The impact of future climate and land use changes on runoff in the Min-Tuo River Basin

With the rapid development of industrial and agricultural production, urbanization, and the significant increase in the demand for water in production and daily life, as well as inadequate management and protection of water resources, the problem of water scarcity is becoming increasingly severe (Huang et al. 2021). Currently, a key focus of water science development lies in the research of watershed hydrological cycles and water resources, with climate change (CC) and land use change (LUCC) being the main factors influencing watershed hydrological cycles and water resources (Yang et al. 2019). CC and LUCC significantly interact to impact the spatiotemporal evolution of surface runoff and will continue to play a role in the next 100 years (Yin et al. 2018). Therefore, simulating and predicting the hydrological responses of watersheds to both factors are of great significance for ensuring sustainable social development and promoting national economic construction (Pratoomchai et al. 2015).

In recent years, many scholars have analyzed the runoff response under the combined influence of CC and LUCC. For instance, Yaru et al. (2023) studied the Huangfuchuan basin, investigating changes in hydrology and land use within the basin. They used the Budyko method and Water and Energy transfer Process (WEP) model to assess the impact of CC and land surface changes on runoff. The results showed a significant decrease in runoff in the study area, with LUCC contributing more than CC. Feiyan et al. (2023) used the SWAT model to simulate runoff in the Xiaoxingkai Lake basin from 1961 to 2017 and estimated the effects of CC and LUCC on runoff. The results indicated that CC had the largest contribution, followed by direct human activities and LUCC. Verma et al. (2023b) used the SWAT model to study the impact of LULC changes and climate on the hydrology of the Mohanedi River Basin. The results showed that urbanization and vegetation reduction increased runoff and reduced evapotranspiration. Lastly, Yu et al. (2023) quantitatively analyzed the effects of LUCC and CC on runoff by constructing the CHESS model and setting scenario modes. They predicted future changes in basin runoff by combining the cellular automata (CA)-Markov model. The results indicated that LUCC and CC respectively inhibited and promoted runoff in the Liuxihe Reservoir basin, with CC having a greater influence, leading to an increasing trend in the average annual runoff in the basin. The hydrological modeling approach, considering the spatial distribution differences of various factors, has been widely applied (Kavetski & Fenicia 1990).

The Climate Model Intercomparison Project (CMIP) is an international collaborative project initiated by the World Climate Research Programme (WCRP) with the goal of understanding past, present, and future climate changes by collecting and comparing simulation results from various global climate models (GCMs). The latest iteration, the Coupled Model Intercomparison Project Phase 6 (CMIP6), represents a significant improvement over CMIP5 in many aspects. It possesses enhanced capabilities to more accurately predict and assess the impacts of CC, featuring higher resolution and more comprehensive data records (Tian et al. 2021; Guo et al. 2022; Verma et al. 2023a). In recent years, scholars (Saeed & Mahdi 2021; Li et al. 2022a, b) have utilized the multi-model ensemble mean (MME) approach to enhance the performance of precipitation and temperature prediction models based on comparisons of GCMs, effectively improving the precision and accuracy of model predictions. Runoff, as a crucial component of the surface water cycle, serves as a comprehensive representation of complex hydrological processes influenced by various factors including CC (Mehta et al. 2023). It holds significant research value, and the study of CC and runoff using CMIP6 and the Soil and Water Assessment Tool (SWAT) model is currently a hot topic in research (Yong et al. 2022; Jianzhu et al. 2023; Wang et al. 2024).

The Min-Tuo River Basin is strategically positioned as a crucial link between the ‘Belt and Road Initiative’ and the Yangtze River Economic Belt. It serves as an important ecological barrier in the upper reaches of the Yangtze River, abundant in water resources. Rational planning and development of water resources in this region are vital for achieving the goals of ‘peak carbon’ and ‘carbon neutrality’. However, due to the complex terrain, monsoons, uneven distribution of annual precipitation and intense rainfall patterns, the region is highly susceptible to CC, experiencing severe water and soil erosion, frequent droughts and floods, and escalating water security concerns. These issues are significantly impeding the socio-economic development and water security of the basin.

