Responses of runoffs and sediment yields in a river basin to land cover/use changes under extreme conditions based on the SWAT model

Abstract

This study employs the Soil and Water Assessment Tool (SWAT) to investigate the dynamics of runoff and sediment in the Fuhe River Basin of Poyang Lake. After calibration and validation, the model's coefficient of determination (R²) for both runoff and sediment yield exceeded 0.9, indicating high model accuracy. During the study period from 2001 to 2010, the cropland area in the Fuhe River Basin decreased by 195.9 km², while forest and urban land areas increased by 105.4 km² and 86.1 km², respectively. By inputting the multi-year LULC data from 2001 to 2010 into the SWAT model, we assessed the impact of LULC changes on the multi-year water and sediment yields of the Fuhe River Basin. The results showed annual differences in water and sediment yields, both below 10 mm and 2 t/ha², respectively. At the sub-basin scale, LULC changes had a significant impact on water and sediment yields. Under the 2010 baseline landuse scenario, the simulated runoff and sediment yields were 837.11 m³/s and 4.32 × 10⁶ t, respectively. Compared to the baseline scenario, the reforestation scenario resulted in reductions in water and sediment yields by −1.0% and −11.6%, respectively. In contrast, the agricultural development scenario exacerbated soil erosion, leading to increases in water and sediment yields by 5.0% and 37.2%, respectively. On a seasonal timescale, results indicated that compared to the baseline scenario, the five extreme landuse scenarios led to increases in runoff and sediment yields in spring and winter by 11.3% and 162.6%, respectively, which were higher than the increases in summer and autumn of 3.0% and 150.0%, respectively, indicating a more significant impact of LULC changes in spring and winter.

Graphical abstract

Keywords

SWAT / SUFI-2 / landuse change / runoff and sediment

1 Introduction

Changes in Landuse and Land Cover (LULC) significantly influence the hydrological and sediment processes within watersheds (Anand et al., 2018). These changes can cause variations in runoff, evapotranspiration, reductions in surface runoff, flood frequency, sediment transport, and annual average runoff volume (Rogger et al., 2017; Li et al., 2018). The primary drivers behind these significant LULC changes include rapid urbanization, national resource conservation and utilization policies, and global climate shifts (Mao and Cherkauer, 2009; Elfert and Bormann, 2010; Dwarakish and Ganasri, 2015). In agricultural regions situated in fertile plains, swift landuse alterations can diminish soil fertility, leading to decreased crop yields and, in extreme cases, projecting a food crisis. It is therefore imperative to evaluate effects of landuse changes on runoff and sediment yield to help the sustainable management of water resources and ecological environment (Ahiablame and Shakya, 2016).

With the progress in information technology, the development of hydrological models has significantly advanced, introducing tools like the Hydrologic Engineering Center’s Hydrologic Modeling System (HEC-HMS) model (Younis and Ammar, 2018), the Soil and Water Assessment Tool (SWAT) model (Bieger et al., 2015), Hydrological Simulation Program-FORTRAN (HSPF) model (Zhang and Ross, 2015), and Hydrological Bureau’s Water Balance Department (HBV) model (Ashagrie et al., 2006). Among these, the SWAT model exhibits substantial applicability in simulating hydrological and sediment processes and understanding the hydrological properties in river and lake system.

For example, Shukla et al. (2023) leveraged the SWAT model to examine the repercussions of LULC changes on runoff within Germany’s Rur watershed. Their findings indicated that conversions from forest to urban land, agricultural land, and grassland escalated total runoff by 43%, 14%, and 4%, respectively. Similarly, a study on Iran’s Anzali watershed assessed the hydrological impacts of both dynamic and static landuse changes, revealing that although no significant long-term shifts were observed in the watershed’s water balance and sediment yield at a broad scale, localized increases in agricultural land at the sub-basin level led to rises in evapotranspiration, water yield, and sediment yield (Aghsaei et al., 2020). Conversely, urban expansion was associated with reductions in these metrics (Aghsaei et al., 2020). Tao et al. (2015) investigated the hydro-logical processes under different landuse scenarios in the Fuhe River Basin, also part of the Poyang Lake catchment, using the SWAT model. Their study showed that increasing forest or grassland areas reduced surface runoff, increased groundwater recharge, and evapotranspiration, indicating higher water retention capacity of forest and grassland (Tao et al., 2015). Gong et al. (2023) used a coupled landuse and global climate model to predict scenarios in the Ganjiang River Basin, another part of the Poyang Lake catchment. They found that runoff sensitivity to climate change was greater than to LULC changes, and there was no significant synergistic effect between the two. Ma et al. (2023) constructed three future landuse scenarios using the SWAT model to evaluate relative changes in water balance components and extreme runoff frequency. They predicted that by 2030, 2060, and 2090, the outflow at the Fuhe River Basin’s outlet would increase by 27%−30%, 24%−39%, and 35%−43%, respectively (Ma et al., 2023).

The Poyang Lake catchment, the largest lake system in China, has experienced significant transformations due to recent shifts in landuse and climate (Ma et al., 2023). These changes have led to more frequent droughts and floods, causing severe soil erosion in the rivers flowing into Poyang Lake (Tao et al., 2015). As a result, these environmental challenges have profoundly affected industrial and agricultural production, as well as water availability for residential purposes in the middle and lower plains of the Poyang Lake basin. It is therefore imperative to initiate targeted research on the hydrological and sediment dynamics in response to landuse changes within the Poyang Lake watersheds. Such research is crucial for developing tailored strategies to mitigate the adverse effects of these environmental shifts, ensuring the sustainable use of water resources and soil conservation in the region. Addressing this research gap will not only improve our understanding of watershed management but also contribute to formulating policies and practices that can effectively address the impacts of climate change and landuse evolution (Guo et al., 2008).

