Preamble

Arid Zone Research Vol. 42 No. 6 Jun. 2025

Applicability Analysis of Hydrological Models for Different Watershed Types in the Eastern Section of the Agro-Pastoral Transitional Zone in Northern China

ZHANG Yajing, HAO Ruifang
(School of Soil and Water Conservation, Beijing Forestry University, Beijing 100083, China)

Abstract: Water resources constitute the foundation for maintaining ecosystem balance and sustaining human livelihoods and economic development. Simulating hydrological processes in arid and semi-arid ecosystems promotes the effective utilization of local water resources. This study analyzed the applicability of two models—the Distributed Hydrology Soil Vegetation Model (DHSVM) and the Soil and Water Assessment Tool (SWAT)—in different watershed types within semi-arid regions. Sensitivity analysis and parameter calibration were performed for both models, which were then used to simulate monthly runoff during the growing seasons (May–September) of 2015–2019 in the upper reaches of the Xar Moron River and Laoha River in the eastern agro-pastoral transitional zone of northern China. The upper Xar Moron River basin is dominated by grassland, whereas the upper Laoha River basin is dominated by forestland and farmland. The results demonstrate that DHSVM exhibits seven primary sensitive parameters in the Xar Moron River basin and six in the Laoha River basin, while SWAT identifies eleven and twelve sensitive parameters, respectively. Following calibration of these sensitive parameters, DHSVM achieved Nash-Sutcliffe efficiency (NSE) coefficients of 0.70 during the calibration period and 0.11 during the validation period in the Xar Moron River basin, compared to 0.43 and 0.04 for SWAT. In the Laoha River basin, DHSVM attained NSE coefficients of 0.56 and 0.70, while SWAT achieved 0.86 and 0.54, respectively. Both models demonstrate satisfactory applicability for hydrological process simulation in the study area, with DHSVM providing more accurate simulation of total runoff volume and SWAT offering more precise simulation of peak monthly runoff.

Keywords: DHSVM model; SWAT model; different watershed types; applicability analysis; runoff simulation

Introduction

Water scarcity poses a severe threat to human life, socio-economic sustainable development, and ecosystem evolution. Climate change and human activities represent the primary drivers of water resource variability. The agro-pastoral transitional zone in northern China serves as a critical ecotone between semi-humid agricultural regions and arid/semi-arid pastoral areas. Characterized by its vulnerability and sensitivity, this zone exhibits pronounced responsiveness to global change. Precipitation in this region demonstrates substantial interannual variability, and the relatively dry climate coupled with scarce water resources creates significant constraints on socio-economic development. Since the 1990s, implementation of major ecological restoration projects has further intensified conflicts between water demands for production and ecological recovery. Consequently, understanding regional hydrological processes to enable rational water allocation and utilization is essential for mitigating local water use conflicts and improving water use efficiency.

Hydrological modeling represents the primary approach for understanding regional hydrological processes. Current hydrological simulation models are diverse, primarily categorized into lumped, distributed, and semi-distributed models. In semi-distributed models, hydrological response units typically correspond to sub-watersheds, whereas distributed models employ uniformly sized grid cells as response units. Distributed hydrological models are thus more compatible with refined water resource management and have gained wider application. The SWAT model stands as the most widely applied distributed hydrological model, capable of simulating various hydrological physicochemical processes including water quantity, sediment transport, and chemical migration and transformation. It has been applied across diverse geographical environments, including tropical, temperate, and alpine regions, and has been extensively used for sediment quantitative assessment in China's Yellow River basin.

The DHSVM model, as a physically based distributed hydrological model, demonstrates greater adaptability to varied geographical settings and has found broad application in agriculture and future water resource prediction. However, both distributed and semi-distributed models possess inherent limitations in practical application. The SWAT model's coupling methodology often neglects dynamic feedback between socio-economic water use processes and natural hydrological processes. DHSVM exhibits high sensitivity to soil and vegetation parameterization, yet parameter configuration is extremely cumbersome. Many studies lack access to long-term, complete observational datasets and must rely on default values, substantially reducing output precision.