However, existing studies on future runoff responses in the Min-Tuo River Basin have primarily focused on historical observational data (Hou et al. 2018; Yuhang et al. 2021) or the impacts of individual elements (Yonggui et al. 2023; Jiang et al. 2024), failing to address the effects of various land use types on runoff. Simultaneously, there is a lack of in-depth research on future LUCC and runoff variations. Therefore, this study aims to simulate future LUCC in the Min River Basin using the CA-Markov model and estimate future CC using four scenarios under the latest CMIP6 models. By inputting future CC and LUCC into the SWAT model, reasonable simulations and predictions of future runoff can be made. The estimated results will provide scientific support for the strategic planning, sustainable development, and water security of the basin, playing a crucial role in achieving China's ‘peak carbon’ and ‘carbon neutrality’ goals, as well as reducing water and soil erosion.

The Min-Tuo River Basin encompasses several populous large and medium-sized cities, marked by high population density and thriving industrial and agricultural sectors. The region boasts abundant natural and hydraulic resources, providing a substantial material foundation for urban socio-economic development and residents’ livelihoods. However, with the advancement of agricultural modernization, industrial agglomeration, urbanization, and the escalating occurrence of regional extreme climate events, the pressure on water resources in the Min-Tuo River Basin is becoming increasingly pronounced (Peng et al. 2021). In 2021, the total water consumption in the basin reached 12.02 × 10¹⁰ m³, accounting for 49.18% of the province's total water consumption. Consequently, there is an urgent need for in-depth research on the impact of runoff in this basin to comprehensively address the challenges facing water resources management.

Modeling data for the SWAT model is detailed in Table 1, encompassing the digital elevation model (DEM), meteorological, and runoff data within the study area. Additionally, daily meteorological data (including precipitation, maximum temperature, minimum temperature, humidity, wind speed, and sunshine hours) from 15 meteorological stations covering the entire basin for the period 1979–2021 are collected. Furthermore, data from CMIP6 GCMs are selected for studying CC impacts.

Historical data in CMIP6 are available until 2014, with future periods spanning from 2015 to 2060, aligning with the goal of achieving ‘carbon neutrality’ by 2060. Six GCMs (as shown in Table 2) that simulate well within the study area, under the ‘low-carbon’ (SSP1-2.6 and SSP2-4.5) and ‘high-carbon’ (SSP3-7.0 and SSP5-8.5) emission pathways, are selected for daily meteorological data (as presented in Table 3) to conduct MME studies, aiming to reduce the uncertainty of future CC projections (Jin et al. 2022; Wu et al. 2023).

The SWAT model, a semi-distributed hydrological model developed by the United States Department of Agriculture (USDA), finds wide application in runoff simulation, non-point source pollution control, and analysis of the impact of climate and underlying surface changes. Favored as a tool to simulate watershed water cycles, the SWAT model takes into account both natural and anthropogenic factors influencing hydrological elements. By incorporating CMIP6 data, it can accurately predict future runoff trends. In the study area, which is located in a high-altitude region with inconvenient transportation, the SWAT model demonstrates remarkable simulation capabilities, effectively compensating for the lack of continuous meteorological and hydrological data.

The hydrological land cycle in the model consists of two main components: the river network confluence stage and the slope runoff generation stage. Slope runoff generation refers to the process where precipitation, after infiltrating through the surface, flows from the slope into rivers. The river network confluence is the process where runoff, formed during the slope runoff generation stage, accumulates and flows to the watershed outlet (Neitsch et al. 2011).

In the formula, ⁠, are the soil moisture content, precipitation, surface runoff, evapotranspiration, infiltration water flow, and groundwater flow in the early and late stages of the i day, respectively, mm; t is time, d.

In the formula, Q_s,i is the simulated runoff, m³/s; Q_m,i is the measured runoff, m³/s; is the average value of the measured runoff, m³/s; is the average value of simulated runoff, m³/s; and i is the length of the simulated sequence.