Aiming to explore the effects of runoff and sediment under various landuse scenarios using the SWAT model, this study focuses on the Fuhe River basin, a vital component of the Poyang Lake watershed system to address the runoff and sediment variations in response to a larger scale of environmental changes. The significance of this research is underscored by the basin’s essential role in supporting diverse ecological systems, agricultural activities, and human livelihoods. By constructing a range of landuse scenarios that include both historical landuse patterns and projections of extreme landuse changes, the study employs the SWAT model to simulate the impacts on runoff and sediment yield.

The objectives of this paper are threefold. First, it seeks to assess the accuracy of the SWAT model in simulating the complex processes of runoff and sediment within the Fuhe River Basin. This involves a detailed evaluation of the model’s simulations compared to observed data, aiming to highlight the model’s precision and identify potential areas for refinement. Second, the study aims to investigate the hydrological and sediment responses to diverse landuse change scenarios. This includes examining how alterations in landuse, such as deforestation, urban expansion, and agricultural intensification, influence water dynamics and soil erosion within the basin. Understanding these impacts is crucial for predicting future changes and managing current challenges. Finally, the research strives to provide a scientific foundation for effective water resource management and landuse planning in the Fuhe River Basin. By elucidating the potential effects of various landuse scenarios on the basin’s hydrology, the study offers insights that can guide policymakers and stakeholders in making informed decisions that promote environmental sustainability and resilience to climate variability.

2 Study area, data, and method

2.1 Study area

The Fuhe River Basin, located in eastern Jiangxi Province, China, spans from 115°30′ E, 26°30′ N to 117°10′ E, 28°37′ N (Fig. 1). As a major tributary of the Poyang Lake system within the Yangtze River Basin, it originates in the Wuyi Mountains, extends 349 km, and covers 17186 km², making it the second-largest river in Jiangxi Province. Situated in a subtropical humid monsoon zone, the Fuhe River Basin experiences distinct seasonal variations with a moist and warm climate. Annual temperatures range from 17.1°C to 18.4°C, and precipitation ranges from 1740.7 mm to 2003.3 mm.

The landscape is predominantly forested (68.4%) and agricultural (27.2%), with soils comprising red and yellow. Significant tributaries, including the Xiuhe, Linshui, Litan, and Dongxiangshui Rivers, contribute to an annual runoff depth of 967.2 mm and an annual runoff volume of 15.3 billion m³. The basin’s hydrological processes are monitored by 16 meteorological stations, with the Lijiadu hydrological station serving as the primary outlet.

2.2 SWAT Model and configuration

The Soil and Water Assessment Tool (SWAT) model, developed by the United States Department of Agriculture (USDA) (Arnold et al., 1998), encompasses eight major components: hydrology, weather, sediment, soil temperature, plant growth, nutrients, agriculture/pesticides, and agricultural management. Surface runoff in SWAT is calculated using the Soil Conservation Service (SCS) curve number method, and potential evapotranspiration is estimated using the Penman-Monteith equation (Allen et al., 1989).

This study employs the Digital Elevation Model (DEM) of the Fuhe River Basin to subdivide it into sub-basins, determining a minimum drainage area threshold. Subsequent extraction of river networks, topographic factors, and river parameters resulted in 29 sub-basins.

Figure 2 illustrates the process flow chart implemented by the SWAT model. The SWAT model configuration includes input data, sub-basin division, HRU determination, sensitivity analysis, and parameter calibration. Input data consist of spatial and attribute databases of the study area, including DEM, landuse/cover maps, soil type maps, soil properties, meteorological data, and hydrological and sediment data for model calibration and validation.

2.3 Data

In this study, DEM data were sourced from the Geospatial Data Cloud platform using ASTER GDEM with a 30 m resolution, comprising five DEM images. The data were preprocessed using ArcGIS 10.2 for masking, projection transformation, and mosaicking to form the DEM of the Fuhe River Basin. Four slope categories were established: 0−5°, 5°−10°, 10°−15°, and greater than 15° (Fig. 3). Hydrological Response Units (HRUs) were defined as combinations of land use, soil, and slope. HRUs were generated by overlaying these factors in ArcSWAT, with thresholds for landuse, soil, and slope set at 0%, 5%, and 5%, respectively.

Meteorological data were sourced from the W3S Historical Weather Data Set on the SWAT official website. Daily precipitation and maximum/minimum temperature data from 2001 to 2010 were selected from 16 meteorological stations covering the entire Fuhe River Basin. A two-year model warm-up period was established.

Hydrological and sediment data included daily runoff volumes and river sediment loads from the Lijiadu Hydrological Station for the years 2001 to 2010, obtained from the Jiangxi Provincial Hydrology Bureau.

2.4 Sensitivity analysis, calibration and validation, and evaluation metrics

The SWAT model incorporates two sensitivity analysis methods: One-at-a-time (OAT) and Global Sensitivity Analysis. This study employed Global Sensitivity Analysis to examine the sensitivity of all parameters. Based on previous studies by Arnold et al. (1998) in the Fuhe River Basin, 28 parameters affecting runoff and sediment sensitivity were initially selected.

The SWAT-CUP (Calibration and Uncertainty Programs) 2012 software (Abbaspour et al., 2007), in conjunction with the SUFI-2 algorithm (McKay et al., 1979), was used to select calibration parameters and establish their initial range values. The Sequential Uncertainty Fitting Version 2 (SUFI-2) algorithm, a gradient search-based program, employs Latin Hypercube Sampling to compute parameters for runoff and sediment through multiple iterations of parameter calibration. Sensitivity analysis results were evaluated using the t-test, where higher t-values and lower P-values indicate greater sensitivity.

