Comprehensive evaluation and scenario simulation of carrying capacity of water resources in Mu Us Sandy Land, China

Water is an essential natural resource that plays a vital role in supporting healthy human survival, regional socioeconomic development, and ecological sustainability. Water resources are widely distributed; however, the freshwater resources available for human use account for only 2% of the total. Although the threshold of freshwater resources (planetary boundary) for human use is being approached (Rockström et al. 2009), water scarcity is a globally growing problem. The water-stressed population has increased from 240 million in 1900 (Kummu et al. 2016) to 2.3 billion in 2022 (Desertification 2022). By 2050, approximately 5.7 billion people worldwide are estimated to face water shortages (Boretti & Rosa 2019). Rosa et al. (2020) found that water-constrained agricultural areas accounted for approximately 25% of global farmland. Water scarcity has become a serious challenge for human survival and sustainable development; therefore, rationally using water resources is a hot topic of concern for experts and scholars.

Water resource carrying capacity (WRCC) is a comprehensive index to evaluate whether water resources, economic society, and ecological environment develop in harmony (Wang et al. 2017b). Since the WRCC concept was proposed in the late 1980 s, research methods have been continuously enriched and improved (Zuo 2017). The analysis of the carrying capacity of water resources can be a basis for the utilization of water resources (Yang et al. 2019). Previous studies have evaluated WRCC at different scales: global (Simonovic 2002), continental (Zhao et al. 2022), national (Simonovic & Rajasekaram 2004), and regional (Yang et al. 2019). As research on WRCC at the regional scale becomes increasingly targeted, evaluation at this scale requires attention (Liao et al. 2020). The study of WRCC encompasses various areas, including natural resources, environmental, social, and economic. In terms of natural resources, WRCC levels can be used to evaluate thresholds for water resource extraction (Feng et al. 2009; Wang et al. 2018). The environmental area of WRCC commonly focuses on pollution of the water environment (Zhang et al. 2021). In social areas, WRCC is primarily used to evaluate the size of the urban population (Song et al. 2011). In the economic areas, WRCC can be used to analyze the scale of agriculture, optimize planting structure, and evaluate the production scale and water demand structure of the industry (Kang et al. 2019). Through the analysis of WRCC, future scenarios can be developed and used to inform different systems, including cities and agriculture (Feng et al. 2009).

In recent years, three main research methods have been used to evaluate WRCC, including empirical formulas, comprehensive evaluations, and system analyses. The empirical formula method includes the ecological footprint (Li et al. 2018) and index system methods (Zhang et al. 2021). Comprehensive evaluation methods include the analytic hierarchy process (AHP) (Wang et al. 2018; Yang et al. 2019), fuzzy comprehensive evaluation (Meng et al. 2009), and entropy weight (EW) (Cui et al. 2018). System analysis methods include the system dynamics (SD) (Yang et al. 2019) and pressure-state-response (Wei et al. 2014) models. Many deep-learning-based approaches, such as artificial neural networks (Shi & Zhang 2021) and the variable precondition S-type cloud algorithm (Wu et al. 2021), have recently been used to analyze WRCC, further enriching the available methods. Specially, the SD model emphasizes feedback mechanisms between factors. The SD model can also treat regional water resources and social, ecological, and economic coordination as a dialectically integrated system, and be used to simulate higher-order and nonlinear systems. Moreover, the SD model has been widely applied to optimize system models with multiple feedbacks and complex spatial and temporal changes (Yang & Wang 2021).