While both models offer distinct advantages in different geographical contexts, current research tends to select hydrological models in isolation without adequately considering regional environmental characteristics. Moreover, applicability analysis of hydrological models for simulating hydrological processes in the eastern agro-pastoral transitional zone of northern China remains relatively scarce. This study selected two distinct watershed types in this region—a grassland-dominated watershed and a forestland/farmland-dominated watershed—to simulate hydrological processes using both DHSVM and SWAT models, aiming to identify the most suitable hydrological model for the study area and improve simulation accuracy. This research is significant for understanding hydrological processes in different watershed types within the eastern agro-pastoral transitional zone and for rationally allocating and utilizing water resources to address local water shortages.

1.1 Study Area Overview

The eastern section of the agro-pastoral transitional zone in northern China (40°–49°N, 115°–125°E), with an area of approximately 35.24×10⁴ km², represents the transition from semi-humid to arid/semi-arid regions, demarcated by the 400 mm isohyet. Located in a semi-arid area with annual precipitation of 200–400 mm, the region receives concentrated precipitation during the growing season, accounting for approximately 70% of annual totals. Summer thunderstorms are common, while spring experiences severe water deficits. Annual evaporation ranges from 1500–1900 mm, yielding an evaporation-to-precipitation ratio of 3.4–4.89 and a humidity coefficient of 0.2–0.3. The Xar Moron River, Xiliao River, and Nen River flow through this zone. Considering hydrological station distribution, the study area encompasses the upper reaches of the Xar Moron River (a northern tributary of the Xiliao River) with the Balin Bridge hydrological station, where grassland dominates both riverbanks and the watershed area is approximately 1.29×10⁴ km². The upper reaches of the Laoha River (a tributary source of the Xiliao River) with the Dianzi hydrological station feature forestland and farmland along both banks, with a watershed area of approximately 0.17×10⁴ km².

1.2 Data Sources

Both models require underlying surface data (DEM, land use, soil type) and meteorological data. Land use data for 2020 were obtained from the Geographic Spatial Data Cloud at 30 m resolution. Soil type data (HWSD V1.2) were sourced from the National Tibetan Plateau Data Center at 250 m resolution. Loam, sandy loam, sand, and silt loam constitute the primary soil types in the Xar Moron River basin, while loam and sandy loam dominate the Laoha River basin. Meteorological data, including daily maximum temperature, minimum temperature, precipitation, wind speed, relative humidity, and sunshine duration, were obtained from the China Surface Climate Data Daily Value Dataset. Leaf Area Index (LAI) data for DHSVM were derived from the Global Land Surface Satellite (GLASS) dataset at 500 m resolution. Runoff data from the Balin Bridge and Dianzi hydrological stations served as validation data (Table 1).

Table 1 Data sources

Data Type Source Spatial Resolution Temporal Resolution DEM Geographic Spatial Data Cloud (https://www.gscloud.cn/) 30 m - Land use China Land Cover Dataset (http://www.geodata.cn) 30 m Annual Soil type HWSD V1.2, National Tibetan Plateau Data Center (https://www.tpdc.ac.cn/) 250 m - Meteorology China Surface Climate Data Daily Value Dataset (https://m.data.cma.cn/) - Daily LAI/Albedo GLASS (http://glass.umd.edu) 500 m 8-day Runoff Hydrological Yearbooks - Monthly

1.3.1 Distributed Hydrology Soil Vegetation Model (DHSVM)

DHSVM is a physically based distributed hydrological model that simplifies actual watersheds into regular computational grids, with each grid cell assigned specific vegetation and soil properties to enable spatially heterogeneous hydrological simulation. At each time step, the model simultaneously solves energy balance and water balance equations for every grid cell, simulating hydrological process dynamics. Water exchange between grids occurs via surface runoff and subsurface flow.

The model requires watershed boundary, river network, soil depth, land use, and soil type data as inputs. Meteorological inputs include average temperature, wind speed, relative humidity, shortwave radiation, longwave radiation, and precipitation. Shortwave radiation is calculated using the Angstrom-Prescott equation:

$$RS = \left(a_s + b_s \frac{n}{N}\right)R_a$$

where $RS$ is incoming shortwave radiation (MJ·m⁻²·d⁻¹), $R_a$ is extraterrestrial shortwave radiation (MJ·m⁻²·d⁻¹), $a_s$ is diffuse shortwave radiation coefficient (0.25 under average climate conditions), $b_s$ is direct shortwave radiation coefficient (0.5 under average climate conditions), $n$ is actual sunshine hours, and $N$ is maximum possible sunshine hours.