Within SWAT-CUP, the SWAT model simulation results were calibrated using observed runoff and sediment data to evaluate the model’s applicability in the Fuhe River Basin. Data from 2001 to 2002 served as the warm-up period to eliminate initial simulation impacts, with 2003 to 2006 as the calibration phase and 2007 to 2010 as the validation phase.

This study selects five indices for evaluating the model results: the coefficient of determination (R²), Nash-Sutcliffe efficiency coefficient (ENS), percent bias (PBIAS).

The calculation formula for the coefficient of determination (R²) is given by

(1)R^2=\frac{{\sum }_i{\left(Q_{m,i}-{\overline{Q}}_m\right)}^2{\left(Q_{s,j}-{\overline{Q}}_s\right)}^2}{{{\sum }_i{\left(Q_{m,i}-{\overline{Q}}_m\right)}^2{\sum }_i\left(Q_{s,i}-{\overline{Q}}_s\right)}^2},

where Q_m,i nd Q_s,j represent the observed and simulated values, respectively, and {\overline{Q}}_mand {\overline{Q}}_sare their averages. R² evaluates the consistency of data variation trends between observed and simulated values, ranging from 0 to 1, with values closer to 1 indicating better fit.

The Nash-Sutcliffe efficiency coefficient (ENS) is calculated as follows:

(2)\text{ENS}=1-\left[{\sum }_{i=1}^n(Q_o-Q_p{)}^2/{\sum }_{i=1}^n(Q_o-Q_{\text{avg}})\right],

where Q_o is the observed value, Q_p is the simulated value, and Q_avg is the average of the observed values. ENS ranges from −∞ to 1, with values above 0.5 indicating an acceptable fit, 0.65−0.75 indicating a good fit, and above 0.75 indicating a very good fit.

The percent bias (PBIAS) is given by

(3)\mathrm{P}\mathrm{B}\mathrm{I}\mathrm{A}\mathrm{S}=100\left[\frac{{\sum }_{i=1}^n{\left(X_i^{\mathrm{o}\mathrm{b}\mathrm{s}}-X_i^{\mathrm{s}\mathrm{i}\mathrm{m}}\right)}^2}{{\sum }_{i=1}^n\left(X_i^{\mathrm{o}\mathrm{b}\mathrm{s}}\right)}\right],

where X_i^{\mathrm{o}\mathrm{b}\mathrm{s}}and X_i^{\mathrm{s}\mathrm{i}\mathrm{m}} represent the observed and simulated values, respectively. PBlAS assesses the trend of the simulated mean compared to the observed values. For runoff simulations, a PBlAS within ± 10% indicates good model performance, while for sediment simulations, a range within ± 15% is satisfactory. PBlAS values greater than 0 indicate underestimation, while values less than 0 indicate overestimation (Moriasi et al., 2007).

3 Model sensitivity and validations

3.1 Sensitivity results

In the SWAT hydrological model, different parameters have varying degrees of influence on the watershed hydrological processes. Therefore, parameter uncertainty in the model may affect the accuracy of the simulation results. In this study, we conducted a sensitivity analysis of hydrological and sediment parameters for the Fuhe River Basin. Initially, we selected 28 parameters and employed the SUFI-2 algorithm to perform the sensitivity analysis. The sensitivity of each parameter was determined based on its t-statistic and P-value, and after multiple rounds of calibration, several key hydrological and sediment parameters were identified, as shown in Table 1.

The three most sensitive parameters are the main channel Manning’s coefficient (CH_N2), the rate of temperature change (TLAPS), and the threshold depth for percolation to the deep aquifer (REVAPMN). Under wet conditions, the SCS runoff curve number (CN2), soil conductivity (SOL_K), and soil evaporation compensation factor (ESCO) are also sensitive to runoff. Among them, the most critical parameter affecting hydrological processes is CH_N2. CH_N2, which represents the Manning’s coefficient, is a crucial parameter that impacts river flow velocity. A higher Manning’s coefficient indicates greater channel roughness and slower flow velocity. Therefore, CH_N2 directly affects runoff, flow regulation, and hydrological response processes. TLAPS, representing the rate of air temperature decrease with elevation, influences evaporation, precipitation, and plant growth, making it critical for hydrological cycle simulation in mountainous terrains. REVAPMN indicates the minimum soil moisture content or depth required for percolation to the deep aquifer, significantly affecting groundwater recharge, evapotranspiration, and soil moisture dynamics. CN2 estimates surface runoff under specific soil moisture conditions, with higher values resulting in greater runoff and reduced infiltration (Teklay et al., 2019).

Due to the tight correlation between sediment yield and runoff volume, four sediment parameters were selected for calibration based on previous studies (Aghsaei et al., 2020). These include the linear index of sediment re-entrainment (SPCON), the maximum flow velocity adjustment factor for sediment computation in sub-basins (ADJ_PKR), the exponent index of sediment re-entrainment (SPEXP), and the conservation practice factor in the USLE equation (USLE_P). After multiple calibrations, SPCON was determined to be the most sensitive parameter affecting sediment yield. SPCON is a linear parameter that controls the re-entrainment of sediment in the channel. It determines the erosion, transport, and deposition patterns of sediment, thereby directly influencing the intensity and extent of the sediment transport process.

3.2 Model calibration and validation

This study utilized ten years of observed monthly average runoff and sediment data from the Lijiadu Hydrological Station, spanning from 1 January 2001 to 31 December 2010. The SWAT model is calibrated and validated using SWAT-CUP 2012 software, achieving optimal results at the 161st iteration of 200. The coefficient of correlation (r) is 0.92, 0.96 for runoff and sediment yield (Fig. 4), demonstrating a strong fit between simulated and observed data. The coefficient of determination (R²) and Nash-Sutcliffe efficiency (ENS) in the calibration and validation period for runoff are 0.93 and 0.92, respectively (Table 2), and the both are 0.90 for the sediment yield, indicating a good match between simulated and observed values. The Percent Bias (PBIAS) for runoff and sediment yield are 6.50% and −3.70%, respectively, within the acceptable range of ± 10%, indicating that the accuracy fully meets the requirements.

