on the impact of climate change on the hydrologic cycle began in the 1980s [14]. In the 21st century, rising temperatures have intensified evapotranspiration without a corresponding increase in precipitation, thereby increasing the intensity and frequency of hydrological droughts worldwide [15]. At the same time, human activities—including water extraction, land-use and land-cover change, urbanization, industrialization, and reservoir regulation—have significantly altered hydrologic cycles [16], particularly as climate change itself is driven by ongoing human activities [17].

However, research on hydrologic cycles in north–south transitional zone regions remains insufficient and often adopts a monolithic perspective, frequently overlooking the distinct characteristics and variations in hydrological patterns between mountainous and plain areas [18,19]. North–south transitional zone regions are generally characterized by the interlacing of climate systems, water resources, and human activities [20]. Due to the influence of these interacting factors, such regions possess unique hydro-meteorological conditions that substantially increase the complexity of their hydrologic cycles [21]. The Huaihe River Basin is situated within China’s north–south transitional zone, with the Shaying River as its largest tributary. Influenced by these climatic conditions, the basin experiences dry winters and springs, hot and rainy summers and autumns, sharp transitions between cold and warm weather, and a rapid onset of flood seasons [22]. These factors result in frequent water-related disasters and an uneven spatiotemporal distribution of precipitation, with 50% to 80% of annual rainfall concentrated between June and September. Furthermore, the Shaying River Basin is characterized by high population density, rapid urbanization, large-scale industrial and agricultural production, and intensive regulation of water conservancy projects, including the widespread presence of sluice gates [23]. The dual influence of climate change and intense human activity causes the hydrologic cycle in this north–south transitional zone region to fluctuate drastically [24], leading to significant spatiotemporal shifts in hydrological elements, altering water resource availability, and ultimately impacting the socioeconomic development of the Shaying River Basin [25].

To address these gaps, this study applies the Simulation Water-cycle and Allocation Model (SWAM) to the Shaying River Basin—the largest tributary basin of the Huaihe River Basin—to simulate hydrologic-cycle changes under baseline and changed-environment conditions. The specific objectives are to: (1) characterize the spatiotemporal changes in major hydrologic-cycle components, including evapotranspiration, surface runoff, soil-water storage, and groundwater storage; (2) quantify the relative contributions of climate change and human activities using scenario-based attribution; and (3) compare the dominant drivers in mountainous and plain areas to provide implications for differentiated water-resource management.

2. Materials and Methods

2.1. Study Area

The Shaying River Basin is the largest tributary basin of the Huaihe River Basin in eastern China. The river originates in the Funiu Mountains of Henan Province, flows through Henan and Anhui Provinces, and joins the main stream of the Huaihe River near Zhengyangguan. The basin is located between approximately 111.57–116.43° E and 32.31–34.52° N, with a drainage area of approximately 40,000 km². The topography generally decreases from the northwestern mountainous region toward the southeastern plain, producing marked differences in runoff generation, groundwater conditions, land use, and human water use between mountainous and plain areas.

The basin is located in a warm-temperate monsoon climate zone and has typical characteristics of a north–south climatic transition region. Winters and springs are generally dry, whereas summers and autumns are hot and humid. Precipitation is highly seasonal, with approximately 50–80% of annual rainfall concentrated between June and September. Flood-season runoff accounts for a large proportion of annual runoff, and alternating drought and flood events occur frequently. The mean annual precipitation is approximately 769.5 mm, generally decreasing from the southeast to the northwest. This uneven seasonal and spatial distribution of precipitation is a key natural factor controlling hydrologic-cycle variability in the basin. The location, topography, and river network of the Shaying River Basin are shown in Figure 1.

The basin contains diverse soil types, including fluvo-aquic soil, brown soil, and Shajiang black soil. Shajiang black soil occupies a substantial part of the basin and is characterized by low hydraulic conductivity and poor drainage capacity. During concentrated rainfall events, rapid runoff from mountainous areas may enter the plains, where low infiltration capacity and weak drainage conditions can intensify waterlogging. These soil and terrain characteristics are important for understanding the spatial differences in surface runoff, soil-water storage, and groundwater recharge.

The plain area of the Shaying River Basin is densely populated and intensively cultivated. It supports large-scale agricultural production and has experienced urban expansion, industrial development, irrigation growth, groundwater abstraction, and intensive hydraulic regulation. Reservoirs, sluices, gates, and inter-basin water transfers have been constructed to support water supply, irrigation, and flood control. Therefore, the basin provides a representative case for examining the combined effects of climate variability and human water use on hydrologic-cycle changes.

2.2. SWAM Model Description and Rationale

SWAM was selected because it is designed to represent the coupled “natural–artificial” water cycle and can simulate multiple hydrologic components, including evapotranspiration, soil-water redistribution, groundwater movement, runoff generation and concentration, water withdrawals, and hydraulic engineering operations. Compared with lumped or purely statistical attribution methods, SWAM is more suitable for this study because the Shaying River Basin contains strong mountain–plain contrasts, intensive irrigation, groundwater abstraction, and extensive regulation by reservoirs, sluices, and water-transfer projects. In this study, SWAM was not used for model intercomparison but as a process-based diagnostic tool to quantify decadal hydrologic-cycle differences under prescribed climate, land-use, and water-use scenarios.

The SWAM framework includes simulations of evapotranspiration within grid cells, soil-water redistribution, groundwater movement, snow accumulation and melting, runoff generation, runoff concentration, and the operation of water conservancy projects. The model further represents hillslope and plain-area confluence, groundwater dynamics in the plains, and hydraulic infrastructures such as lakes, reservoirs, sluices, dams, and inter-basin water-transfer projects. These modules allow the model to describe both natural hydrological processes and artificial water-use processes within a unified water-balance framework [26,27].

2.2.1. Evapotranspiration

The model considers three types of evapotranspiration: evaporation from bare soil, transpiration from various vegetation types, and canopy evaporation induced by different vegetation cover types. Evapotranspiration within each computational grid cell is calculated as the weighted sum of contributions from all underlying vegetation cover types. Vertically, each grid cell consists of a three-layer soil column and a two-layer canopy structure (or bare soil), with the upper and lower canopy layers further subdivided into dry and wet components. Evapotranspiration for each vegetation type is computed based on parameters including surface evaporation resistance, leaf stomatal resistance, potential evaporation capacity, and aerodynamic resistance. For bare soil, only evaporation from the upper soil layer is considered, which varies according to underlying surface heterogeneity—such as infiltration conditions, soil type, and evaporation factors. It is stipulated that when a snow layer exists on the surface, the snow completely covers the bare soil layer and the lower canopy layer; therefore, evaporation from the lower canopy layer and bare soil is not considered under such conditions.

