Measurement and spatiotemporal differentiation of green water resources utilization efficiency in the Yellow River Basin

With the growth of population size and socio-economic conditions, various types of water consumption have sharply increased. Research shows that China's water resources utilization rate is only around 40%, while developed countries reach around 80% (Liu et al. 2023; Yoshe 2023). With the continuous increase in water demand and the intensification of climate change, the pressure on water resources utilization continues to rise. In addition, the extensive economic growth model has also brought serious water pollution problems to various regions, and the discrepancy between the supply and demand of water resources has emerged as a pivotal global concern. As an important economic zone, the Yellow River occupies an important position in the social and economic growth pattern of China. Promoting high-quality circular utilization of water resources in the Yellow River Basin (YRB) is an inherent necessity for promoting regional economic transformation and development and can also provide basic industries and energy security for national economic development. However, frequent water disasters, significant flood risks, fragile ecological environment, severe water resources security situation, and the need to improve development quality are prominent problems in the YRB (Yang et al. 2022). As the economy advances and population grows, the contradiction between the increasing requirement for water resources in the basin and the limited supply of water resources is becoming increasingly prominent. In the last few years, the interruption of water flow in the middle reach (MR) and lower reach (LR) of the YRB has been increasing, and the reasonable growth and utilization of its water resources have become increasingly difficult. This has led to the current shortage of water resources and environmental pollution in the YRB, which seriously restricts economic development. There are significant differences in the water resources utilization efficiency (WRUE) within the basin, and there is no coordination between resource utilization efficiency (UE) and economic development (Rong et al. 2021; Hu et al. 2023). Therefore, improving the green utilization efficiency (GE) of water resources in the YRB is an objective requirement to solve the problem of local water resources and facilitate sustainable development. Under the background of sustainable development, the measurement of WRUE has new economic and environmental significance and can provide new solutions for regional economic development.

Under the background of climate change, the issue of water shortage is becoming more grave, and enhancing the WRUE represents a crucial strategy for alleviating this problem. Li et al. used the super-directional distance function (Super-DDF) model to measure WRUE in the YRB. From the perspective of the whole basin, the digital economy raised WRUE. From a regional angle, the digital economy had a significant promoting effect on WRUE in the upper reach (UR) and MR of the Yellow River, and a restraining effect in the LR of the Yellow River, but the results were not significant (Li & Liang 2022). Zhang et al. used panel data from 30 provinces in China from 2006 to 2019 to study the impact of environmental regulation on green water efficiency. The results indicated that the WRUE in China was relatively low, and there were significant differences among provinces. Environmental regulation has a positive impact on the efficiency of national green water use. In view of regional variations, it can be seen that environmental regulation exerts a considerable positive impact on the efficiency of green water use in both the eastern and western regions (Zhang & Qiu 2022). Mu et al. used a multi-stage difference model to evaluate the impact of water resources taxes on WRUE. The results indicated that water resources policies could effectively improve the WRUE by optimizing resource allocation. Besides, spatial heterogeneity analysis denoted that the water resources tax policy had a stronger effect on raising the WRUE in the eastern region of China than in the central and western regions (Mu et al. 2022).