From Fig. 5, it can be observed that during the calibration period, the trend of simulated runoff matches well with the observed runoff. The model simulates the peak runoff values effectively during the flood season (April, May, June), but the simulation performance declines during the dry season (November, December, January), with observed values being higher than the simulated ones. However, the overall result is consistent, and the sediment peak values are also accurately simulated. After calibration, parameters were updated in the SWAT database and rerun to validate using data from 2007 to 2010. In Table 2, validation metrics showed R² and ENS of 0.96 and 0.95 for runoff, respectively, and 0.96 and 0.84 for sediment yield, with PBIAS values of −4.60% for runoff and −0.10% for sediment yield, indicating the model’s under-simulates runoff and over-simulated sediment yield compared to observed data.

4 Responses of runoffs and sediment yields to LULC changes

4.1 LULC changes in the Fuhe River basin

The LULC classification data for the years 2001, 2004, 2007, and 2010 were obtained from the China Land Cover Data set (CLCD) (Yang and Huang, 2021) with an annual 30 m resolution from 1990 to 2020, achieving an overall accuracy of 80%. The original data set’s LULC classifications were reclassified in ArcGIS to generate six land use types for the SWAT model: cropland, forest, grassland, urban area, water, and bare land. The landuse transition matrix was then used to calculate the area conversions between different landuse types, thereby obtaining the multi-year LULC change status.

In 2010, the Fuhe River Basin was predominantly covered by forest land, accounting for 68.38% of the entire basin area. This forest land was extensively distributed in the upper and middle reaches of the Fuhe River. Following this, cropland and residential areas covered 27.24% and 2.72% of the basin area, respectively. Residential areas and cropland were primarily located in the plains of the lower reaches and the mountainous river valleys of the middle reaches of the Fuhe River. Grasslands, water bodies, and bare land accounted for 0.09%, 1.55%, and 0.01% of the basin area, respectively. The main spatial distributions of Landuse and Land Cover (LULC) for the years 2001, 2004, 2007, and 2010 are presented in Fig. 6.

Comparing LULC data from 2001, 2004, 2007, and 2010 (Table 3), the area of cropland continuously decreased, with a reduction of 195.95 km² from 2001 to 2010. In contrast, the coverage of forest land steadily increased from 67.67% in 2001 to 68.38% in 2010, representing an increase of 105.40 km². This increase can be attributed to the Grain for Green policy, under which a large amount of cropland was restored to forest land. With economic development, the area of residential zones also saw a rapid increase, expanding by 86.10 km² over ten years. The areas of water bodies, grasslands, and bare land saw minor changes over this decade, overall maintaining a balance.

The areas of conversion between different landuse types are listed in Table 4. The reduction in cropland area was primarily converted into forest land, with a conversion area of 479.86 km², accounting for 86.93% of the total cropland conversion area, followed by conversion to residential land use (9.42%). There was also a significant area of forest land converted to cropland, totaling 595.89 km², indicating mutual land conversion between forest and cropland. The largest source of increase in residential area was cropland, with 120.28 km² converted, accounting for 75.47% of the total input area to residential land. There were also minor areas of cropland and forest land converted to water bodies, grasslands, and bare land. In summary, from 2001 to 2010, the main landuse type area conversions occurred among cropland, forest land, and residential areas.

4.2 Historical and extreme landuse scenario settings

Based on the multi-year land use changes in the Fuhe River Basin, land use data from the years 2001, 2004, 2007, and 2010 were selected as historical landuse scenarios. The year 2010 was chosen as the baseline year because, during the study period, the areas of cropland, forest land, and residential land in 2010 had the highest proportions. This selection allows for a better investigation of the hydrological and sediment conditions during the LULC changes across different years. The area of landuse types is then statistically analyzed according to the distribution of landuse types at different times (Table 3).

Five different landuse scenarios are designed: conversion of cropland to forest (S1), conversion of cropland to grassland (S2), urban development (S3), agricultural development (S4), and soil and water loss (S5). The study quantitatively investigates the changes in runoff and sediment load under different landuse scenarios, with specific proportions of different extreme landuse types presented in Table 5.

The extreme landuse scenarios specifically reflect the outcomes of accelerating or reversing the observed landuse change trends. For instance, scenarios converting cropland to forest (S1) or grassland (S2) explore the hydrological implications of further afforestation or natural vegetation restoration, akin to the effects of the Grain for Green policy. The urban development (S3) scenario examines the hydrological impact of continued urban expansion, while the agricultural development (S4) scenario considers the effects of increasing agricultural land, contrary to the observed trend. The soil and water loss (S5) scenario likely investigates the consequences of inadequate land management practices leading to degradation.

4.3 Hydrological and sedimental responses to the LULC changes

To investigate the impact of LULC changes on runoff and sediment yield in the Fuhe River Basin, this study examines three scales: watershed, sub-watershed, and seasonal. At the watershed scale, the study uses landuse classification maps from four different historical periods to analyze the effects of LULC changes on water resources and sediment yield in the basin. At the sub-watershed scale, the 2nd sub-watershed, where the Lijiadu hydrological station is located, is selected as the study area to explore the runoff and sediment responses to extreme LULC changes. Finally, the study examines the seasonal runoff and sediment output under extreme LULC changes.