2.2.2. Soil-Water Redistribution

As the core link of soil-water redistribution, the infiltration process takes precipitation, snowmelt water and drainage from adjacent grids as three core recharge sources in the model, and a layered soil model is adopted for refined simulation of water movement in the unsaturated zone. When simulating the soil water within the computational grid, the model follows the bottom-up water movement logic and prioritizes the calculation of water transfer from each soil layer to its upper layer (the calculation process is shown in Figure 2), with engineering simplification applied to the upward recharge process: upward water transfer is only triggered when the moisture content of the lower soil layer exceeds its porosity, and the moisture content of the lower layer remains at the porosity threshold after transfer.

($i=NRL-1$, NRL is the number of soil layers, $RDth\left[i\right]$ is the thickness of each layer, $Porosisty\left[i\right]$ is the porosity of each layer, $moist\left[i\right]$ is the moisture content of each layer, DLD is the soil depth of the bottommost layer, and DP is the porosity of the bottommost layer.)

2.2.3. Groundwater Movement

Groundwater movement was calculated for hydrologic-cycle units in the plain area, where groundwater abstraction, shallow groundwater evaporation, and groundwater–surface-water exchange are important components of the water balance. The boundary conditions for groundwater calculations are divided into three main types:

(1)

Type I boundary (given head boundary)

$$H\left|B_1\right.=H_1\left(x,y,t\right)\hspace{1em}\hspace{1em}x,y\in B_1$$

(1)

where $H_1$ is a known head function on $B_1$. This type of boundary is usually expressed as the dividing line (surface) between the surface water body and the seepage area.

(2)

Type II boundary (given flow boundary)

$$T{\displaystyle \frac{\partial H}{\partial n}}\left|B_2\right.=q\left(x,y,t\right)\hspace{1em}\hspace{1em}x,y\in B_2$$

(2)

where H and n are the head and the outer normal direction of the boundary respectively; $-{\displaystyle \frac{\partial H}{\partial n}}$ is the component of the hydraulic gradient in the boundary direction; and q is the single-width flow.

(3)

Third type of boundary (mixed boundary conditions)

Mixed boundary conditions are usually used to solve the equations when some of the conditions at the boundary of the study area are known for head variation cases or flow cases.

2.2.4. Snow Accumulation and Melting Process on the Surface and Water Balance

The snow accumulation and melting module couples mass-balance and energy-balance calculations using a two-layer snow representation. The mass-balance component describes changes in snow cover, snow water equivalent, snowmelt, and snowmelt-generated runoff, while the energy-balance component describes ice formation, melting, and thermal storage changes within the snowpack. Although snowfall and snowmelt are not dominant hydrologic processes in most parts of the Shaying River Basin, this module was retained because it is an integrated component of SWAM and can account for occasional winter snow or freezing processes. The contribution of snow processes to the basin-scale attribution results is expected to be limited compared with precipitation, evapotranspiration, irrigation, and groundwater abstraction. The two-layer snow accumulation and melting model is shown in Figure 3.

2.2.5. Runoff Generation

In the construction of the runoff module, the Curve Number (CN)–Topographic Index method is adopted to define the initial spatial distribution of the excess storage runoff generation zones in the watershed. During the runoff generation calculation, the infiltration-excess runoff generation grids and saturation-excess runoff generation grids are dynamically identified based on the relationship between rainfall and soil moisture content.

The following assumptions are followed in the division of excess storage grids: ① Given the flat terrain and relatively high soil moisture content near river channels, all river channel grids are defined as saturation-excess grids, while non-river channel grids can be set as infiltration-excess or saturation-excess grids according to the corresponding discrimination criteria. ② In the calculation of non-river channel grids, infiltration-excess and saturation-excess grids are dynamically identified by judging whether rainfall exceeds the infiltration capacity and whether soil moisture content reaches the field capacity. The core principles are as follows: in the runoff generation simulation, a grid is identified as an infiltration-excess grid when the rainfall intensity exceeds the soil infiltration capacity; a grid is converted to a saturation-excess grid once its soil moisture content reaches the field capacity.

2.2.6. Runoff Concentration

Runoff concentration in the plain area was calculated according to the principle of nearest inflow. When sufficient drainage-network information was available, the nodal drainage method was used. When detailed drainage information was unavailable, the side-inflow method was adopted. This treatment allows the model to represent confluence processes in both data-rich and data-limited plain-area units. The generalized confluence process for plain-area hydrologic units is shown in Figure 4.

2.2.7. Simulation of Hydraulic Engineering and Water-Transfer Processes

The fundamental approach for simulating hydraulic engineering involves generalizing these structures as computational watershed nodes within the river network. Subsequently, the water balance equation is solved for each individual node to maintain mass conservation; the conceptual schematic for this process is illustrated in Figure 5.

In alignment with the operational characteristics of the basin, the model treats small- and medium-sized reservoirs with limited data as simplified generalized nodes for flow calculation. In contrast, large-scale reservoirs with comprehensive datasets are subject to high-fidelity, real-time simulation. Regarding water diversion projects, these are categorized into intra-basin and inter-basin transfers. For intra-basin transfers, the model designates distinct source and destination nodes within the river network. For inter-basin transfers, a source or destination node is integrated into the river network as appropriate to facilitate the simulation of river reach routing and evolution.

2.3. Data Sources and Processing

2.3.1. Data Preparation

Meteorological data for the Shaying River Basin were sourced from the Daily Value Dataset of Chinese Terrestrial Climate Data (V3.0), encompassing precipitation; maximum, minimum, and mean temperatures; average wind speed; relative humidity; and sunshine duration. Ten representative meteorological stations were identified within the basin using Thiessen polygons; subsequently, the Kriging interpolation method was employed to distribute these data across the individual hydrological units. Based on these datasets, the analysis utilized continuous time series from two decades (1971–1980 and 2001–2010), specifically incorporating net radiation, daily precipitation, thermal extremes, sunshine duration, wind speed, temperature diurnal range, and relative humidity. Data from each station were further processed using a weighted averaging technique to represent the basin-wide conditions.