The Mu Us Sandy Land is located in the agro-pastoral ecological region of northern China; its environment is ecologically fragile and water-scarce (Liu et al. 2020). Many researchers have focused on the impact of sandy vegetation on groundwater (Cheng et al. 2017), exploring the optimal use of soil and water resources in the region (Zhao et al. 2021a). For example, Wang et al. (2014) proposed an integrated land and water development framework that combined irrigation management and cultivation practices. Despite widespread attention paid to water resources in Mu Us Sandy Land, the WRCC of the region is still unknown and future optimal water resource use patterns has yet to be explored. In this study, we first assess changes in the WRCC based on socioeconomic and water resources data from 2000 to 2019. We then explore the WRCC under six development scenarios: situation continuation scheme (SCS), high economic scheme (HES), industrial readjusted scheme (IRS), ecological development scheme (EDS), efficient use scheme (EUS), and comprehensive scheme (CS). Our aim is to explore development strategies that balance resources, production, society, and ecology based on the WRCC, thus providing scientific guidance for sustainable regional development.

The Mu Us Sandy Land in Inner Mongolia (107 °20′–111 °30 E, 37 °27′–39 °22′ N) is located in the transitional zone between the Ordos Plateau and Loess Plateau. The elevation ranges from 1000–1800 meters, with the terrain sloping from northwest to southeast. The region has a temperate arid and semiarid continental monsoon climate (Liu et al. 2020), with mean annual precipitation is 287.2 mm, mainly concentrated between July and September, when over 70% of the annual precipitation occurs. The average annual temperature is 6–9 °C. Prior to the 1980s, the Mu Us Sandy Land was one of China's most severely desertified areas (Figure 1). In 2000, China launched a number of ecological restoration projects to control desertification. Subsequently, desertification in the region has decreased, with only 19.1% of the area covered by sand in 2015 (Guo et al. 2021). The region has experienced rapid economic development and steady population growth since the early 2000s. For example, Gross Domestic Product has increased 34.92 times from RMB 4.40 billion in 2000 to RMB 153.660 billion in 2019, while population growth has increased over this period by 20%, from 395,500 to 475,400.

We used an SD model as they are widely applied in water resource management and water environment safety assessments (Chen & Wei 2014; Zhang et al. 2021). Concepts used to build a SD model include boundaries and hierarchy, feedback loops, level variables, and rate variables. The SD model is divided into two stages: model construction and simulation. The model construction stage is based on causal relationships and reflects the internal mechanism of the system through a system structure flowchart. In the latter simulation stage, responses in a system are simulated by adjusting sensitive parameters using different development strategies in order to determine the optimal development model (Zomorodian et al. 2018).

We divided the SD model in our study into water resources, production, domestic, and ecological subsystems. The water resource subsystem considers surface, ground, and recycled water, where recycled water mainly refers to the recycling of production and domestic wastewater. The production subsystem considers water demand for animal husbandry, agriculture, and production in secondary and tertiary industries. The domestic subsystem considers the domestic water of urban and rural residents. The ecological subsystem mainly consists of grassland, woodland, and urban green area water consumption.

There are two important variables in SD models: a level variable and a rate variable. Level variables represent the cumulative result of material or information changing with time; rate variables reflect the speed of the cumulative effect in the system. Through the analysis of the internal dynamic structure of the subsystem, the total population, the number of sheep, etc., are selected as level variables, and the birth rate, the change rate of the average annual number of sheep, etc., are selected as rate variables.

Scientific and reasonable judgment of an evaluation index weight plays a vital role in determining the accuracy and credibility of the quantitative results of the studied objectives (Yang et al. 2019). Current methods for obtaining weights of evaluation indicators are divided into two categories: subjective and objective measurements. AHP is a commonly used subjective calculation method (Yan et al. 2021), and objective calculation methods include EW (Cui et al. 2018) and fuzzy comprehensive evaluation (Gong & Jin 2009). As WRCC is a system affected by many aspects, such as the economy, society, ecology, and other elements, it is complex and uncertain. To improve the scientific and accuracy of the WRCC assessment results, we integrated subjective and objective methods using a comprehensive index calculation formula to combine the weights determined from AHP and EW methods.

We designed the following six scenarios to explore the changes of WRCC under different schemes (Table 2).