Longwave radiation is calculated as:

$$RL = f \cdot \epsilon_a \cdot \sigma \cdot T_{max}^4 + (1-f) \cdot \epsilon_a \cdot \sigma \cdot T_{min}^4$$

where $RL$ is incoming longwave radiation (MJ·m⁻²·d⁻¹), $\sigma$ is the Stefan-Boltzmann constant (5.67×10⁻⁸ W·m⁻²·K⁻¹), $\epsilon_a$ is atmospheric emissivity, $f$ is clear-sky fraction, $T_{max}$ is daily maximum temperature (K), and $T_{min}$ is daily minimum temperature (K). All raster input data were projected to a unified coordinate system and resampled to 30 m resolution.

1.3.2 Soil and Water Assessment Tool (SWAT)

SWAT is a distributed watershed hydrological model operating at a daily time step. The model first delineates the watershed into sub-basins, which are further subdivided into Hydrologic Response Units (HRUs) based on land use and soil type combinations, making HRUs the smallest simulation units. SWAT comprises two phases: the land phase, which controls water, sediment, nutrient, and chemical inputs to the main channel within each sub-basin, and the routing phase, which determines the transport of these materials through the river network to the watershed outlet.

The hydrological cycle is based on the water balance equation:

$$SW_t = SW_0 + \sum_{i=1}^{t}(R_{day} - Q_{surf} - E_a - W_{seep} - Q_{gw})$$

where $SW_t$ is final soil water content (mm), $SW_0$ is initial soil water content (mm), $t$ is time (days), $R_{day}$ is precipitation on day $i$ (mm), $Q_{surf}$ is surface runoff on day $i$ (mm), $E_a$ is evapotranspiration on day $i$ (mm), $W_{seep}$ is water entering the vadose zone from the soil profile on day $i$ (mm), and $Q_{gw}$ is return flow on day $i$ (mm).

SWAT requires DEM, land use, and soil type data. Soil types were reclassified to reduce categories, and soil parameters were calculated using the SPAW tool to build the soil database. Meteorological inputs include daily maximum temperature, minimum temperature, precipitation, relative humidity, solar radiation, and wind speed, formatted according to SWAT requirements and organized with a weather generator.

1.3.3 Parameter Calibration and Model Operation

DHSVM parameters are categorized into global parameters, vegetation parameters, and soil parameters. Vegetation parameters (e.g., stomatal resistance, vegetation height) and soil parameters (e.g., porosity, lateral saturated hydraulic conductivity, exponential decay rate) are particularly sensitive for hydrological simulation. This study employed the Extended Fourier Amplitude Sensitivity Test (EFAST) for sensitivity analysis. EFAST is a variance-based global method that integrates the advantages of the Sobol method.

SWAT contains numerous uncertain parameters, primarily calibrated automatically using the SUFI-2 algorithm. SUFI-2 offers five algorithms, including Particle Swarm Optimization (PSO) and Markov Chain Monte Carlo (MCMC). This study utilized the PSO algorithm, which employs Latin hypercube sampling to obtain parameter values for model simulation, with sensitivity analysis implemented through iterative algorithms.

Model performance was evaluated using the Nash-Sutcliffe efficiency coefficient (NSE) and coefficient of determination (R²). NSE ranges from -∞ to 1, with values closer to 1 indicating better fit between simulated and observed monthly runoff:

$$NSE = 1 - \frac{\sum_{i=1}^{n}(q_{obs}^i - q_{sim}^i)^2}{\sum_{i=1}^{n}(q_{obs}^i - q_{mean})^2}$$

where $q_{obs}$ is observed runoff, $q_{sim}$ is simulated runoff, and $q_{mean}$ is mean observed runoff.

2.1 Parameter Sensitivity Analysis

2.1.1 DHSVM Model

Using EFAST, parameters with sensitivity indices greater than 0.2 were classified as sensitive, while those below 0.2 were considered insensitive. In the upper Xar Moron River basin, seven primary sensitive parameters were identified: minimum stomatal resistance, bulk density, decay coefficient, soil bubbling pressure, lateral saturated hydraulic conductivity, leaf area index, and field capacity. Among these, bulk density, decay coefficient, soil bubbling pressure, lateral saturated hydraulic conductivity, and field capacity are soil parameters, while minimum stomatal resistance and leaf area index are vegetation parameters. In the upper Laoha River basin, six sensitive parameters were identified: minimum stomatal resistance, soil thermal conductivity, lateral saturated hydraulic conductivity, decay coefficient, field capacity, and leaf area index.