To improve data transparency and reproducibility, the main meteorological, hydrological, and groundwater observation data used in this study are summarized in

Table 1. Summary of the main data sources used for model forcing, calibration, validation, and scenario analysis.

Data Category	Station/Data Source	Period	Variables and Use in This Study
Meteorological data	Ten representative stations selected using Thiessen polygons	1971–1980; 2001–2010	Precipitation, maximum/minimum/mean temperature, wind speed, relative humidity, and sunshine duration; used for meteorological forcing, Kriging interpolation, and regional weighted averaging
Runoff data	Zhoukou, Baiguishan, and Zhaoping hydrological stations	1971–1980; 2001–2010	Monthly runoff; used for model calibration and validation
Groundwater data	Selected observation wells in the plain area of the Shaying River Basin	1971–1980; 2001–2010	Groundwater dynamics and evaporation-related groundwater parameters; used for groundwater-resource validation and hydrogeological parameter estimation
Water-use data	Huaihe River Basin water-resources bulletins, planning documents, and administrative-region statistics	1971–1980; 2001–2010	Urban, rural, irrigation, industrial, and domestic water use; used for the water-resource development scenario and water-use correction

.

Land-use data were obtained from the Resource and Environment Science Data Center of the Chinese Academy of Sciences. For model application, land-use types were reclassified into six categories: forest land, grassland, water bodies, urban construction land, unused land, and arable land. The 2005 land-use distribution used in this study is shown in Figure 6. The 1980 land-use dataset was used to represent the baseline land-surface condition, whereas the 2005 land-use dataset was used to represent the changed-environment period because it is close to the midpoint of 2001–2010 and provides spatially complete land-cover information for this period. This treatment may smooth intra-decadal land-use changes, and this uncertainty is discussed in Section 5. Soil-type data were also used to support the parameterization of infiltration, soil-water redistribution, and runoff-generation processes in the SWAM model. The distribution of major soil types in the Shaying River Basin is shown in Figure 7.

Socioeconomic water-use data were compiled from the Water Resources Bulletin of the Huaihe River Basin, the Comprehensive Water Resources Plan of the Huaihe River Basin, and water-resource bulletins of administrative regions within the Shaying River Basin. For administrative units only partially located within the basin, urban and rural water-use data were adjusted using correction factors based on the proportion of residential or cultivated land area within the basin. These data were used to represent irrigation, industrial, and domestic water withdrawals in the water-resource development scenario.

The water-supply degree in the hydrogeological data of the plain area was calculated based on the groundwater dynamics and evaporation data of selected observation wells in the Shaying River Basin in Henan Province; the water-supply degree of four lithologies (chalky sand, sub-sand, sub-clay, sub-sand and sub-clay interlayer) was calculated by the graphical method according to the Averyanov formula $E={E_0(1-H/H_0)}^n$; and the water-supply degree of some water sources was calculated by the cylinder measurement method, the pumping test method, the use pumping test, indoor instrument determination, the field concentric ring or test pit injection test, and other methods of permeability to determine the coefficient K value.

The hydrogeological data of the plain area, in terms of water-feeding degree and permeability coefficient, are important hydrogeological parameters:

(1)

Water-feeding degree

According to the groundwater dynamics and evaporation data of selected observation wells in Henan Province of the Shaying River Basin and according to the Averyanov formula, $E={E_0(1-H/H_0)}^n$, a relationship diagram of $\Delta H/E_0~H$ was made using the graphical method, and the water-feeding degrees of four lithologies (chalky sand, sub-sand, sub-clay, sub-sand and sub-clay interlayer) were calculated and combined with the water-feeding degrees determined by the cylinder measurement and pumping test method at some water sources. The groundwater-feeding degrees in the plain area of the Shaying River Basin are shown in Figure 8.

(2)

Permeability coefficient

To determine the permeability coefficient (K), various methods, such as pumping tests, laboratory instrumentation, and field-based concentric ring or test pit injection tests, are typically employed. By referencing the experimental results from specific water sources within the Shaying River Basin and integrating the empirical K values associated with diverse lithological characteristics, the determined permeability coefficients were established, as presented in

Table 2. Permeability coefficient.

Lithology	Hydraulic Conductivity K (m·d−1)
Clay	<0.1
Sub-clay	0.1~0.25
Sub-sand clay	0.25~0.50
Fine sand	1.0~8.0
coarse sand	5.0~10.0
Medium-fine sand	8~15
Medium-coarse sand	15~25
Gravelly medium-fine sand	30
Sand gravel	50~100
Sandy pebble gravel	100~200

.

2.3.2. Hydrologic-Cycle Unit Construction

The spatial discretization of the Shaying River Basin was executed through a multi-stage process to define the hydrologic-cycle calculation units. The initial phase involved delineating mountainous units by intersecting sub-basin boundaries with administrative divisions. Subsequently, plain areas were discretized using a standardized rectangular grid. Each grid cell was systematically coded with irrigation area, sub-basin, and administrative attributes to ensure uniform data processing.

(1)

Topographic Boundary Delineation

The fundamental prerequisite for unit processing is the delineation of the mountain–plain interface. This boundary was established using a slope threshold analysis applied to adjacent terrain raster datasets.

(2)

Mountainous Unit Delineation

By applying slope dispersion analysis, the mountainous and plain regions were partitioned into discrete spatial maps. The mountainous units were then finalized by overlaying administrative boundaries on the identified terrain, as illustrated in Figure 9.

(3)

Discretization of Plain Units

The plain areas were divided into rectangular cells with a spatial resolution of 1 km × 1 km. To facilitate finite-difference groundwater calculations, the grid encompasses the entire plain area, with zero-value boundary cells appended to the periphery for computational stability, as shown in Figure 10.

(4)

Junction Unit Calibration

For cells situated at the junction of mountainous and plain regions, ArcGIS version 10.8 (Esri, Redlands, CA, USA) was employed to calculate the actual area of the hydrological plain component. This value replaced the original geometric cell area, resulting in topological units that maintain strict attribute and relational integrity while accounting for irregular boundary areas.