The error analysis results indicated that the absolute values of the errors were within 10% (Appendix C). However, the sensitivity of each parameter did vary widely. Although most parameters were not sensitive, individual parameters generally were more sensitive. Therefore, the SD model was found to be robust and modifying parameters was possible to investigate the six scenarios (Table 3). In other words, the simulation results of the SD model fit the collated data, and could be used for forecasting the WRCC of the Mu Us Sandy Land.

Table 3

Sensitivity analysis results for some representative variables used in the SD model

View Large

The three water demand subsystems are production, domestic, and ecological subsystems. Among them, the production subsystem accounted for the largest proportion of water consumption, with ecological and domestic subsystems ranking second and third over the past 20 years, respectively (Figure 4(b)). The carrying capacity of the production, domestic, and ecological subsystems had an overall decreasing trend (Figure 4(a)). The carrying capacity of the production subsystem showed a fluctuating decline, with a carrying capacity indicator of 0.2567 by 2019, representing a poor carrying status. The carrying capacity of the domestic subsystem declined at an even rate, with a carrying capacity of 0.0804 by 2019, representing a weak carrying status. The ecological subsystem maintained an optimal carrying capacity in 2000, but by 2019, the carrying capacity was weak, showing the largest decline and a carrying capacity indicator of 0.0802 (Figure 4(a)).

In particular, we updated the actual WRCC in 2020 based on the latest data. The results show that the WRCC is 0.4106, and the error compared with the SCS is 4.33%. Obviously, the result is less than 10%, further demonstrating that our simulations are robust and reliable.

The HES scenario results showed increasing the development speed of various sectors in society is conducive to rapid economic development. Under this development scenario, the production output value will be 2.08 times higher than that of SCS by 2030 (Figure 7(c)), dramatically improving social productivity. However, the ratio of water supply to water demand will decrease to 0.6698 times that of SCS (Figure 7(a)), exacerbating water stress. The volume of COD emissions will reach its highest level; thus, the water pollution problem will become severe, with COD rising to 2.01 times the SCS scenario (Figure 7(b)).

The results of the IRS scenario showed that the ratio of water supply to water demand will be lower and higher than that of the SCS in the short- and long-term, respectively (Figure 7(a)). Similarly, the production value in the IRS scenario will begin to exceed that of SCS in 2027, increasing to 1.21 times that of SCS of 2030 and continuing to rise, which may contribute to the economy to some extent (Figure 7(c)). The volume of COD emissions will be lower than HES in the long term and 1.5777 times higher than SCS (Figure 7(b)).

The EDS scenario results showed the ratio of water supply to water demand will be lower than the SCS (Figure 7(a)), indicating that the demand for water increases when the ecological area reaches a certain size. This scenario suggested that implementing environmental protection policies may slow COD emissions to a certain extent (Figure 7(b)). The EDS scenario results suggest total ecological water consumption will substantially increase 1.61 times that of SCS by 2030 (Figure 7(d)).

The results for EUS showed that this scenario will remarkably improve the ratio of water supply to water demand by only saving water. However, with the development of the socio-economy, the demand for water resources will increase, which will reduce the optimization effect (Figure 7(a)). The impact of EUS on environmental protection is pronounced. It will beneficially reduce COD emissions (Figure 7(b)) and effectively reduce the use of ecological water resources, being approximately 20.01% lower than the SCS scenario (Figure 7(d)).

The CS scenario results showed that this scenario will optimize the ratio of water supply to water demand, and a balance between supply and demand can be achieved in the long term (Figure 7(a)). The volume of COD emissions will be much lower than that of SCS, being approximately 0.53 times lower than the SCS scenario by 2030 (Figure 7(b)). The production output will be higher than that of SCS by 2030 (Figure 7(c)). The ecological water demand will be stabilized by controlling the ecological water quota while maintaining the current ecological status, reducing water consumption by 35.53% compared with the SCS scenario (Figure 7(d)).