2.1.2 SWAT Model

The SUFI-2 algorithm conducted iterative sensitivity analysis on SWAT parameters including soil, land use, groundwater, snowmelt, and surface characteristics. With 1000 model runs and 5 iterations, sensitive parameters were selected from 26 base parameters for each watershed. The upper Xar Moron River basin yielded 11 sensitive parameters, while the upper Laoha River basin produced 12. Lateral saturated hydraulic conductivity, field capacity, and bulk density were calculated using the SPAW tool, with other parameters referenced from literature and model presets (Table 2, Table 3).

Table 2 Sensitive parameter values of DHSVM model in different watersheds

Parameter Xar Moron River Laoha River Lateral saturated hydraulic conductivity (m·s⁻¹) 1.05×10⁻⁵ 1.0×10⁻⁵ Field capacity (fraction) 0.34 0.30 Soil thermal conductivity (W·m⁻¹·K⁻¹) 1.1 1.1 Decay coefficient 3.4×10⁻⁶ 3.4×10⁻⁶ Bulk density (kg·m⁻³) 1.0×10³ 1.1×10³ Minimum stomatal resistance (s·m⁻¹) 150 150 Leaf area index GLASS LAI GLASS LAI

Table 3 Main sensitive parameter values of SWAT model in different watersheds

Parameter Xar Moron River Laoha River Curve number (CN2) 65 70 Surface runoff lag time (SURLAG) 0.5 0.8 Soil evaporation compensation coefficient (ESCO) 0.8 0.9 Groundwater evaporation coefficient (GWQMN) 500 600 Shallow groundwater runoff coefficient (ALPHA_BF) 0.2 0.3 Groundwater delay time (GW_DELAY) 15 20 Main channel Manning's n (CH_N2) 0.03 0.035 Main channel hydraulic conductivity (CH_K2) 5 8 Saturated hydraulic conductivity (SOL_K) 5.0 6.5 Soil available water capacity (SOL_AWC) 0.12 0.15 Soil bulk density (SOL_BD) 1.4 1.5

2.2 Model Calibration and Validation

Due to flow intermittence in rivers of the eastern agro-pastoral transitional zone of northern China, the growing season (May–September) was selected as the study period to improve simulation accuracy. Monthly observed runoff data from 2012–2019 were collected from hydrological yearbooks, with 2012–2014 used as warm-up, 2015–2017 for calibration, and 2018–2019 for validation.

Observed monthly runoff from Balin Bridge station (upper Xar Moron River) and Dianzi station (upper Laoha River) were used to calibrate and validate both models. In the upper Xar Moron River basin, DHSVM achieved NSE values of 0.70 and 0.11, and R² values of 0.73 and 0.12, during calibration and validation periods, respectively. SWAT attained NSE values of 0.43 and 0.04, and R² values of 0.45 and 0.05. In the upper Laoha River basin, DHSVM yielded NSE values of 0.56 and 0.70, and R² values of 0.58 and 0.72, while SWAT produced NSE values of 0.86 and 0.54, and R² values of 0.87 and 0.58 (Table 4).

Table 4 NSE and R² values between simulated and observed monthly runoff

Watershed Model Calibration NSE Calibration R² Validation NSE Validation R² Xar Moron River DHSVM 0.70 0.73 0.11 0.12 Xar Moron River SWAT 0.43 0.45 0.04 0.05 Laoha River DHSVM 0.56 0.58 0.70 0.72 Laoha River SWAT 0.86 0.87 0.54 0.58

2.3 Results and Analysis

2.3.1 Xar Moron River Basin

DHSVM demonstrated superior performance in simulating monthly runoff compared to SWAT, though both models exhibited relatively poor fit with observed flow. DHSVM's simulated monthly runoff trends closely matched observed trends, particularly during 2015–2017. However, both models performed poorly in 2018, with negative NSE coefficients that affected overall simulation results.

The models showed distinct advantages: DHSVM better simulated baseflow, while SWAT more accurately captured peak flows. SWAT's average simulated peak of 624.39 m³·s⁻¹ closely approached the observed average peak of 697.45 m³·s⁻¹, whereas DHSVM's average peak was 580.46 m³·s⁻¹. The performance differences may stem from the Xar Moron River's characteristics—from Baicha Estuary to Balin Bridge, the channel is wide and shallow with dispersed flow. DHSVM's input files contain detailed river network descriptions, enabling more refined watershed representation. However, DHSVM's soil parameterization emphasizes soil water regulation, causing overestimation during the 2018 validation period, with maximum discrepancies reaching 377.55 m³·s⁻¹ (Figure 4).