The CA-Markov model is currently extensively utilized for simulating future LUCC (Ackom et al. 2020; Wei et al. 2021).

The methodology of this study involves several key steps. Initially, actual LUCC data for the years 2000, 2010, and 2020 are imported to establish the LUCC transition matrix of the watershed. Subsequently, by integrating factors such as urban development within the watershed, topographical conditions, and applying constraints based on transportation routes, elevation, slope, and other relevant data, a set of suitability maps for LUCC simulation is generated. Following this, based on the measured LUCC data for 2000 and 2010, the LUCC transition matrix, and the suitability map set, the CA-Markov model is employed to simulate the LUCC for the study area in 2020. A comparative analysis is then conducted between the simulated LUCC for 2020 and the actual LUCC data to assess the applicability of the CA-Markov model. Subsequent simulations are performed for the years 2030–2050 based on the LUCC trends from 2000 to 2020, the LUCC transition matrix, and the suitability map set.

In the formula, and are the correct simulated grid ratio and the correct simulated grid ratio under random conditions, respectively. The research shows that when > 0.75, the measured LUCC2020 and the simulated LUCC2020 have significant consistency, and the simulation effect is good.

In the formula, and are the reconstructed future precipitation and temperature series, respectively; and are the multi-year average precipitation and temperature of the observation field during the reference period; and are the future precipitation and temperature series predicted by the climate model, respectively; and is the multi-year average precipitation and temperature simulated by the climate model during the reference period.

In the formula, is the result of multi-mode ensemble average; is the simulation result of each single model, and n is the total number of MME research models.

Baseline scenario S0: Base period 1979–2020 meteorological data and measured LUCC2020.

Using the combination of GCMs and LUCC, the following 11 combination scenarios were set up.

• S1: LUCC2020 is combined with meteorological data in the future 2021–2030;

• S2: LUCC2020 is combined with meteorological data in the future 2031–2040;

• S3: LUCC2020 is combined with meteorological data in the future 2041–2050;

• S4: LUCC2020 is combined with meteorological data in the future 2051–2060;

• S5: LUCC2020 is combined with meteorological data in the future 2021–2060.

• S6: Baseline meteorological data combined with future LUCC2030;

• S7: Baseline meteorological data combined with future LUCC2040;

• S8: Baseline meteorological data combined with future LUCC2050.

• S9: The future 2031–2040 meteorological data combined with the future LUCC2030;

• S10: The future 2041–2050 meteorological data combined with the future LUCC2040;

• S11: The future 2051–2060 meteorological data combined with the future LUCC2050.

The Mann-Kendall (MK) rank correlation test method is a widely used trend test method. Its advantage is that the sample does not need to follow a certain distribution and is not disturbed by a few outliers. It is more suitable for the trend test of non-normally distributed data such as hydrometeorological sequences. The calculation method is as follows:

The monthly runoff data from seven hydrological monitoring stations in the Min-Tuo River Basin were selected for calibration and validation. Due to variations in the years of observed runoff data at each monitoring station, the calibration and validation periods differ across sites, yet this discrepancy does not compromise the accuracy of calibration and validation. Parameter sensitivity analysis was conducted based on the observed monthly runoff data, and parameters exhibiting strong sensitivity were chosen for calibration adjustments (Table 4).

According to the results derived from six applicable GCMs for the Min-Tuo River Basin, the future CC scenarios for the basin under four emission pathways were obtained. Statistics were conducted on the corrected scenarios against the S0 scenario within the MME, revealing the mean annual precipitation, mean annual temperature, and their respective changes for the basin in the future years (Table 6). Under the low-carbon and high-carbon pathways, the future period projections for annual precipitation and temperature in the Min-Tuo River Basin show increases compared with the baseline period as follows: SSP1-2.6: 171.85 mm (+18.89%) and 1.74 °C, respectively, compared with the baseline period. SSP2-4.5: 65.9 mm (+7.24%), 1.78 °C and SSP3-7.0: 156.47 mm (+17.19%), 1.85 °C; SSP5-8.5: 178.65 mm (+19.62%), 2.13 °C.