4.3.1 Watershed scale

To simulate the water balance components, total water yield, and total sediment yield in the Fuhe River Basin using landuse data from different periods, this study calculates the annual averages. As shown in Table 6, the annual average changes in water yield and sediment yield at the watershed scale are both less than 5%, and the long-term interannual differences in water balance components are each less than 10 mm. This indicates that landuse changes at the watershed scale do not have a significant impact on runoff and sediment yield. The outputs of the SWAT hydrological model show that groundwater flow (GW_Q), lateral flow (LATQ), and total water yield (WYLD) decreased by 6.06 mm, 1.56 mm, and 5.81 mm, respectively, while actual evapo-transpiration (ET) and total sediment load (SYLD) increased by 3.84 mm and 0.86 t/ha², respectively. The conversion of forest land to cropland may be one reason for the increase in evapotranspiration. The observed decrease in groundwater flow and lateral flow could be due to reduced soil infiltration and increased surface runoff.

Since 2000, comprehensive soil and water conservation efforts, including reforestation and grassland restoration, have been strengthened in the Fuhe River Basin. This has led to an increase in forest and grassland areas and higher vegetation coverage, resulting in higher actual evapotranspiration (ET). Consequently, this may have led to a reduction in runoff under the 2010 scenario.

The impact of land use changes (LULC) on water balance components and sediment yield at the watershed scale is minimal (Fohrer et al., 2001). Other researchers have also considered incorporating landuse change information into the SWAT hydrological model, but the simulation results of water balance components indicate that the impact of landuse changes is small. In a medium-sized watershed in India, Wagner et al. (2016) found no significant impact on hydrological responses at the watershed scale. Similarly, Wang et al. (2018) pointed out that monthly hydrological responses are more affected by the implementation of landuse dynamics compared to annual responses.

4.3.2 Sub-watershed scale

At the sub-basin scale, landuse changes significantly affect hydrological and sediment responses within the watershed. A simulation study was conducted to examine the runoff and sediment responses of five different land use types—cropland, forest, grassland, urban, and bare land—under extreme use scenarios.

In Figs. 7 and 8 and Table 7, simulation results revealed the runoff volumes for the five extreme landuse scenarios as follows: 828.49, 852.49, 828.73, 878.20, and 830.00 m³/s. The agricultural development scenario (S4) had the highest runoff volume, while the cropland-to-forest scenario (S1) had the lowest. Compared to the 2010 baseline scenario, the transformation from cropland to forest and urban land resulted in decreased runoff volumes (−1.03% and −1.00%, respectively), while conversion to grassland increased runoff volume (1.84%). This indicates that grasslands have a higher water production capacity than urban and forested lands, likely due to the greater ability of urban structures and forest vegetation to absorb and obstruct rainfall compared to grasslands. Furthermore, converting forest to cropland significantly increased runoff volume (4.91%), demonstrating the superior runoff generation capacity of cropland over bare land. This reflects the contribution of high soil moisture content in croplands to surface runoff formation, whereas the dry and loose soil of bare lands more readily absorbs moisture, promoting the formation of subsurface flow.

Regarding sediment changes, the variations in scenarios S1, S2, and S3 were relatively minor, fluctuating within ± 15%, while scenarios S4 and S5 showed significant changes. Notably, in scenario S5, sediment transport exceeded more than twice the baseline scenario (over 200%), indicating a substantial increase in soil erosion for bare land conversion. In the 2010 baseline scenario, the total sediment load for sub-basin 2 was 4.32 × 10⁶ tons. The lowest sediment load was observed in the cropland- to-forest scenario (S1) at 3.80 × 10⁶ tons, highlighting the significant soil conservation effects of extensive forest vegetation. Conversely, the forest-to-bare land scenario (S5) resulted in the highest sediment load at 9.65 × 10⁶ tons, demonstrating that bare land, lacking vegetation protection, is prone to generating substantial sediment under rainfall erosion. The variation in sediment loads across different scenarios further confirms the critical impact of landuse type on watershed sediment responses, with the highest sediment production capacity observed in the bare land scenario.

4.3.3 Intra-annual scale

In the Fuhe River Basin, the extreme unevenness and seasonal variation in the distribution of runoff and sediment throughout the year underscore the importance of studying their seasonal allocation. This study selects Sub-basin 2, where the Lijiadu Hydrological Station is located, as a case study. It simulates monthly runoff and sediment load under five extreme landuse scenarios to calculate their seasonal averages for comparison with the baseline scenario of 2010.

In Fig. 9, simulation results indicate that the spring runoff volume change rate ranges from −3.62% to 11.02%, with the agricultural development scenario (S4) showing the highest rate, reflecting the significant impact of landuse change on runoff during the dry season. In summer, the rainy season, the change rates for all scenarios are relatively minor, mostly around ± 5%, suggesting that the abundant water diminishes the impact of landuse changes. Autumn shows a wider range of runoff change rates, from −7.68% to 7.24%, with the largest decrease in the cropland-to-grassland scenario (S2) and a slight decrease in the agricultural development scenario (S4). In winter, change rates range from −12.79% to 6.44%, with the largest increase in the agricultural development scenario (S4), highlighting the sensitivity of landuse changes during the dry season. Thus, the average change rates for runoff in summer and autumn are 3.0%, while in spring and winter, they are 11.3%, indicating greater changes in runoff volumes during spring and winter than in summer and autumn.

The seasonal variation rates of sediment load significantly differ across landuse scenarios. In spring, the lowest change rates are observed, yet the bare land scenario (S5) reaches up to 113.56%, illustrating the significant impact of exposed soil on sediment transport. During summer, when river sediment transport is at its highest, changes across scenarios are minimal. In autumn, sediment change rates switch from positive to negative, with the largest decrease in S3 (−12.97%), indicating a relative reduction in sediment production. In winter, the highest change remains in S5 (154.61%), while scenarios converting cropland to forest and grassland (S1, S2, S3) show a significant decrease in sediment load, reflecting the importance of reforestation and grass restoration policies in protecting against soil erosion. The average change rates of sediment load in summer and autumn are 150.0%, and in spring and winter, 162.6%, suggesting greater sediment load changes in spring and winter than in summer and autumn. These results highlight the significant impact of seasonality and landuse changes on the hydrological and sediment responses in the Fuhe River Basin, as well as the effectiveness of corresponding soil and water conservation measures.