2.3.2 Laoha River Basin

Both models performed better in the upper Laoha River basin than in the Xar Moron River basin. DHSVM achieved NSE values of 0.56 (calibration) and 0.70 (validation), while SWAT attained 0.86 and 0.54, respectively. SWAT produced the best simulation results during the 2015–2017 calibration period, with an NSE of 0.86. DHSVM performed best during the 2018–2019 validation period, with an NSE of 0.70. Both models simulated monthly runoff peaks more accurately in this basin. However, simulated peak timing differed from observations: DHSVM and SWAT peaks occurred in July, while observed peaks appeared in August. Precipitation peaks occurred in July, suggesting that the models did not fully account for human impacts such as reservoir regulation (Figure 5).

Figure 4 DHSVM and SWAT simulated versus observed monthly runoff in the Xar Moron River basin

Figure 5 Precipitation-runoff processes in different watershed types

3 Discussion

From its source to Baicha Estuary, the Xar Moron River features steep slopes and rapid flow, primarily recharged by springs and groundwater. Downstream of Baicha Estuary to Balin Bridge, the channel widens and flow becomes dispersed. Current hydrological models primarily consider precipitation-driven recharge. DHSVM's flexible soil parameterization allows consideration of soil water regulation, and its detailed river network representation enables more accurate baseflow simulation compared to SWAT. Consequently, SWAT's baseflow simulation showed larger deviations from observed values, particularly during the validation period, affecting its overall accuracy. In contrast, SWAT's peak flow simulation was more accurate, with simulated peaks closely matching observed values.

Both models demonstrate applicability for hydrological process simulation in different watershed types, albeit with limitations. In parameter calibration, DHSVM relies on limited tools (e.g., maximum simulation iterations and parallel options) that cannot be adjusted based on data availability, constraining precision improvement. DHSVM also lacks robust sensitivity analysis methods; this study employed EFAST through algorithmic iteration, which may introduce errors. With over 100 adjustable parameters, many of which are difficult to obtain and rely on empirical values, simulation accuracy is inevitably affected.

Furthermore, both models inadequately represent human activity impacts. While SWAT includes a reservoir module, its operational rules are simplified and only represent seasonal and interannual variations. Standard DHSVM lacks a reservoir module, though modified versions have incorporated reservoir effects, improving simulation accuracy. Nevertheless, human impacts on watershed hydrology are diverse and extend beyond reservoirs. The upper Xar Moron River basin experiences minimal human disturbance, while the upper Laoha River basin contains farmland whose irrigation effects were not considered, potentially compromising simulation accuracy.

Comparatively, DHSVM's applicability extends beyond runoff simulation accuracy. Northern China's agro-pastoral transitional zone constitutes an important agricultural and pastoral production base severely constrained by water resources, with groundwater over-exploitation and drying rivers. DHSVM, using grid cells as its smallest unit, can output multiple raster datasets including evapotranspiration and soil moisture content at precise soil layer resolutions, facilitating understanding of spatiotemporal hydrological characteristics and comprehensive grasp of hydrological dynamics. Previous studies have successfully applied DHSVM in arid and semi-arid regions, such as estimating wheat yield through soil moisture monitoring and analyzing climate change impacts on waterlogging.

4 Conclusion

This study selected the physically based distributed DHSVM model and the SWAT model to simulate monthly runoff during the growing season in different watershed types within the eastern agro-pastoral transitional zone of northern China. The applicability of both models was evaluated, yielding the following conclusions:

1. Both models exhibit good applicability for hydrological process simulation in the study area, each with distinct advantages. DHSVM better simulates baseflow, while SWAT more accurately simulates peak flows. In the Xar Moron River basin, DHSVM's simulated peak (580.46 m³·s⁻¹) differed from the observed peak (697.45 m³·s⁻¹) by only 73.06 m³·s⁻¹.

2. In the Xar Moron River basin, DHSVM achieved NSE coefficients of 0.70 and 0.11 during calibration and validation periods, respectively, while SWAT attained 0.43 and 0.04. In the Laoha River basin, DHSVM yielded NSE coefficients of 0.56 and 0.70, compared to SWAT's 0.86 and 0.54. Overall, DHSVM demonstrates better applicability than SWAT across both watershed types.