The Kappa coefficient was used to calculate the accuracy of the measured LUCC2020 and the simulated LUCC2020 in the Min-Tuo River Basin. The Kappa coefficient was 0.87 (>0.75), which was significantly consistent, indicating that the CA-Markov model can be used to predict the future LUCC in the Min-Tuo River Basin.

From Table 7, it can be seen that the area of forest land, water area, construction land, and unused land in the Min-Tuo River Basin will increase by 2.09, 0.84, 2.39, and 0.29%, respectively, from 2020 to 2050; cultivated land and grassland area will be reduced by 5.45 and 0.16%. Among them, the increase of construction land is the most significant, and the area of cultivated land is greatly reduced.

Under the dual-carbon pathway, the future annual average runoff in the Min-Tuo River Basin shows a trend of decrease followed by an increase compared with the baseline period (S0). The minimum runoff occurs between 2031 and 2040, with S2 (SSP1.2-6, SSP2.4-5) decreasing by −3.11 and −3.36% compared with S0; the maximum runoff occurs between 2051 and 2060, with S4 (SSP2.4-5) increasing by 6.09% compared with S0. The average annual runoff changes over the 40-year period from 2021 to 2060 are −5.25 and −1.97 m³/s, with change rates of −1.35 and −0.51%.

Under the high-carbon pathway, the future annual average runoff in the Min-Tuo River Basin also shows a trend of decrease followed by an increase compared with the baseline period (S0). The minimum runoff occurs between 2031 and 2040, with S2 (SSP3.7-0, SSP5.8-5) decreasing by −5.21 and −2.41% compared with S0; the maximum runoff occurs between 2051 and 2060, with S4 (SSP3.7-0) increasing by 0.98% compared with S0. The average annual runoff changes over the 40-year period from 2021 to 2060 are −12.86 and −3.72 m³/s, with change rates of −3.31 and −0.96%.

As shown in Table 9, compared with simulated runoff under Scenario S0, the runoff volumes in the Min-Tuo River Basin vary by +0.05, +0.07, and +0.07% under different LUCC scenarios in the future. Moreover, there is a greater increase in runoff during the flood season, with respective increments of 0.11, 0.12, and 0.13%. Conversely, runoff decreases during the non-flood season, with reductions of −0.05, −0.08, and −0.07%, respectively. This phenomenon is expected to exacerbate flood control pressure and water resources management in the Min-Tuo River Basin in the future, making wet seasons wetter and dry seasons drier.

The increase in runoff resulting from LUCC in the Min-Tuo River Basin is primarily attributed to the changes in precipitation patterns associated with the projected LUCC scenarios. The reduction in arable land and grassland, coupled with the expansion of other land cover types, alters the surface characteristics and vegetation composition within the basin. These changes can lead to modifications in the infiltration capacity of the soil, surface runoff generation, and evapotranspiration rates, ultimately impacting the overall hydrological cycle.

Specifically, the conversion of arable land and grassland to other land cover types may reduce the overall vegetation cover and root depth, potentially decreasing the capacity of the land to intercept and retain water. This can result in increased surface runoff, as water is more readily able to flow over impermeable surfaces or through shallow-rooted vegetation. Additionally, changes in land use patterns may contribute to changes in soil compaction, erosion rates, and land surface roughness, further influencing runoff generation processes.

In the context of future climate and land use change data, analysis was conducted on the trends of runoff variations in the Min-Tuo River Basin for the period 2021–2060 under four shared socio-economic pathway (SSP) scenarios using the SWAT model. It was observed that the impact of future CC outweighs that of future LUCC on runoff, indicating an overall decreasing trend in runoff volume.