5 Discussions

5.1 Uncertainty in the simulation process

In this study, the SWAT model was constructed for the Fuhe River Basin to quantify the impact of LULC changes on regional runoff and sediment yield from 2001 to 2010. While higher spatial resolution data are generally believed to enhance model performance, excessively high resolution may result in slower model operation and increased uncertainty. To ensure model accuracy, we tested landuse and soil data with different spatial resolutions and ultimately found that a 30-m resolution provided the best results.

However, despite the SWAT model’s capability to accurately simulate runoff and sediment yield, there are inherent uncertainties. First, the model has intrinsic uncertainties. During calibration and validation, the SUFI-2 algorithm in SWAT-CUP was used for evaluation. Nevertheless, the inherent errors in the SUFI-2 algorithm led to uncertainties in model parameterization (Ashraf Vaghefi et al., 2014; Abbaspour et al., 2017). To further reduce parameter uncertainty, we adopted a multi-objective parameter calibration approach. Key hydrological parameters, such as CN2, CH_N2, ESCO, and SPCON, were subjected to global and single-factor sensitivity analyses and optimized. Additionally, the complexity of hydrological and sediment processes, along with errors in hydrological and sediment data and model structural defects, introduces uncertainties in SWAT model parameterization, thereby affecting the accuracy of simulation results. To address high-uncertainty para-meters, such as soil conductivity and the evaporation compensation factor, we implemented multiple rounds of random sampling to mitigate the impact of local optima on overall model performance. Ultimately, the calibrated and validated simulation results aligned with the observed data from the Fuhe River Basin and with previous studies (Lu et al., 2020), demonstrating that while the SWAT model effectively simulates runoff and sediment yield, its results remain constrained by the aforementioned sources of uncertainty.

5.2 Impact of LULC changes on runoff and sediment yield

The study results indicate that LULC changes have a significant impact on runoff and sediment yield in the Fuhe River Basin. During the study period, the urbanization process accelerated, leading to a continuous decrease in cropland area and a rapid increase in urban impervious surfaces. The LULC in the Fuhe River Basin mainly transitioned from cropland, forest, and grassland to residential areas, consistent with previous research findings (Tao et al., 2015). To comprehensively analyze the impact of different LULC types on hydrological and sediment processes, we used the SWAT model to incorporate multi-temporal landuse data and simulate the effects of LULC structural changes at multiple time points on the basin's hydrological processes.

In analyzing the response of hydrological processes to changes in different LULC types, we focused on five land use types: cropland, forest, grassland, urban land, and bare land. By constructing a land use transition matrix, we simulated various extreme LULC scenarios to explore the effects of specific landuse changes on hydrological processes. The results demonstrated that an increase in forest and grassland areas typically reduces surface runoff and sediment yield, as these land cover types have high interception and evapotranspiration capacities, which enhance soil water storage. Conversely, the expansion of urban areas with more impervious surfaces accelerates surface runoff and reduces groundwater recharge, leading to increased sediment output (Getachew and Manjunatha, 2022; Yi et al., 2023; Liu et al., 2024).

Based on 2010 land use types, five extreme LULC scenarios were developed in this study, and the changes in runoff and sediment yield under these scenarios were examined. The simulation results aligned with the findings of other scholars (de Oliveira Serrão et al., 2022; Shukla et al., 2023), further highlighting the significant influence of LULC changes on hydrological and sediment processes. Therefore, future development in the Fuhe River Basin should control forest area and increase vegetation cover to mitigate flooding, soil erosion, and other ecological issues.

5.3 Limitations and constraints

Despite considering various LULC change scenarios and seasonal analyses, this study has some limitations. First, there is a lack of long-term data analysis. Limited by the availability of sediment and meteorological data, the analysis only covers the period from 2001 to 2010, which may restrict the understanding of long-term trends and patterns. Nkwasa et al. (2022) used 30 years of sediment data from 1989 to 2019 to construct the SWAT + model to predict sediment yield in data-scarce regions. Secondly, this study did not consider the impact of climate change on runoff and sediment yield, which might not accurately reflect the hydrological and sediment processes in the basin. Climate change directly affects the hydrological cycle and sediment transport processes in watersheds by altering precipitation patterns, temperature, evaporation, and transpiration. Several studies have shown that increased frequency and intensity of extreme rainfall events under climate change may lead to higher peak flow and intensified sediment transport. Additionally, rising temperatures will affect increased evaporation, thereby altering the temporal and spatial distribution of water resources in the watershed (Mundetia et al., 2023). The increase in rainfall intensity due to climate change will directly impact runoff, subsequently enhancing the erosive capacity of rivers and intensifying sediment resuspension and transport (Liu et al., 2022a). Other researchers have integrated climate change models with LULC changes to study regional hydrological processes (Liu et al., 2022b; Bennour et al., 2023; Gong et al., 2023). Lastly, this study lacks consideration of socio-economic factors. It primarily focuses on the impact of natural factors on runoff and sediment yield in the Fuhe River Basin, without accounting for socioaa-economic factors (such as population growth, land management policies), which could influence the driving factors of LULC changes. Verburg et al. (2004) demonstrated that socio-economic factors are crucial drivers of landuse changes and should be considered in the model.

Therefore, future research should incorporate more influencing factors, such as integrating climate change parameters, analyzing model uncertainties, and utilizing long-term data, to provide more scientific and rational support for the management and policymaking in the Fuhe River Basin.