3. Both models show better performance in the Laoha River basin than in the Xar Moron River basin, indicating superior applicability in watersheds dominated by forestland and farmland. The primary reason is that the Laoha River's recharge depends mainly on precipitation, whereas the Xar Moron River receives diverse recharge from springs and groundwater that are poorly represented in both models.

4. DHSVM's grid-based structure enables more detailed spatial analysis of hydrological processes, making it particularly suitable for regions requiring comprehensive water resource management strategies.

References

[1] Li M S, Yang X H, Wang K W, et al. Exploring China's water scarcity incorporating surface water quality and multiple existing solutions[J]. Environmental Research, 2024, 246: 118191.

[2] Scanlon B R, Fakhreddine S, Rateb A, et al. Global water resources and the role of groundwater in a resilient water future[J]. Nature Reviews Earth & Environment, 2023, 4(2): 87-101.

[3] He C Y, Liu Z F, Wu J G, et al. Future global urban water scarcity and potential solutions[J]. Nature Communications, 2021, 12(1): 1-10.

[4] Zhang Y J, Hao R F, Qin Y. Temporal and spatial variation of agricultural and pastoral production in the eastern section of the agro-pastoral transitional zone in northern China[J]. Agriculture Basel, 2024, 14(6): 829.

[5] Li Qiuyue, Pan Xuebiao. The impact of climate change on boundary shift of farming-pasture ecotone in northern China[J]. Journal of Arid Land Resources and Environment, 2012, 26(10): 1-6.

[6] Xie P, Wu Z Y, Sang Y F, et al. Evaluation of the significance of abrupt changes in precipitation and runoff process in China[J]. Journal of Hydrology, 2018, 560: 451-460.

[7] Jayathilake D I, Smith T. Understanding the role of hydrologic model structures on evapotranspiration driven sensitivity[J]. Hydrological Sciences Journal, 2020, 65(9): 1474-1489.

[8] Beven K J, Warren R, Zaoui J. SHE: Towards a methodology for physically based distributed forecasting in hydrology[J]. IAHS, 1980, 129: 133-137.

[9] Feyen L, Vázquez R, Christiaens K, et al. Application of a distributed physically based hydrological model to a medium size catchment[J]. Hydrology and Earth System Sciences, 2000, 4(1): 47-63.

[10] Zhang Z P, Wang Q Z, Guan Q Y, et al. Research on the optimal allocation of agricultural water and soil resources in the Heihe River Basin based on SWAT and intelligent optimization[J]. Agricultural Water Management, 2023, 279: 108177.

[11] Soufiane T, Lamia E, Youssef A, et al. The application of SWAT model and remotely sensed products to characterize the dynamic of streamflow and snow in a mountainous watershed in the High Atlas[J]. Sensors, 2023, 23(3): 1246.

[12] dos Santos F M, Pelinson N D, de Oliveira R P, et al. Using the SWAT model to identify erosion prone areas and to estimate soil loss and sediment transport in Mogi Guaçu River basin in Sao Paulo State, Brazil[J]. Catena, 2023, 222: 106872.

[13] Anteneh Y, Alamirew T, Zeleke G, et al. Modeling runoff sediment influx responses to alternative BMP interventions in the Gojeb watershed, Ethiopia, using the SWAT hydrological model[J]. Environmental Science and Pollution Research, 2023, 30(9): 22816-22833.

[14] Mabood F, Hadi F, Jan A U, et al. Assessment of Pb and Ni and potential health risks associated with the consumption of vegetables grown on the roadside soils in District Swat, Khyber Pakhtunkhwa, Pakistan[J]. Environmental Monitoring and Assessment, 2022, 194(12): 906.

[15] Zhang X Q, Chen P, Dai S N, et al. Analysis of non-point source nitrogen pollution in watersheds based on SWAT model[J]. Ecological Indicators, 2022, 138: 108881.

[16] Gu P F, Wu Y X, Liu G D, et al. Application of meteorological element combination driven SWAT model based on meteorological datasets in alpine basin[J]. Water Supply, 2022, 22(3): 3307-3324.

[17] Yang S, Tan M L, Song Q X, et al. Coupling SWAT and Biome-BGC models for improving daily scale hydro-climatic simulation and climate change impact assessment in a tropical river basin[J]. Journal of Environmental Management, 2023, 330: 117244.