(5)

Integration of Irrigation and Administrative Districts

Following the spatial distribution of the irrigation districts (Figure 11), these zones were superimposed onto the existing grid. A final overlay of administrative districts was conducted to generate the comprehensive hydrologic-cycle calculation units. This process yielded a total of 37,389 computational units, as depicted in Figure 12.

2.3.3. Scenario Design and Attribution Calculation

To separate the effects of climate change and human activities on hydrologic-cycle changes, four scenarios were designed. Scenario S1 represents the baseline condition, using meteorological data from 1971 to 1980, land-use data from 1980, and water-resource development conditions from 1971 to 1980. Scenario S2 represents the climate-change scenario, in which meteorological data were replaced by those from 2001–2010, while land-use and water-resource development conditions were kept the same as in S1. Scenario S3 represents the land-use-change scenario, in which land-use data were replaced by the 2005 land-use dataset while meteorological and water-use conditions were kept the same as in S1. Scenario S4 represents the water-resource development scenario, in which water-use conditions were replaced by those from 2001–2010, while meteorological and land-use conditions were kept the same as in S1.

For a hydrologic variable, X, the climate-induced change was calculated as $\Delta XC=X_{S2}-X_{S1}$. The land-use-induced change was calculated as $\Delta XL=X_{S3}-X_{S1}$, and the water-resource-development-induced change was calculated as $\Delta XW=X_{S4}-X_{S1}$. The total human-activity-induced change was estimated as $\Delta XH=\Delta XL+\Delta XW$. The relative contributions of climate change and human activities were then calculated as:

$$CC ={\displaystyle \frac{\left|\Delta XC\right|}{\left(\right|\Delta XC|+ |\Delta XH\left|\right)}}\times 100\%$$

(3)

$$CH={\displaystyle \frac{\left|\Delta XH\right|}{\left(\right|\Delta XC|+|\Delta XH\left|\right)}}\times 100\%$$

(4)

where CC and CH represent the relative contributions of climate change and human activities, respectively. This scenario-based attribution approach assumes first-order separability among climate change, land-use change, and water-resource development. Potential nonlinear interactions among these drivers are not fully resolved and are discussed as a limitation in Section 5.

2.4. Model Calibration and Validation

2.4.1. Calibration and Evaluation Metrics

The model was calibrated for the baseline period of 1971–1980 and validated for the changed-environment period of 2001–2010. Monthly runoff at key hydrological stations, annual groundwater resources, and annual total water resources were used to evaluate model performance. The correlation coefficient (R) and Nash–Sutcliffe efficiency coefficient (Ens) were used as performance metrics. After calibration, the model parameters were kept unchanged during the validation period to test the transferability of the parameter set under changed environmental conditions.

The formula for the correlation coefficient, $R^2$, is given in (3):

$$R^2={\displaystyle \frac{\left({\sum }_{i=1}^n(Q_{sim,i}-\overline{Q_{sim}}\right)\left(Q_{obs,i}-\overline{Q_{obs}}\right){)}^2}{{\sum }_{i=1}^n(Q_{sim,i}-\overline{Q_{sim}}{)}^2{\sum }_{i=1}^n(Q_{obs,i}-\overline{Q_{obs}}{)}^2}}$$

(5)

The formula for calculating the Nash coefficient, $E_{ns}$, is shown in (4):

$$E_{ns}=1-{\displaystyle \frac{{\sum }_{i=1}^n(Q_{obs,i}-Q_{sim,i}{)}^2}{{\sum }_{i=1}^n(Q_{obs,i}-\overline{Q_{obs}}{)}^2}}$$

(6)

Here, $Q_{obs,i}$ denotes the value of the measured series; $Q_{sim,i}$ denotes the value of the simulated series; $\overline{Q_{sim}}$ denotes the average value of the simulated series; and $Q_{obs}$ denotes the average value of the measured series.

Because this study focuses on decadal scenario-based attribution rather than parameter-sensitivity diagnosis or model intercomparison, the model evaluation was designed to test whether SWAM could reproduce the key runoff, groundwater, and total-water-resource characteristics required for the attribution analysis. A separate parameter-sensitivity analysis was not included in the present study because many model parameters were derived from local hydrogeological surveys, soil and land-use datasets, hydraulic-engineering information, and previous SWAM applications. Nevertheless, parameter uncertainty may influence the quantitative attribution results, and systematic sensitivity analysis and uncertainty propagation should be incorporated in future studies when more detailed observations and computational resources are available.

2.4.2. Calibration and Validation Results

Monthly Runoff Validation

Zhoukou, Baiguishan, and Zhaoping stations were selected as key hydrological stations for runoff validation. The simulated and observed monthly runoff series showed good agreement during both the calibration and validation periods. The correlation coefficients were greater than 0.89, and the Ens values were greater than 0.77, indicating that the model can reasonably reproduce monthly runoff dynamics at the selected stations. The runoff calibration and validation statistics are summarized in

Table 3. Runoff rate determination and validation for each site.

Station:		Zhoukou Station	Baigushan Station	Zhaopeng Station
Correlation coefficient	Rate Period	0.8989	0.8976	0.8977
	Validation Period	0.8987	0.8972	0.8974
NSE	Rate Period	0.7782	0.7953	0.7959
	Validation period	0.7973	0.7949	0.7951

.

Groundwater Validation

Groundwater resources were validated using available groundwater-resource assessments for 1971–1980 and 2001–2010. The correlation coefficients between simulated and assessed annual groundwater resources were 0.8895 and 0.8971 for the two periods, respectively. These results suggest that the model can capture the main interannual variation in groundwater resources at the basin scale. However, localized groundwater-level changes and seasonal overdraft may still contain uncertainty because of the limited spatial resolution of groundwater observations.

Water Resources Validation

Observed water resources data were synthesized from relevant technical reports and the published literature. Statistical analysis indicates that the correlation coefficient (R) between the measured and simulated total annual water resources for the 1971–1980 period was 0.8988. For the 2001–2010 period, the correlation coefficient was 0.8723. These results reflect a high degree of consistency between the simulated outputs and the empirical data, verifying the model’s robustness in capturing the long-term variability of the basin’s total water resources.