Data envelopment analysis (DEA) employs linear programming to assess the comparative efficacy of comparable units of the same type. It is widely used in solving various efficiency problems. Liang et al. combined DEA and analytical hierarchy process (AHP) methods to propose the WRUE index (WEI) to assess the WRUE of various regions in the Tumen River Basin. Quantitative analysis was conducted on domestic, agricultural, and industrial water UE in the Three Gorges Reservoir area using the DEA-Charnes, Cooper, and Rhodes (CCR) model. The findings indicated that from 2014 to 2019, the WEI values of most areas along the ‘the Belt and Road’ showed an upward trend (Liang et al. 2022). Wang et al. measured the WRUE of 30 provinces in China using a DEA model and then analyzed the influence of environmental regulations on regional WRUE in China using the generalized method of moments (GMM) model. The outcomes denoted that the overall WRUE was low, and regional WRUE was unbalanced. The areas with higher efficiency were concentrated in the east, while the areas with lower efficiency were concentrated in the west. Weak environmental regulation intensity was not conducive to improving WRUE (Wang & Wang 2021). Fu et al. used Mann–Kendall and multiple linear regression analysis to study the regional and temporal dynamic changes in WRUE in China. According to research results, the annual average water use efficiency in China from 2000 to 2017 was 0.998 gC/mm·m². The results of this study may help China select appropriate regional vegetation and sustainably manage local water resources (Fu et al. 2024). Shi et al. investigated the impact of China's new urbanization development on industrial water efficiency. On this basis, a spatial econometric model and a panel threshold model were constructed to explore the spatial spillover effects and threshold characteristics of new urbanization development on industrial water use efficiency. The results indicate that the development of new urbanization presents a significant positive spatial correlation and has a positive spillover effect on the improvement of industrial water use efficiency (Shi et al. 2023). Liang et al. assessed the industrial water UE of more than 200 cities in the Chinese Mainland from 2012 to 2016 under the slack-based measure (SBM) model and analyzed the industrial water use efficiency of China from the urban level. Subsequently, the Tobit regression model was employed to ascertain the pivotal determinants of industrial water UE from the vantage points of natural, social, municipal, and industrial structural factors. The outcomes indicated that per capita gross domestic product (GDP), the total length of drainage pipelines, and the presence in coastal areas had a considerable positive influence on WRUE, whereas the proportion of the secondary industry to GDP had a considerable negative impact (Liang & Zhou 2022).

In summary, extensive research has been conducted on the WRUE, including the impact of the digital economy on WRUE in the YRB, the impact of environmental regulation on green water use efficiency, and the impact of water resource taxes on WRUE. Various factors have been analyzed for the WRUE in different regions. However, these influencing factors are all singular and isolated from other factors during the analysis process. Previous studies have failed to effectively consider multiple influencing indicators when analyzing WRUE, and the analysis results obtained need further optimization. To address this issue, the study first constructs a measurement index for WRUE and then combines DEA and SBM to construct a calculation model for WRUE based on DEA-SBM. It comprehensively considers the impact of multiple factors on UE, while also taking into account the influence of expected and unexpected output factors on WRUE, to achieve a more accurate and effective analysis of the UE of green water resources in the YRB, and facilitate the sustainable development of regional economic circulation.

The study designed a method for measuring the WRUE in the YRB. First, the first section constructed measurement indicators for WRUE and comprehensively analyzed the input and output of WRUE. Secondly, in the second section, a measurement model for WRUE was constructed with the DEA model, and an SBM model with non-radial and non-angle was introduced to comprehensively consider the unexpected outputs in the output factors.

To better analyze the efficiency of green water resource utilization in the YRB, this study first collected and organized relevant sample data. This study mainly focuses on nine provinces that flow through the YRB. Given that the analysis object, namely the WRUE in the YRB, has certain spatiotemporal characteristics, this study adopts a systematic sampling method to ensure that the selected samples can cover the spatiotemporal variation characteristics within the region. This study uses a systematic sampling method to analyze relevant data from 2011 to 2022 as samples. First, the collected data are processed to correct outliers and missing values. Then, the data are classified based on characteristics such as time and region. Finally, all data are normalized to facilitate subsequent analysis. The relevant indicator data are all from the ‘China Environmental Statistical Yearbook’, ‘China Statistical Yearbook’, and the ‘Water Resources Bulletin’ of various provinces and cities in the YRB (Li et al. 2022).

In formula (1), T means the total water footprint, T₁ represents the polluted water footprint, T₂ represents the water footprint of the domestic and ecological environment, T₃ represents the water footprint of agricultural, forestry, and animal husbandry products, and T₄ represents the water footprint of industrial products. In human resources indicators, relevant practitioners are applied to measure the actual labor input in the production process. Based on the health, social services, and education-related data released by the National Bureau of Statistics over the years, this study uses the society development index (SDI) to measure the social development and improvement of people's livelihoods in the YRB region (Li et al. 2023). The indicator system is denoted in Table 1.

In formula (6), Unex represents the total amount of undesired GW footprint, B₁ represents industrial GW, B₂ represents domestic GW, and B₃ represents agricultural GW footprint. Based on the above indicator analysis, the selection of input and output variables is based on principles such as scientific rigor, systematicity, practical needs, and feasibility, while also referring to the indicator system constructed in existing research. After a comprehensive analysis, the following indicator system is obtained. The indicator system for measuring the green WRUE in the YRB is shown in Table 2 (Zhe & Tonghui 2023).