[18] Tian P P, Lu H W, Feng W, et al. Large decrease in streamflow and sediment load of Qinghai-Tibetan Plateau driven by future climate change: A case study in Lhasa River Basin[J]. Catena, 2020, 187: 104340.

[19] Zhang X C, Yuan X, Liu H R, et al. Soil moisture estimation for winter wheat waterlogging monitoring by assimilating remote sensing inversion data into the Distributed Hydrology Soil Vegetation Model[J]. Remote Sensing, 2022, 14(3): 792.

[20] Ma Y, Xiong Q X, Zhu J Q, et al. Early warning indexes determination of the crop injuries caused by waterlogging based on DHSVM model[J]. The Journal of Supercomputing, 2020, 76(4): 2435-2450.

[21] Ridgeway J B, Surfleet C G. Effects of streamside buffers on stream temperatures associated with forest management and harvesting using DHSVM-RBM; South Fork Caspar Creek, California[J]. Frontiers in Forests and Global Change, 2021, 4: 611380.

[22] Sun N, Yearsley J, Voisin N, et al. A spatially distributed model for the assessment of land use impacts on stream temperature in small urban watersheds[J]. Hydrological Processes, 2015, 29(10): 2332-2346.

[23] Xu Y P, Gao X C, Zhu Q, et al. Coupling a regional climate model and a distributed hydrological model to assess future water resources in Jinhua River Basin, East China[J]. Journal of Hydrologic Engineering, 2015, 20(4): 04014054.

[24] Yang Mingzhi, Xu Jijun, Sang Lianhai, et al. Development and application of the distributed water resources allocation and regulation model based on hydrological cycle[J]. Journal of Hydraulic Engineering, 2022, 53(4): 456-470.

[25] Zhang Liren, Guo Zhenmin, Zeng Hongwei, et al. Application of multi-objective genetic algorithm on parameter optimization of DHSVM: A case study in Shihmen reservoir catchment[J]. Journal of Taiwan Agricultural Engineering, 2019, 65(1): 18-35.

[26] Wang Jingai, Shi Peijun. On the utilization of land resources and regional development strategies in the ecotone of agriculture and animal husbandry in Inner Mongolia[J]. Areal Research and Development, 1988, 7(1): 24-28.

[27] Zhang Jianchun, Chu Shaolin, Chen Quangong. Advances in defining the boundary of farming-grazing transition zone in China[J]. Pratacultural Science, 2008, 25(3): 78-84.

[28] Jin Yinghua, Zhou Daowei, Jiang Shicheng, et al. Research on climate changes and combinational law of temperature and precipitation during growing season in the eastern ecotone between agriculture and animal husbandry in Northern China[J]. Journal of Northeast Normal University (Natural Science Edition), 2007, 39(4): 137-142.

[29] Yang J, Huang X. The 30 m annual land cover dataset and its dynamics in China from 1990 to 2019[J]. Earth System Science Data, 2021, 13(8): 3907-3925.

[30] Zhao Yi, Nan Zhuotong, Li Xiangfei, et al. On applicability of a fully distributed hydrological model in the cold and alpine watershed of Northwest China[J]. Journal of Glaciology and Geocryology, 2019, 41(1): 147-157.

[31] Angstrom A. Solar and terrestrial radiation. Report to the international commission for solar research on actinometric investigations of solar and atmospheric radiation[J]. Quarterly Journal of the Royal Meteorological Society, 1924, 50(210): 121-126.

[32] Prescott J A. Evaporation from a water surface in relation to solar radiation[J]. Transactions of the Royal Society of South Australia, 1940, 64: 114-118.

[33] Xu Zongxue. Hydrological Models[M]. Beijing: Science Press, 2009.

[34] Tan M L, Gassman P W, Yang X Y, et al. A review of SWAT applications, performance and future needs for simulation of hydro-climatic extremes[J]. Advances In Water Resources, 2020, 143: 103664.

[35] Lan C, Giambelluca T W, Ziegler A D. Lumped parameter sensitivity analysis of a distributed hydrological model within tropical and temperate catchments[J]. Hydrological Processes, 2011, 25(15): 2405-2421.

[36] Zhao Yi. Research on the Distributed Model Simulation of the Spatial/Temporal Hydrological Characteristics in a Cold Alpine Basin[D]. Nanjing: Nanjing Normal University, 2019.