3. Spatiotemporal Characteristics of Hydrologic-Cycle Components in the Shaying River Basin

3.1. Basin-Scale Hydrologic-Cycle Fluxes During the Baseline Period

Based on the datasets for the Shaying River Basin during the period 1971–1980, the SWAM model was used to simulate the hydrologic-cycle evolution. The simulated hydrological fluxes were quantitatively derived and are presented in

Table 4. Hydrologic-cycle fluxes in the Shaying River Basin during the historical period (unit: billion m³).

Water Circulation System	Input Items		Output Items		Storage Change
Surface Water	Precipitation from rivers, lakes and reservoirs	4.9	River, lake and reservoir evaporation	4.8	Surface water storage change
	Runoff to rivers	76.2	River, lake and reservoir seepage	10.3
	Industrial and domestic groundwater withdrawal	1.8	Irrigation surface-water withdrawal	6.7
	Water transfer from outside the district	0.5	Industrial domestic water consumption	1.9
		9.2	Outbound flow	67.9
	Total	92.6	Total	91.6		1.0
Soil water	Agricultural precipitation	277.8	Evapotranspiration from farmland	223.7	Soil storage change
	Non-agricultural precipitation	65.4	Non-farm evapotranspiration	44.2
	Surface irrigation water	6.7	Evapotranspiration	76.2
	Underground irrigation water	2.4	Soil seepage recharge to the subsurface	24.7
	Phreatic evaporation	18
	Total	370.2	Total	368.8		1.4
Groundwater	Seepage from rivers, lakes and reservoirs	10.3	Irrigation shallow subsurface withdrawal	2.34	Groundwater storage
	Soil seepage recharge to underground	25	Industrial domestic shallow subsurface abstraction	1.71
			Phreatic evaporation	18
			Groundwater cross-flow recharge to deeper layers	2.6
				9.2
	Total	35.0	Total	33.9		1.1
Basin-wide	Precipitation	348.1	Evaporation	272.8	Surface-water storage change	1.0
	Deep groundwater extraction	0.1	Industrial domestic water consumption	1.9	Soil-water storage change	1.4
	Water transfer from outside the area	0.5	Surface runoff outflow	67.9	Groundwater storage change	1.1
			Groundwater overflow recharge to deep layer	2.6
	Total	348.7	Total	345.2	Surface-water storage change	3.5

.

During the baseline period, precipitation was the dominant input to the basin-scale hydrologic cycle, whereas evapotranspiration and outbound runoff were the major output components. The basin showed a positive total storage change of approximately 0.35 billion m³, indicating a net water surplus at the decadal mean scale. Surface water, soil water, and groundwater all showed positive storage changes in the basin-scale water balance. It should be noted that storage change (ΔS) can be either positive or negative; positive values indicate storage accumulation, whereas negative values indicate storage depletion.

The proportions of hydrologic-cycle flux inputs and outputs directly influenced by anthropogenic factors are illustrated in Figure 13. It can be observed that surface-water storage changes are significantly modulated by human activities, primarily driven by water withdrawals for irrigation, industrial processes, and domestic consumption. Conversely, the direct anthropogenic impact on groundwater storage changes is predominantly manifested through groundwater extraction. Overall, the output fluxes of each hydrological layer exhibit greater sensitivity to human activities compared to the corresponding input fluxes.

3.2. Spatial Characteristics of Major Hydrologic-Cycle Components

3.2.1. Evapotranspiration

During the baseline period, the mean annual precipitation in the Shaying River Basin was 832 mm, with values ranging from 618 to 1050 mm. The mean annual evapotranspiration was 665 mm, ranging from 543 to 838 mm. Spatially, evapotranspiration was lower in the mountainous area than in the plain area. This difference is mainly related to topographic effects, lower temperature, and higher vegetation coverage in mountainous areas. In contrast, the plain area had higher evapotranspiration because of warmer conditions, extensive farmland, irrigation, and stronger human water use. In the northern plain, evapotranspiration was relatively high despite lower natural precipitation, reflecting the influence of urbanization, artificial water supply, and water-resource regulation. The spatial distribution of total evapotranspiration during 1971–1980 is shown in Figure 14.

3.2.2. Surface Runoff Generation

Mean annual surface runoff depth ranged from 140 to 225 mm during the baseline period, showing clear spatial heterogeneity. Higher runoff depths occurred in the southern part of the basin and in urbanized northern plain areas. The southern basin received greater precipitation, whereas the northern plain was affected by impervious surfaces and artificial drainage conditions. In the northwestern mountainous area, forest and grassland coverage contributed to moderate runoff generation, whereas the central plain, dominated by arable land, showed relatively lower runoff depth because of infiltration and agricultural water retention. The spatial distribution of surface runoff depth is shown in Figure 15.

3.2.3. Soil Water

Soil-water storage in the Shaying River Basin fluctuates under the dual influence of natural precipitation and irrigation water withdrawals, generally exhibiting an increasing trend that aligns with interannual variations in precipitation. With the exception of 1971 and 1976, which recorded slight decreases, soil-water storage remained in surplus across all other years. In terms of intra-annual variation, soil-water storage is negative from November to May, turns positive from June to August, and reaches its maximum in July.

3.2.4. Groundwater

From 1971 to 1980, the average annual groundwater recharge in the Shaying River Basin was greater than the average annual groundwater abstraction at the basin scale. However, groundwater storage change could still be negative in some areas or seasons because groundwater abstraction was spatially concentrated in agricultural and urbanized plain areas, whereas recharge was unevenly distributed in space and time. In addition, shallow groundwater evaporation, delayed recharge response, and cross-flow to deeper aquifers may reduce effective groundwater storage. Therefore, the apparent difference between basin-scale annual recharge and localized groundwater-storage decline reflects the spatial and seasonal complexity of groundwater dynamics rather than a contradiction in the water balance.

4. Impacts of Climate Change and Human Activities on Hydrologic-Cycle Changes

Based on the scenario design described in Section 2.3.3, this section compares hydrologic-cycle responses under climate change, land-use change, and water-resource development conditions. The analysis focuses on differences between the baseline period and the changed-environment period, rather than on continuous long-term trend detection. Therefore, terms such as “increase” and “decrease” refer to differences between the designed scenarios or between the two decadal periods.

The data parameters used for the baseline scenario and the three comparative scenarios are summarized in

Table 5. Quantitative simulation scheme of hydrologic cycle in the Shaying River Basin.