6 Conclusions

This study systematically analyzed the impact of landuse changes (LULC) on hydrological and sediment responses in the Fuhe River Basin from 2001 to 2010 using the SWAT model. The results demonstrate that transitions between different landuse types significantly influence runoff and sediment yield within the basin. Specifically, the expansion of cropland and residential areas has led to a reduction in forest cover, resulting in a marked increase in runoff and sediment loads. Conversely, increasing forest and grassland areas effectively reduced runoff and sediment generation, highlighting the critical role of afforestation and grassland restoration in preventing soil erosion and maintaining watershed ecological stability.

By simulating five extreme landuse change scenarios, the study found that the agricultural development scenario (S4) resulted in the highest runoff volume (878.20 m³/s), while the cropland-to-forest scenario (S1) exhibited the lowest runoff volume (828.49 m³/s). This variation reflects the capacity of increased cropland areas to promote runoff generation, whereas forest and grassland areas play a significant role in reducing runoff and sediment loads. Additionally, the bare land scenario (S5) led to a substantial increase in sediment load, with a change rate of 113.56%, underscoring the crucial role of exposed soil in sediment transport. Seasonal analysis revealed that changes in runoff and sediment loads were more pronounced in spring and winter. The runoff change rates in spring and winter were 11.3%, compared to 3.0% in summer and autumn. Similarly, sediment load changes were more significant in spring and winter, at 162.6%, compared to 150.0% in summer and autumn. These results indicate that seasonal variations have a significant impact on the hydrological and sediment responses of the watershed, particularly during the dry season, where extreme LULC scenarios show heightened sensitivity.

In conclusion, this study highlights the substantial spatial and temporal impacts of landuse changes on water resources and sediment yield in the Fuhe River Basin. The findings provide scientific evidence for policymakers to develop effective land and water resource management strategies, ensuring the ecological security of the Poyang Lake watershed.

References

[1] Abbaspour K C, Vaghefi S A, Srinivasan R (2017). A guideline for successful calibration and uncertainty analysis for soil and water assessment: a review of papers from the 2016 International SWAT Conference.Water, 10(1): 6

[2] Abbaspour K C, Yang J, Maximov I, Siber R, Bogner K, Mieleitner J, Zobrist J, Srinivasan R (2007). Modelling hydrology and water quality in the pre-alpine/alpine Thur watershed using SWAT.J Hydrol (Amst), 333(2−4): 413–430

[3] Aghsaei H, Mobarghaee Dinan N, Moridi A, Asadolahi Z, Delavar M, Fohrer N, Wagner P D (2020). Effects of dynamic land use/land cover change on water resources and sediment yield in the Anzali wetland catchment, Gilan, Iran.Sci Total Environ, 712: 136449

[4] Ahiablame L, Shakya R (2016). Modeling flood reduction effects of low impact development at a watershed scale.J Environ Manage, 171: 81–91

[5] Allen R G, Jensen M E, Wright J L, Burman R D (1989). Operational estimates of reference evapotranspiration.Agron J, 81(4): 650–662

[6] Anand J, Gosain A K, Khosa R (2018). Prediction of land use changes based on Land Change Modeler and attribution of changes in the water balance of Ganga basin to land use change using the SWAT model.Sci Total Environ, 644: 503–519

[7] Arnold J G, Srinivasan R, Muttiah R S, Williams J R (1998). Large area hydrologic modeling and assessment part l: model development.J Am Water Resour Assoc, 34(1): 73–89

[8] Ashagrie A G, de Laat P J, de Wit M J, Tu M, Uhlenbrook S (2006). Detecting the influence of land use changes on discharges and floods in the Meuse River Basin–the predictive power of a ninety-year rainfall-runoff relation.Hydrol Earth Syst Sci, 10(5): 691–701

[9] Ashraf Vaghefi S, Mousavi S J, Abbaspour K C, Srinivasan R, Yang H (2014). Analyses of the impact of climate change on water resources components, drought and wheat yield in semiarid regions: Karkheh River Basin in Iran.Hydrol Processes, 28(4): 2018–2032

[10] Bennour A, Jia L, Menenti M, Zheng C, Zeng Y, Barnieh B A, Jiang M (2023). Assessing impacts of climate variability and land use/land cover change on the water balance components in the Sahel using Earth observations and hydrological modelling.J Hydrol Reg Stud, 47: 101370

[11] Bieger K, Hörmann G, Fohrer N (2015). The impact of land use change in the Xiangxi Catchment (China) on water balance and sediment transport.Reg Environ Change, 15(3): 485–498

[12] de Oliveira Serrão E A, Silva M T, Ferreira T R, Paiva de Ataide L C, Assis dos Santos C, Meiguins de Lima A M, de Paulo Rodrigues da Silva V, de Assis Salviano de Sousa F, Cardoso Gomes D J (2022). Impacts of land use and land cover changes on hydrological processes and sediment yield determined using the SWAT model.Int J Sediment Res, 37(1): 54–69

[13] Dwarakish G S, Ganasri B P (2015). Impact of land use change on hydrological systems: a review of current modeling approaches.Cogent Geosci, 1(1): 1115691

[14] Elfert S, Bormann H (2010). Simulated impact of past and possible future land use changes on the hydrological response of the Northern German lowland “Hunte” catchment.J Hydrol (Amst), 383(3−4): 245–255

[15] Fohrer N, Haverkamp S, Eckhardt K, Frede H G (2001). Hydrologic response to land use changes on the catchment scale. Phys Chem Earth, Part B Hydrol Oceans Atmos, 26(7−8): 577–582

[16] Getachew B, Manjunatha B R (2022). Impacts of land‐use change on the hydrology of Lake Tana Basin, Upper Blue Nile River Basin, Ethiopia.Glob Chall, 6(8): 2200041