[37] Wang Jiandong, Guo Weidong, Li Hongqi. Application of extended Fourier amplitude sensitivity test (EFAST) method in land surface parameter sensitivity analysis[J]. Acta Physica Sinica, 2013, 62(5): 35-41.

[38] Zuo Depeng, Xu Zongxue. Distributed hydrological simulation using SWAT and SUFI-2 in the Wei River Basin[J]. Journal of Beijing Normal University (Natural Science), 2012, 48(5): 490-496.

[39] Liu Ning, Zhang Xia, Zhu Xueping, et al. Runoff simulation using SWAT model and SUFI-2 algorithm in Biliu River Basin[J]. Water Power, 2019, 45(3): 18-22, 89.

[40] Pan S L, Liu L, Bai Z X, et al. Integration of remote sensing evapotranspiration into multi-objective calibration of Distributed Hydrology Vegetation Model (DHSVM) in a Humid Region of China[J]. Water, 2018, 10(12): 1841.

[41] Stieglitz M, Rind D, Famiglietti J, et al. An efficient approach to modeling the topographic control of surface hydrology for regional and global climate modeling[J]. Journal of Climate, 1997, 10(1): 118-137.

[42] Meyer P D, Rockhold M L, Gee G W. Uncertainty analyses of infiltration and subsurface flow and transport for SDMP sites[J]. Office of Scientific and Technical Information, 1997, 51: 541818.

[43] Spellman P, Webster V, Watkins D. Bias correcting instantaneous peak flows generated using a continuous, semi-distributed hydrologic model[J]. Journal of Flood Risk Management, 2018, 11(4): e12342.

[44] Javaheri A, Babbar-Sebens M. On comparison of peak flow reduction effects of restored wetlands on channel flooding[J]. Ecological Engineering, 2014, 73: 132-145.

[45] Liu Q, Liang L Q, Cai Y P, et al. Assessing climate and land change impacts on streamflow in a mountainous catchment[J]. Journal of Water and Climate Change, 2020, 11(2): 503-513.

[46] Safeeq M, Fares A. Hydrologic effect of groundwater development in a small mountainous tropical watershed[J]. Journal of Hydrology, 2012, 428: 51-67.

[47] Lan C, Lettenmaier D P, Mattheussen B V, et al. Hydrologic prediction for urban watersheds with the Distributed Hydrology Vegetation Model[J]. Hydrological Processes, 2008, 22(21): 4205-4213.

[48] Nguyen T V, Dietrich J, Dang T D, et al. An interactive graphical interface tool for parameter calibration, sensitivity analysis, uncertainty analysis, and visualization for the Soil and Water Assessment Tool[J]. Environmental Modelling & Software, 2022, 156: 105497.

[49] Voisin N, Liu L, Hejazi M, et al. One-way coupling of an integrated assessment model and a water resources model: Evaluation and implications of future changes over the US Midwest[J]. Hydrology and Earth System Sciences, 2013, 17(11): 4555-4575.

[50] Zhao G, Gao H L, Naz B S, et al. Integrating a reservoir regulation scheme into a spatially distributed hydrological model[J]. Advances in Water Resources, 2016, 98: 16-31.

[51] Liu Zhe, Lancuo. Investigating the mechanisms of streamflow change in the Beichuan River Basin, Qinghai Province: Based on modeling and statistic analyses[J]. Progress in Geography, 2022, 41(2): 304-315.

[52] Liu Z J, Liu Y S, Li Y R. Anthropogenic contributions dominate trends of vegetation cover change over the farming-pastoral ecotone of northern China[J]. Ecological Indicators, 2018, 95: 370-378.

[53] Chawanda C J, Nkwasa A, Thiery W, et al. Combined impacts of climate and land use change on future water resources in Africa[J]. Hydrology and Earth System Sciences, 2024, 28(1): 117-138.

[54] Mahdian M, Hosseinzadeh M, Siadatmousavi S M, et al. Modelling impacts of climate change and anthropogenic activities on inflows and sediment loads of wetlands: Case study of the Anzali wetland[J]. Scientific Reports, 2023, 13(1): 5399.

[55] Van Wie J B, Adam J C, Ullman J L. Conservation tillage in dryland agriculture impacts watershed hydrology[J]. Journal of Hydrology, 2013, 483: 26-38.

[56] Ren Feiying, Xiong Qinxue. Response of wheat sub-surface waterlogging to climate changing using DHSVM model Simulations[J]. Journal of Triticeae Crops, 2020, 40(10): 1265-1274.