According to the analysis results, basic indicators for measuring WRUE are obtained, which comprehensively consider expected and unexpected output factors under input factors, thus comprehensively measuring the effective utilization of water resources.

In formula (8), is the constraint condition added when the return to scale changes, which is the front line of CRS. The frontier forms under the above two conditions are shown in Figure 3.

In formula (12), and represent the number of DMUs invested. g, b represent the number of variables, and S^g and S^b, respectively, represent slack variables. After considering unexpected outputs comprehensively, the computational effectiveness of WRUE can be further optimized.

Based on relevant data analysis, the study first constructs a measurement index system for WRUE. Based on this indicator system, research constructs a DAE-SBM model for specific analysis and calculation of WRUE. The limitation of this method design lies in simplifying complex water resource systems, ignoring the impact of some less prominent factors on WRUE, and only considering key factors.

The study first introduced the sources of experimental data in the first section and conducted a significance test on the constructed indicator parameters, verifying the effectiveness of the selected indicators in the study. Then, using the research-designed DAE-SBM model, the UE and spatiotemporal differentiation of water resources in the YRB were analyzed, and the characteristics of the UE and spatiotemporal distribution of green water resources in the YRB were obtained.

To analyze the WRUE in the YRB, a WRUE measurement method based on an improved DEA model was designed on the basis of constructing UE measurement indicators. MaxDea8.0 was used to measure the WRUE in the YRB. The WRUE of provinces flowing through the YRB were analyzed, including Qinghai, Sichuan and other provinces. Using the period of 2011–2022 as the research time range, all input and output indicator data were from the China Environmental Statistical Yearbook and the China Statistical Yearbook (Mingyue et al. 2023). First, to explore the impact of each variable on efficiency analysis, sensitivity analysis was conducted in this study. The analysis results are shown in Table 3. From Table 3, among the input factors, the public awareness of water conservation and the completeness of the management system had little impact on the WRUE, while water resource and economic benefits had a significant impact on the UE.

The green WRUE in each province during this time period is shown in Table 4. Based on the data shown in Table 3, the average WRUE of Sichuan and Shaanxi was below 0.6, indicating low UE. In Qinghai, Gansu, and Henan provinces, some years had UE higher than 1, while most years had UE lower than 1. The overall WRUE of each province ranged from 0.3 to 1.2, and its UE could be further optimized. The WRUE in Inner Mongolia, Shanxi, and Shandong was all above 1, indicating an effective utilization state.

The Malmquist index is an indicator utilized to evaluate the dynamic changes in DMU total factor productivity (TFP). The study used it to evaluate the changes in the TFP index of WRUE in the YRB. According to Table 5, the average TFP was 1.078, which showed a decreasing trend in 2014 and 2015. The overall WRUE of the nine provinces in the YRB denoted an upward trend. The technology UE was relatively low, with an average annual rate of less than 1 in 2014, 2016, 2017, 2018, 2021, and 2022. Technological changes decreased significantly in 2012 and 2013, while also showing a declining trend in 2016. The efficiency of scale declined significantly in 2014 and 2015. This indicated that the water resources technology of various provinces in the YRB needs to be further improved, and their scale production can also be further enhanced. The optimization of technology UE can be carried out from two aspects: PTE and SE.

To further analyze the spatiotemporal differentiation in WRUE in the UR, water resources, and LR of the YRB, a detailed analysis was conducted on the internal differences in WRUE. The contribution rates of each part to the overall UE were analyzed separately, and the results are denoted in Table 6. Overall, the contribution rate of green efficiency of water resources in the UR, MR, and LR of the Yellow River was ranked from high to low in the LR, water resources, and UR. Compared to the MR and LR, the UR was significantly lower. The contribution rate range of the UR was 7.94%, while the contribution rate range of the MR was 16.9% and the LR contribution rate range was 26.51%. The contribution rates of each region were relatively stable in the time dimension.

To further validate the performance of DEA-SBM, it was compared with other models, including the CCR model, AHP, and fuzzy comprehensive evaluation (FCE) method. The comparative analysis between DEA-SBM and other models is shown in Table 7. In Table 7, the F1 score of the DEA-SBM model reached 0.93, with an accuracy of 0.92, a recall rate of 0.93, and an Areas Under the Curve (AUC) as high as 0.95, indicating that the model has high accuracy and reliability in efficiency analysis tasks. In contrast, although the CCR model, AHP, and FCE model also performed well, they were slightly inferior to the DEA-SBM model in all indicators. This further proved the superior performance of the research method.