The WRCC is often used to evaluate the role of water resources in supporting social life, economic development, and ecological construction within a specific time and region by providing the basis for implementing sustainable development strategies (Liao et al. 2020). Previous studies quantifying WRCC show wide regional heterogeneity, categorized poor (Yang & Wang 2021), general (Yang et al. 2019), and positive (Peng et al. 2021) states. Even in humid and semi-humid regions where water resources are abundant, water demand can be higher than water supply, and therefore, the WRCC is weak (Feng et al. 2009; Yang & Wang 2021). Our results suggest the WRCC of the Mu Us Sandy Land has declined from a positive to general carrying capacity over the period from 2000 to 2019 (Figure 5). If the current trend continues, WRCC will deteriorate further and reach a poor status by 2024 (Figure 6). These findings are similar to those of previous studies on the Loess Plateau (Gong & Jin 2009), Guanzhong Plain (Yang et al. 2019), Tarim River Basin (Meng et al. 2009), Bosten Lake Basin (Wang et al. 2018), and Weihe River Basin (Yan et al. 2021) in arid and semiarid regions in China. Therefore, the WRCC status of arid and semiarid regions in China is generally poor and requires further attention.

Many researchers have found that social and economic factors, as opposed to natural factors, are often the leading causes of the decline in regional WRCC (Zhao et al. 2021b). Based on our analysis of subsystems, rapid social and economic development appears to be the main factor causing the overall decline in WRCC in the Mu Us Sandy Land over the 20-year period. The main city in Mu Us Sandy Land is Ordos, which is a crucial energy-rich area and heavy-industry base in China. Coal is the primary energy source in the region, and each ton of mined coal requires 0.18 m³ of water. Since 2000, the coal production in the area has increased from 10.18 million tons to 293.13 million tons, a 28.779-fold increase (Appendix D. Figure 1). In addition, the main industry in the region revolves around the coal-chemical industry, with an industrial water consumption rate of approximately 40.02 m³/RMB 1 million. The production subsystem consumes a tremendous amount of water (accounting for approximately 60% or more of the total water consumption) and is the main use of the water resources (Figure 4(b)). Consequently, rapid economic development has led to the deterioration in regional WRCC. Furthermore, the number of people in the region has increased annually over the 20-year period, with the urbanization rate continuing to increase (Appendix D. Figure 2). The increase in population has led to a gradual deterioration in the WRCC of the domestic subsystem. Social development has also led to a decline in the regional WRCC.

China has implemented numerous ecological restoration projects to restore degraded ecosystems (Liu et al. 2020). Ecological restoration projects help prevent wind and sand pollution and improve the ecological environment (Su et al. 2022). However, recent studies have begun to increasingly pay attention to the impact of ecological restoration on water resources, arguing that ecological restoration requires large amounts of water consumption, which affects soil moisture and exacerbates regional drought (Xiao et al. 2020). Ecological restoration increases vegetation growth and has been shown to lead to an ecosystem water imbalance in the Loess Plateau and Kubuqi Sands (Li et al. 2016; Chen et al. 2022). Our results suggested the WRCC of ecological subsystems decreased from excellent in 2000 to weak by 2019 (Figure 4(a)). Future scenario simulations showed the EDS scenario could help to improve the WRCC in the short term but might increase pressure on WRCC at longer time scales (Figure 6). This is likely due to ecological restoration of Mu Us Sandy Land in recent years has led to increases in the vegetated area and in the ecological water demand of plants, which exceeds the regional ecological threshold. A previous study by Zheng et al. (2020) also suggested the ecological restoration of the Mu Us Sandy Land could exacerbate water scarcity in the region. Therefore, in addition to social and economic factors, ecological restoration might contribute to the deterioration in WRCC. Considering the possible opportunity costs of ecological restoration in arid and semiarid areas, we recommend ecological restoration should be conducted under the WRCC threshold to avoid the overexploitation of groundwater resources (Lu et al. 2018). We also advocate the promotion of sand-tolerant plants with low water consumption and high survival rates, ensuring a balance between water resources and regenerating vegetation (Zhao et al. 2021a).