[17] Gong L, Zhang X, Pan G, Zhao J, Zhao Y (2023). Hydrological responses to co-impacts of climate change and land use/cover change based on CMIP6 in the Ganjiang River, Poyang Lake basin.Anthropocene, 41: 100368

[18] Guo H, Hu Q, Jiang T (2008). Annual and seasonal streamflow responses to climate and land-cover changes in the Poyang Lake Basin, China.J Hydrol (Amst), 355(1−4): 106–122

[19] Li P, Li H, Yang G, Zhang Q, Diao Y (2018). Assessing the hydrologic impacts of land use change in the Taihu Lake Basin of China from 1985 to 2010.Water, 10(11): 1512

[20] Liu W, Wu J, Xu F, Mu D, Zhang P (2024). Modeling the effects of land use/land cover changes on river runoff using SWAT models: a case study of the Danjiang River source area, China.Environ Res, 242: 117810

[21] Liu Y, Wu G, Fan X, Gan G, Wang W, Liu Y (2022a). Hydrological impacts of land use/cover changes in the Lake Victoria basin.Ecol Indic, 145: 109580

[22] Liu Y, Xu Y, Zhao Y, Long Y (2022b). Using SWAT Model to assess the impacts of land use and climate changes on flood in the Upper Weihe River, China.Water, 14(13): 2098

[23] Lu J, Liu Z, Liu W, Chen X, Zhang L (2020). Assessment of CFSR and CMADS weather data for capturing extreme hydrologic events in the Fuhe River Basin of the Poyang Lake.J Am Water Resour Assoc, 56(5): 917–934

[24] Ma H, Zhong L, Fu Y, Cheng M, Wang X, Cheng M, Chang Y (2023). A study on hydrological responses of the Fuhe River Basin to combined effects of land use and climate change.J Hydrol Reg Stud, 48: 101476

[25] Mao D, Cherkauer K A (2009). Impacts of land-use change on hydrologic responses in the Great Lakes region.J Hydrol (Amst), 374(1−2): 71–82

[26] McKay M D, Beckman R, Conover W (1979). A comparison of three methods for selecting values of input variables in the analysis of output from a computer code.Technometrics, 21: 239–245

[27] Moriasi D, Arnold J, Liew M, Bingner R, Harmel R D, Veith T L (2007). Model evaluation guidelines for systematic quantification of accuracy in watershed simulations.Trans ASABE, 50(3): 885–900

[28] Mundetia N, Sharma D, Sharma A, Dubey S K, Mitra B K, Dasgupta R, Jeong H (2023). Assessment of hydrological response with an integrated approach of climate, land, and water for sustainable water resources in the Khari River Basin, India.Anthropocene, 41: 100373

[29] Nkwasa A, Chawanda C J, van Griensven A (2022). Regionalization of the SWAT+ model for projecting climate change impacts on sediment yield: an application in the Nile Basin.J Hydrol Reg Stud, 42: 101152

[30] Rogger M, Agnoletti M, Alaoui A, Bathurst J C, Bodner G, Borga M, Chaplot V, Gallart F, Glatzel G, Hall J, Holden J, Holko L, Horn R, Kiss A, Kohnová S, Leitinger G, Lennartz B, Parajka J, Perdigão R, Peth S, Plavcová L, Quinton J N, Robinson M, Salinas J L, Santoro A, Szolgay J, Tron S, van den Akker J J H, Viglione A, Blöschl G (2017). Land use change impacts on floods at the catchment scale: challenges and opportunities for future research.Water Resour Res, 53(7): 5209–5219

[31] Shukla S, Meshesha T W, Sen I S, Bol R, Bogena H, Wang J (2023). Assessing impacts of land use and land cover (LULC) change on stream flow and runoff in Rur Basin, Germany.Sustainability (Basel), 15(12): 9811

[32] Tao C, Chen X, Lu J, Gassman P W, Sabine S, José-Miguel S P (2015). Assessing impacts of different land use scenarios on water budget of Fuhe River, China using SWAT model.Int J Agric Biol Eng, 8: 95–109

[33] Teklay A, Dile Y T, Setegn S G, Demissie S S, Asfaw D H (2019). Evaluation of static and dynamic land use data for watershed hydrologic process simulation: a case study in Gummara Watershed, Ethiopia.Catena, 172: 65–75

[34] Verburg P H, Schot P P, Dijst M J, Veldkamp A (2004). Land use change modelling: current practice and research priorities.GeoJournal, 61(4): 309–324

[35] Wagner P D, Bhallamudi S M, Narasimhan B, Kantakumar L N, Sudheer K P, Kumar S, Schneider K, Fiener P (2016). Dynamic integration of land use changes in a hydrologic assessment of a rapidly developing Indian catchment.Sci Total Environ, 539: 153–164

[36] Wang Q, Liu R, Men C, Guo L, Miao Y (2018). Effect of dynamic land use inputs on improvement of SWAT model performance and uncertainty analysis of outputs.J Hydrol (Amst), 563: 874–886

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

[38] Yi H, Zhang X, He L, He J, Tian Q, Zou Y, An Z (2023). Detecting the impact of the “Grain for Green” program on land use/land cover and hydrological regimes in a watershed of the Chinese Loess Plateau over the next 30 years.Ecol Indic, 150: 110181

[39] Younis S M Z, Ammar A (2018). Quantification of impact of changes in land use land cover on hydrology in the upper Indus Basin, Pakistan.Egypt J Remote Sens Space Sci, 21(3): 255–263

[40] Zhang J, Ross M (2015). Hydrologic modeling impacts of post-mining land use changes on streamflow of Peace River, Florida.Chin Geogr Sci, 25(6): 728–738

RIGHTS & PERMISSIONS

Higher Education Press