Finally, the significance of the obtained analysis results was analyzed, and the results are shown in Table 8. In the table, there is a clear correlation between each result and the research object, which has varying degrees of impact on WRUE and can effectively reflect the reliability of the results.

To better analyze the WRUE in the YRB, the study first constructed a corresponding indicator system. On this basis, researchers designed a WRUE measurement model based on DEA and optimized it using the SBM model. The experiment outcomes denoted that the average WRUE in Sichuan, and Shaanxi in the YRB was less than 0.6, indicating low UE. This may be related to the overall allocation, utilization, and investment ratio of water resources in the region. If the input ratio is high but the output efficiency is low, it directly reduces the UE of regional green water resources. In Qinghai, Gansu, and Henan provinces, some years had UE higher than 1, while most years had UE lower than 1. The overall WRUE of each province ranged from 0.3 to 1.2. From the perspective of spatial differences, water resources in different regions contributed to the utilization of green water resources in the YRB to varying degrees. Specifically, the contribution rate of green WRUE in the UR, water resources, and LR was ranked from high to low in the LR, MR, and UR. In contrast to the MR and LR, the UR was significantly lower. The contribution rate of the UR was extremely poor at 7.94%, while the contribution rate of the MR was extremely poor at 16.9%, and the LR contribution rate was extremely poor at 26.51%. The contribution rates of each region were relatively stable in the time dimension. This may be because the overall economic development level of the upstream and downstream regions is relatively higher, which allows the invested resources to be better converted into economic benefits. Similarly, the economic development level in the MR and UR of the region was slightly lower than that in the LR, which led to lower efficiency in the transformation of economic benefits. Therefore, under the same investment ratio, LR regions might achieve better efficiency in utilizing green resources, resulting in a higher contribution rate. The scale, structure, and efficiency of economic development in the YRB were basically consistent with the trend of changes in WRUE, showing an upward trend. The UE of green water resources in the YRB met the economic development needs of the region and could achieve a balance between the growing economic development needs and green environmental development to a certain extent.

By analyzing the UE of green water resources in the YRB, the dynamic changes in the UE of green water resources in the basin over the years can be effectively analyzed, providing a reference for subsequent water conservation and environmental protection work. Secondly, the research results can provide effective suggestions for the future utilization of green water resources in the basin, further improving UE and promoting regional sustainable development. In addition, the research results can effectively provide useful guidance for the sustainable utilization of water resources in the YRB.

There are still shortcomings in the research. This study comprehensively considers the measurement indicators of the social development index, but the exploration of how this index specifically affects the efficiency of green water resource utilization is relatively insufficient. Second, how to further improve the UE of green water resources is also a direction that needs to be further explored in the future. Potential sources of bias in research include data collection and processing, as well as the use of stakeholders, which may have an impact on the accuracy of analysis results.

There are relatively significant differences in economic growth among different reaches in the YRB, and there are also significant differences in the demand for water resources. To explore the UE of green water resources in the YRB, a measurement method for green WRUE based on DEA-SBM was designed. Moreover, relevant data from 2011 to 2022 were used for statistical analysis. The experimental results showed that there were significant temporal and spatial differentiations in the WRUE in the YRB. From a spatial perspective, the UE of LR regions was higher than that of MR and UR regions. The contribution rates of green efficiency of water resources in the basin, from high to low, were LR, MR, and UR. In terms of time, from 2011 to 2022, the UE showed an increasing trend. In addition, the scale, structure, and efficiency of economic development in the YRB were basically consistent with the trend of changes in WRUE, showing an overall upward trend. However, there are still shortcomings in the research. In the analysis of the indicator system, more unexpected output factors can be taken into account to further optimize the calculation accuracy of WRUE. In addition, future research can further explore the specific mechanisms for improving the efficiency of green water resource utilization.

This research was supported by Henan Provincial Key R&D and Extension Special Project 2023: Study on the Measurement of Water Resources Utilization Efficiency and Spatial and Temporal Patterns in the Yellow River Basin under Environmental Constraints. (232400411180).

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

The authors declare there is no conflict.