Period in Which Data Are Available	Meteorological Conditions	Land Use	Water Resources Development and Utilization Conditions
Baseline Scenario (Historical Scenario)	1971–1980	1980	1971–1980
Contrast Scenario I	2001–2010	1980	1971–1980
Comparison Scenario II	1971–1980	2005	1971–1980
Comparison Scenario III	1971–1980	1980	2001–2010

Note: In this paper, after comprehensive consideration based on the available remote sensing data, the 1980 land-use data is proposed to be used for the unchanged period, and the 2005 land-use data is proposed to be used for the changed period.

. All model simulation conditions were kept consistent across scenarios, except for the specific variable changes indicated in

Table 6. Comparison analysis of different simulation scenarios with the baseline scenario.

Driver	Data series	Baseline Scenario	Contrast I	Contrast II	Contrast III
Climatic conditions Change	Meteorological data for the unchanged period	✓		✓	✓
	Meteorological data for the changed period		✓
Land-use change	Meteorological data for the unchanged period	✓	✓		✓
	Meteorological data for the changed period			✓
Water Resources Development and Use	Meteorological data for the unchanged period	✓	✓	✓
	Meteorological data for the changed period				✓

.

4.1. Changes in the Hydrologic Cycle Under Climate Change Conditions

4.1.1. Changes in Meteorological Elements Under Climate Change Conditions (Plain vs. Mountainous Areas)

The comparison of meteorological conditions between the periods 1971–1980 and 2001–2010 shows clear spatial differences between the plain and mountainous areas. Air temperature increased in both regions, indicating a warmer climatic condition during the changed-environment period. Precipitation increased in the plain area but decreased in the mountainous area, suggesting that climate-change effects were spatially heterogeneous within the basin. These contrasting precipitation changes are important for explaining the different hydrologic responses of the two physiographic zones. Net radiation did not show a uniform increase across the basin; therefore, changes in evapotranspiration should be interpreted as the combined result of temperature, precipitation, land-surface conditions, and water availability rather than radiation alone. The changes in net radiation, precipitation, and mean temperature are compared in Figure 16, Figure 17 and Figure 18, respectively.

4.1.2. Hydrologic-Cycle Change Patterns Under Climate Change Conditions (Plain Areas and Mountainous Areas)

Evapotranspiration Response Under Climate Change

Under the climate-change scenario, evapotranspiration changed differently between the plain and mountainous areas. In the plain area, increased precipitation and temperature enhanced water availability and evaporative demand, leading to higher evapotranspiration. In the mountainous area, reduced precipitation limited water availability despite rising temperature, resulting in a weaker or opposite evapotranspiration response. This contrast indicates that evapotranspiration in the basin is controlled not only by atmospheric demand but also by regional water availability.

Surface-Runoff Response Under Climate Change

Surface runoff responded to changes in both precipitation and evapotranspiration. In the plain area, increased precipitation did not necessarily translate into proportional runoff increase because evapotranspiration, irrigation, infiltration, and hydraulic regulation modified runoff generation and routing. In the mountainous area, reduced precipitation contributed to lower runoff generation. These results indicate that climate-driven runoff changes are spatially dependent and strongly conditioned by local terrain and land-surface properties.

Soil-Water Response Under Climate Change

Soil-water storage dynamics in the Shaying River Basin fluctuate under the combined influence of natural precipitation and agricultural irrigation withdrawals. On an interannual basis, storage changes exhibit clear sensitivity to precipitation variability. In 2008, storage remained essentially balanced, with only a marginal overall decline. In contrast, 2002 recorded an increase in soil-water storage, while a decreasing trend was observed in all other years during the study period.

Regarding the intra-annual distribution of soil-water storage across the basin, values remained negative from September through June of the following year. As rainfall-driven recharge intensified, storage transitioned to positive values during July and August, peaking in August. Notably, these peak values remained lower than the storage changes recorded for the corresponding months during the historical baseline period.

Groundwater Soil-Water Response Under Climate Change

During the period from 1971 to 1980, the average annual groundwater resources in the Shaying River Basin amounted to 48.3 million m³, whereas from 2001 to 2010, this value decreased to 46.0 million m³. This decline reflects a discernible overall downward trend in groundwater availability across the two study periods.

4.2. Hydrologic-Cycle Changes in Soil and Water Resources Development and Utilization Conditions

4.2.1. Analysis of the Development and Utilization of Soil and Water Resources (Plain Areas and Mountainous Areas)

Analysis of Land-Use Change

By comparing the land-use maps from 1980 and 2005 for both mountainous and plain areas, the spatial variations in land use between the two periods were identified. The analysis reveals that the Shaying River Basin is predominantly characterized by farmland. Specifically, the plain areas consist mainly of cultivated land and residential land, while the mountainous areas are composed primarily of cultivated land, forest land, and grassland. A comparison of the two periods indicates that, across the entire basin, the areas of residential land and forest land increased, whereas the areas of farmland, grassland, and water decreased. Furthermore, the magnitude of these changes differed between mountainous and plain areas, with residential land in the plain area exhibiting the most significant increase and farmland in the plain area showing the most pronounced decrease.

Analysis of Changes in Water Resources Development and Utilization

A comparison of water resources development and utilization between the periods 1971–1980 and 2001–2010 reveals that groundwater withdrawal in the plain area increased rapidly, significantly outpacing the growth rate observed in the mountainous area. Specifically, groundwater extraction for irrigation increased tenfold, while that for industrial and domestic use rose more than sevenfold.

Water-resource development intensified markedly during the changed-environment period. Groundwater withdrawal increased more rapidly in the plain area than in the mountainous area, especially for irrigation and industrial–domestic water use. This increase was associated with population growth, agricultural irrigation demand, urbanization, industrial development, and the expansion of hydraulic engineering regulation. These factors increased artificial water inputs and outputs and strengthened the influence of human activities on surface-water, soil-water, and groundwater storage.

4.2.2. Hydrologic-Cycle Responses to Land-Use and Water-Use Scenarios

Land-Use Scenarios

Under the land-use-change scenario, evapotranspiration changed only slightly relative to the baseline condition. Surface runoff decreased slightly, while groundwater resources increased. These changes suggest that land-use change alone had a relatively moderate effect on basin-scale hydrologic-cycle components compared with water-resource development. However, the effect of land-use change differed spatially because changes in farmland, construction land, forest land, and grassland modified infiltration, evapotranspiration, and runoff generation differently in the plain and mountainous areas.