After quantitatively evaluating regional WRCC, previous studies have explored future strategies for WRCC under different local development scenarios based on the current WRCC. In arid and semiarid regions, optimal development patterns include scenarios such as increasing water resource supply (Wang et al. 2018), optimizing agricultural water consumption (Yang et al. 2019), implementing water conservation measures (Meng et al. 2009), and optimizing and upgrading industrial structures (Yang & Wang 2021). We found that under the CS scenario, the WRCC of the Mu Us Sandy Land remains in a general status, showing a stable upward trend (Figure 6). The ratio of water supply to demand is good (Figure 7(a)), the production value develops more rapidly (Figure 7(c)), and the volume of COD emissions (Figure 7(b)) and ecological water consumption (Figure 7(d)) are low. This strategy combines the advantages of the IRS and EUS scenarios and is more effective at enhancing WRCC than a single strategy. Compared with the single-strategy design, EUS is much better at improving WRCC on short time scales and IRS on long time scales (Figure 6). The former confirms the effectiveness of water conservation measures. However, the rapid increase in water demand cannot be balanced by water-saving policies and technologies in the long term because of the further increase in water demand as society develops. The latter reflects the current industrial structure of the Mu Us Sandy Land will likely require further adjustment and optimization. People should actively seek industrial transformation in favor of the tertiary sector, which consumes less water. Given the above findings, a comprehensive strategy based on water conservation and an industrial adjustment plan is the most appropriate development strategy for the region (Figure 6).

Combining the comprehensive index evaluation and the SD model is a widely used method to scientifically and objectively evaluate regional WRCC (Yang et al. 2019). Although we used these two methods, our study still has limitations. First, the Ordos Statistical Yearbook and the Ordos Water Resources Bulletin were published from the year 2000 only. This means we were not able to utilize earlier historical data, limiting the model construction of WRCC for the Mu Us Sandy Land to only 20 years. Notably, some of the ecological restoration projects in the area, such as the restoration of farmland to forest and grass, began around 2000. Therefore, obtaining data before 2000 would have been useful in gaining a longer-term understanding of WRCC dynamics in the region. Second, studies dealing with higher-order, nonlinear, and spatially and temporally complex system problems have commonly used the SD model to describe and simulate water resource systems for WRCC analysis (Chen & Wei 2014). However, the SD model has certain constraints, mainly owing to the complexity and uncertainty of the real-world situation. Although our study involved multiple elements of natural resources, production, society, and ecology, the complex system we constructed is still a simplified model compared to the actual, real-world system. Therefore, in future research, we recommended building further integrated models that combine SD with other methods, such as dynamic neural networks (Zomorodian et al. 2018), to assess the carrying capacity of local water resources more accurately.

In this study, we considered the Mu Us Sandy Land as the research area and used the comprehensive index evaluation and SD model trained on natural-society-economic data for the Mu Us Sandy Land in Inner Mongolia to analyze WRCC from 2000 to 2019 and under six future scenarios out to 2030. Our results found the WRCC has been declining in the region since 2000, and the carrying state has changed from general (0.6786) to poor (0.4235). We found an integrated strategy of industrial restructuring and water conservation measures is the most appropriate development strategy for this area moving forward. By 2030, WRCC will increase by 1.84 times compared with the situation continuation scheme under this strategy. Our findings provide theoretical guidance on the WRCC situation of the Mu Us Sandy Land and can be used to guide policy implementation to optimize water use. This complements the weaknesses in assessing the WRCC of sandy land, thereby providing an essential contribution to the achievement of rational water resource use and sustainable development in arid and semiarid regions. However, the WRCC is the result of the long-term comprehensive effect on nature and society. This paper focuses on the results of the disturbance of human activities to the water resources system. In the future, it is necessary to combine integrated models and establish a WRCC early warning system platform.

This work was supported by the Major program of Inner Mongolia (2021ZD0008), Key Science and Technology Program of Inner Mongolia (2019ZD007), the National Natural Science Foundation of China (32071582) and the Grassland Talents Program of Inner Mongolia (CYYC9013).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.