Based on estimations of future runoff in China under high temperature scenarios (SSP2-4.5 and SSP5-8.5) from CMIP6, Zhou et al. (2023a, b) suggested a long-term declining trend in runoff for the Yangtze River Basin. Ruijie et al. (2022) highlighted the threat to water security posed by decreased runoff in the upper Yangtze River, with the Minjiang River Basin contributing 22.1% of the runoff decline in the Yangtze River Basin. Studies by Xia & Wang (2008) noted a significant decrease in runoff in the Min and Tuo River basins, while Wang et al. (2008) analyzed the changing characteristics of annual runoff sequences in major rivers in the upper Yangtze River, indicating a decreasing trend in annual runoff sequences in the Jialing, Min, Tuo, and Fu Rivers located in the eastern and central parts of Sichuan.

These research findings align with the projected decrease in runoff in the Min-Tuo River Basin as outlined in this study. Considering the broader context of global CC, it is anticipated that future changes in precipitation and temperature may exacerbate water stress (Skii 2018; Mohmedraffi et al. 2022), increase the likelihood of severe events, and further decrease runoff, aligning with the outcomes of this research. Thus, there is a need to promote mitigation strategies and management measures to reduce these adverse impacts.

Existing research has overlooked the influence of LUCC changes and alterations in vegetation species on the hydrological cycle processes within the watershed. Furthermore, accurately disentangling the impacts of LUCC and vegetation species changes on hydrological cycle processes, especially the repercussions of LUCC, poses a significant challenge (Lv et al. 2023). Previous studies predominantly categorized the driving forces behind hydrological changes into two main factors: CC and anthropogenic activities, often treating LUCC as a proxy for human-induced alterations (Zhang et al. 2001). Nevertheless, LUCC modifies hydrological processes by influencing parameters such as infiltration, permeability, and soil water retention capacity during runoff generation, along with surface roughness, surface water retention, and river network connectivity during flow routing. Consequently, these alterations lead to an increase in canopy interception and the conversion of precipitation to evapotranspiration (Yang et al. 2017). Hence, human activities play a pivotal role in reshaping watershed hydrological cycles (Yongxin et al. 2022). Nonetheless, investigations into the mechanisms driving changes in hydrological processes due to LUCC are relatively scarce and warrant further exploration. The SWAT and CA-Markov models employed in this study exhibit inherent systematic errors. Despite their utility as reference points for future investigations into runoff dynamics and associated drivers, their precision necessitates enhancement (Guangxing et al. 2021; Makumbura et al. 2022). Therefore, comprehensive and precise delineation of DEM depressions, coupled with the refinement of SWAT model code and integration with high-resolution LUCC data, is imperative for capturing runoff characteristics that faithfully represent real-world basins and yield dependable future forecasts (Changzheng et al. 2022; Jiashuo et al. 2022).

Furthermore, CC research conforms to the ‘if-then-how’ paradigm, encompassing future emission scenarios, hydrological model selection, and land use dynamics (Zhang et al. 2010; Schneider et al. 2017). This adds layers of uncertainty to the impacts of CC on water flow. Although climate model-based predictions inherently harbor uncertainty, scenario-based simulations can aid in projecting future CC trends (Chevuturi et al. 2022). Preliminary investigations underscore the substantial contribution of climate model selection to the uncertainty associated with assessing CC impacts (Joseph et al. 2018). Hence, when evaluating the influence of regional CC on water resources, the judicious selection of future climate models and multi-scenario simulations is imperative to ensure the robustness of conclusions.

In four future scenarios, only under the SSP5-8.5 scenario does the future runoff slightly exceed historical levels, while in the other scenarios, it remains below historical levels. This indicates that future CC may exacerbate water scarcity in the basin. However, this study suggests that future LUCC changes will help alleviate water resource pressures. If policies aim to address the imbalance between water supply and demand in the basin, the current policy trajectory should be continued, which involves increasing future urban land area and promoting more conversion of cropland to grassland. If policies aim to alleviate flood pressure in the basin, the continuous expansion of future urban land area should be controlled, and more emphasis should be placed on converting cropland back to forest (Limin et al. 2022). In the future, under the influence of CC, the annual distribution of water resources will become increasingly uneven. It is recommended to construct large reservoirs and underground reservoirs within and around the basin to optimize water storage and distribution, ensuring water availability throughout the year and improving water resource utilization (Koulelis et al. 2023).