Water Utilization Scenarios

Evapotranspiration exhibited an upward trend, with the multi-year average under water resources development and utilization conditions reaching 643 mm—a 1.2% increase relative to the historical period. Concurrently, irrigation water withdrawal increased substantially, and surface runoff depth also rose; under the development scenario, the multi-year average runoff depth was 184.2 mm, representing a 0.6% increase over the historical baseline.

Although the volume of irrigation water applied to cropland increased considerably, soil-water storage showed only a slight upward trend. Under the development conditions, the multi-year average soil-water storage was recorded at 310 million m³, which is 4.2% lower than that of the historical period. Furthermore, increased groundwater extraction for irrigation and industrial use led to a reduction in groundwater recharge, resulting in an overall declining trend in groundwater resources. Under the water resources development and utilization scenario, groundwater resources decreased to 3.00 billion m³—a substantial decline of 16.3% compared to the historical period.

4.3. Attribution Analysis of Hydrologic-Cycle Evolution to Changing Environments

The relative contributions of climate change and human activities were calculated using the scenario-based first-order attribution method described in Section 2.3.3. Human activities include both land-use change and water-resource development. Because the method is based on prescribed scenario differences, the calculated percentages should be interpreted as diagnostic contribution estimates rather than as complete representations of all nonlinear climate–land–water interactions.

4.3.1. Plain Area Attribution Analysis

The results of the hydrologic-cycle attribution analysis in the Shaying River plain area under a changing environment are shown in

Table 7. Results of the attribution analysis of the evolution of the hydrologic cycle in the plains.

	Analysis Items	Scenario				Climate Change	Human Activities
		Benchmark S1	Climate S2	Land S3	Water S4	S2 − S1	S3 − S1 + S4 − S1
Changing Environment	Precipitation (mm)	845	967	845	845	14.4%	-
	Air temperature (°C)	15.2	15.3	15.2	15.2	0.7%	-
	Lower bedding surface (%)	Farmland: 86 Residential site: 11	Same as S1	Farmland: 80	Same as S1	-	Agricultural land decreased by 6% and Residential land increased by 4%
	Artificial water (mm)	35	35	35	164	-	369%
Cyclic response results	Evapotranspiration (mm)	690	712	677	715	22	12
	Surface runoff depth (mm)	208	207	206	204	−2	−7
	Outbound water volume (mm)	229	227	237	211	−2	−9
	Storage change (mm)	11	3	9	−4	−8	−17
Percentage		-	-	-	-	33%	67%

.

The relative contributions of climate change and human activities to hydrologic-cycle processes in the plain area are 33% and 67%, respectively, indicating that human activities play a dominant role in shaping hydrologic-cycle variations within this region. Among the four response factors examined, the impact of human activities on evapotranspiration is relatively limited, accounting for only 35%. In contrast, human activities exert a more substantial influence on outbound water (86%), followed by surface runoff depth (79%) and storage changes (68%).

4.3.2. Mountain Attribution Analysis

Table 8. Calculation of the attribution analysis for the evolution of the hydrologic cycle in the mountains.

	Analysis Items	Scenario				Climate Change	Human Activities
		Benchmark S1	Climate S2	Land S3	Water S4	S2 − S1	S3 − S1 + S4 − S1
Changing Environments	Precipitation (mm)	792	758	792	792	−4.3%	-
	Air temperature (°C)	14.1	14.7	14.1	14.1	4.3%	-
	Lower bedding surface (%)	Agricultural land: 60 Residential site: 3%	Same as S1	Agricultural land: 52 Residential site: 5%	Same as S1	-	Agricultural land decreased by 6.4% and residential sites increased by 4%.
	Artificial water (mm)	24	24	24	79	-	229%
Cyclic response results	Evapotranspiration (mm)	679	711	668	695	31	4
	Surface runoff depth (mm)	184	177	191	173	−6	−4
	Outbound water volume (mm)	183	163	179	159	−20	−29
	Storage change (mm)	4	2	16	−9	−3	−2
Percentage		-	-	-	-	61%	39%

reveals the relative contributions of climate change and human activities to the hydrologic cycle in the mountainous areas of the Shaying River Basin under changing environmental conditions.

The relative contributions of climate change and human activities to hydrologic-cycle processes in the mountainous areas are 61% and 39%, respectively, indicating that climate change exerts a greater influence on hydrologic-cycle elements in this region than human activities. Among the four response factors examined, climate change has the most pronounced impact on evapotranspiration, accounting for 88%. It also exerts a substantial influence on surface runoff depth (59%), followed by storage changes (56%) and outbound water volume (42%).

4.3.3. Comprehensive Attribution Analysis

As illustrated in Figure 19, the evolution of the hydrologic cycle in the plain areas is primarily driven by human activities, whereas in the mountainous areas, it is predominantly influenced by climate change. The intensity of human impact varies considerably between these regions: compared to mountainous areas, the influence of human activities on surface runoff and outbound water in the plains is stronger by 38% and 27%, respectively. Furthermore, the degree of human influence on storage change and evapotranspiration in the plain areas exceeds that in the mountainous areas by 24% and 23%, respectively.

5. Discussion

5.1. Interpretation of Regional Differences in Hydrologic-Cycle Attribution

The attribution results reveal distinct differences between the plain and mountainous areas of the Shaying River Basin. In the plain area, human activities were the dominant driver of hydrologic-cycle changes. This is mainly because the plain area has intensive agricultural irrigation, dense population, urban expansion, groundwater abstraction, and extensive hydraulic regulation. These activities directly modify surface-water withdrawals, groundwater storage, soil-water conditions, and outbound flow. In contrast, the mountainous area is less affected by direct water withdrawals and urban development, and its hydrologic response is more strongly controlled by precipitation, temperature, terrain, vegetation, and natural runoff-generation processes. Therefore, climate change showed a higher relative contribution in the mountainous area.

The basin-scale result, in which human activities contributed more than climate change, is consistent with the strong anthropogenic modification of the Shaying River Basin. However, the spatial contrast between plains and mountains indicates that basin-mean attribution may obscure important regional differences. This finding supports the need to analyze hydrologic-cycle changes separately for physiographic zones, especially in transitional basins where climate gradients and human water use coexist.