This study processed the biased CMIP6 models using MME to effectively alleviate the structural differences among individual models. Studies by Zhao et al. (2021) and Li et al. (2022a, b) based on the corrected MME for the Yellow River upstream and Yangtze River Basin precipitation changes, respectively, have demonstrated the superiority of MME in predicting future climate changes compared with single climate models. While MME has become a research hotspot, excessive focus on integrated models may limit the consideration of the impacts on the performance of individual climate models. Therefore, it is necessary to conduct research on the performance of independent climate models to better understand the differences and similarities among different climate models, thereby supporting more accurate climate predictions.

This study only predicted the first-level LUCC type in the Min-Tuo River, with a spatial resolution of 1 × 1 km, and there is room for improvement in fine characterization of LUCC. However, the characteristics of actual individual LUCC are changes in scattered small patches, which may affect the prediction of future LUCC in the river basin (Fan & Ding 2016; Gedefaw et al. 2023). Furthermore, based on runoff simulation results under different scenarios, this study quantitatively differentiated the contributions of LUCC and CC but did not further explore their impact mechanisms. It also did not quantify the sensitivity of runoff to different levels of LUCC types, which requires further investigation. Additionally, the impact of human activities on runoff changes has been increasing in recent years. However, this study did not consider the impact of decreasing cropland area on future water use, which could affect future runoff changes.

This study aims to analyze the comprehensive impact of LUCC and LUCC on runoff variations in the Min-Tuo River Basin and propose strategies for the future rational utilization of water resources. The key findings are summarized as follows:

• (1) Future CC-induced runoff response: Across four emission scenarios, a consistent trend of decreasing runoff is observed, with a notably higher runoff reduction rate under the high-carbon pathway compared with the dual-carbon pathway.

• (2) Future LUCC-induced runoff response: Relative to the baseline scenario (S0), projected LUCC scenarios in the Min-Tuo River Basin demonstrate an increasing trend in runoff, with changes of +0.05, +0.07, and +0.07%, respectively. Notably, runoff increases during flood seasons while decreasing during non-flood seasons. This pattern is anticipated to exacerbate the seasonal disparity in water availability, thus intensifying flood control and water resource management challenges.

• (3) Combined impact of future CC and LUCC on runoff response: With the exception of periods with increased runoff, the reduction in runoff under the combined influence of CC and LUCC is less pronounced compared with CC alone. The interplay between CC and LUCC mutually influences runoff dynamics, with CC exerting a greater influence. Therefore, future flood control measures should consider both CC and LUCC impacts on runoff.

• (4) Projected changes in CC are anticipated to exacerbate the uneven distribution of water resources throughout the year in the Min-Tuo River Basin, intensifying water scarcity issues. Nevertheless, LUCC may offer some mitigation potential. However, it is acknowledged that this study has limitations, including the need for improved accuracy in predicting land use changes and further investigation into the impact of human activities on runoff dynamics.

N.J.: Data curation, Formal analysis, Investigation, Methodology, Resources, Software, Supervision, Validation, Visualization, Writing – original draft. Q.N.: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – review and editing. Y.D.: Data curation, Formal analysis, Investigation, Methodology, Resources. M.W.: Formal analysis, Investigation, Methodology, Resources, Software. M.Z.: Formal analysis, Investigation, Methodology, Resources, Software. Y.W.: Data curation, Formal analysis, Investigation, Methodology, Software. H.R.: Formal analysis, Investigation, Methodology, Resources, Software. Z.Y.: Data curation, Formal analysis, Investigation, Methodology, Software.

This work was supported by the Department of Education ‘Rural Water Security’ Engineering Research Center Project, Sichuan Province, China (035Z2289).

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.