Compared with classical basin-scale attribution approaches, the SWAM-based scenario framework provides a way to examine multiple hydrologic-cycle components, including evapotranspiration, surface runoff, soil-water storage, groundwater storage, artificial water withdrawals, and hydraulic engineering effects. This is particularly useful for the Shaying River Basin, where natural and artificial water-cycle processes are closely coupled. Nevertheless, the contribution rates should be interpreted as scenario-based diagnostic estimates rather than universal constants for all transitional basins.

5.2. Methodological Assumptions and Uncertainties

Several methodological assumptions and uncertainties should be acknowledged. First, the scenario-based attribution method assumes first-order separability between climate change, land-use change, and water-resource development. In reality, these drivers may interact nonlinearly. For example, land-use change can modify evapotranspiration and local climate feedbacks, while water-resource regulation can alter soil moisture and groundwater recharge. Therefore, the attribution percentages reported in this study should be interpreted as first-order diagnostic estimates under prescribed scenarios.

Second, a full parameter-sensitivity analysis and model intercomparison were not conducted in this study. This is because the objective of the present work is to diagnose decadal hydrologic-cycle attribution using an established SWAM framework rather than to compare alternative models or identify parameter sensitivity rankings. Many model parameters were derived from local hydrogeological surveys, soil and land-use datasets, hydraulic-engineering information, and previous SWAM applications. Model credibility was evaluated using monthly runoff, groundwater resources, and total water resources during calibration and validation periods. Nevertheless, parameter sensitivity and uncertainty propagation may affect the quantitative contribution estimates, and future studies should incorporate systematic sensitivity analysis, multi-model comparison, and ensemble simulation when sufficient data and computational resources are available.

Third, the comparison between the periods 1971–1980 and 2001–2010 provides a decadal-scale contrast between baseline and changed-environment conditions, but it does not fully describe continuous long-term trends or abrupt change points. The 2005 land-use dataset was used to represent the changed-environment period because it is close to the midpoint of 2001–2010 and provides complete spatial coverage. However, this approach may smooth intra-decadal land-use changes. Future work should use higher-temporal-resolution land-use datasets to examine continuous land-cover trajectories.

Fourth, the present analysis focuses on historical attribution up to 2010. Since 2010, continued urbanization, changes in agricultural water use, ecological restoration, hydraulic-engineering operation, and more frequent extreme hydro-climatic events may have further altered hydrologic-cycle processes in the basin. Extending the SWAM simulations to 2010–2024 would help evaluate whether the contribution of human activities has continued to increase in the plain area and whether climate-driven extremes have become more important in the mountainous area.

Finally, the separation of land-surface factors and climatic factors in the scenario framework is a simplification. Land-surface conditions and climate are not fully independent in coupled hydro-climatic systems. For example, vegetation, soil moisture, and interception storage can affect evapotranspiration and local water availability. Therefore, equations involving interception, storage, and evaporation should be interpreted within the water-balance assumptions of SWAM. Future studies should further examine coupled land–atmosphere feedbacks and interception processes using more detailed field observations and physically based parameterization.

5.3. Management Implications

The attribution results have important implications for differentiated water-resource management. In the plain area, where human activities dominate hydrologic-cycle changes, management should focus on controlling groundwater abstraction, improving irrigation efficiency, optimizing the joint use of surface water and groundwater, and strengthening regulation of sluices, reservoirs, and water-transfer projects. Because water-resource development caused marked storage depletion, especially in groundwater, future policies should prioritize groundwater conservation, water-saving irrigation, industrial water-use efficiency, and stricter control of excessive withdrawals.

In the mountainous area, where climate change plays a larger role, management should emphasize ecological conservation, soil and water retention, vegetation protection, and monitoring of climate-driven runoff changes. Maintaining forest and grassland cover can help regulate runoff generation, reduce soil erosion, and improve water retention. Because mountainous runoff contributes to downstream plain hydrology, improved monitoring of precipitation, runoff, and extreme events in the mountainous area is important for flood control and basin-wide water allocation.

At the basin scale, the contrasting drivers in mountainous and plain areas suggest that uniform water-management strategies may be ineffective. Instead, adaptive management should distinguish between climate-dominated upstream or mountainous zones and human-activity-dominated downstream or plain zones. Such differentiated strategies can better support flood control, drought mitigation, groundwater protection, and sustainable water-resource development in the Shaying River Basin.

6. Conclusions

This study applied SWAM to investigate hydrologic-cycle changes in the Shaying River Basin under baseline and changed-environment conditions. By designing comparative scenarios for climate change, land-use change, and water-resource development, the study quantified the relative contributions of climate change and human activities to hydrologic-cycle changes in the whole basin, plain area, and mountainous area. The main conclusions are as follows.

(1) During the baseline period of 1971–1980, precipitation was the dominant input to the basin hydrologic cycle, whereas evapotranspiration and outbound runoff were the major output components. Surface-water, soil-water, and groundwater storage showed positive storage changes at the basin scale, indicating a net water surplus during the baseline period.

(2) Hydrologic-cycle responses differed markedly between the plain and mountainous areas. The plain area was more strongly affected by irrigation, groundwater abstraction, urban expansion, and hydraulic regulation, whereas the mountainous area was more strongly controlled by precipitation, temperature, terrain, and vegetation conditions.

(3) At the basin scale, human activities contributed 59% and climate change contributed 41% to hydrologic-cycle changes. In the plain area, human activities were the dominant driver, contributing 67%, whereas in the mountainous area, climate change was dominant, contributing 61%. These results indicate that physiographic differences should be considered when attributing hydrologic-cycle changes in transitional basins.

(4) Water-resource development reduced surface-water, soil-water, and groundwater storage, especially in the plain area. Future water management should therefore focus on controlling groundwater abstraction, improving irrigation efficiency, optimizing hydraulic regulation, and implementing differentiated management strategies for mountainous and plain areas.

The contribution rates reported in this study should be interpreted as scenario-based first-order attribution estimates for the Shaying River Basin. Future work should extend the analysis to more recent periods, incorporate higher-resolution land-use data, and conduct systematic sensitivity and uncertainty analyses.