A multi-method integrated simulation system for water resources allocation

The emergence of the water crisis makes people begin to pay attention to water resources management. The shortage of water resources highlights the necessity of sustainable development and the optimal allocation of water resources. To comprehensively strengthen the conservation, protection, and utilization efficiency of water resources and accelerate the construction of the water-saving society, China is implementing the most stringent water resources management system. Water resources allocation is a method and a significant prerequisite for achieving total quantity control of water resources utilization, as well as a significant way to improve the efficiency of water resources utilization (Wen et al. 2019; Yu et al. 2020).

The research on water resources allocation began at the beginning of this century. The goal of water resources allocation in the initial stage is to maximize economic benefits. The allocation of water resources in different regions, departments, or seasons is realized by constructing economic benefit optimization models (Reca et al. 2001; Quba'a et al. 2002). These researches have achieved good application results. However, these researches have a single objective, and the allocation process does not take factors such as fairness, efficiency, and sustainability into account. Therefore, some scholars have researched the allocation of environmental water resources from the perspective of ecology (Voogt et al. 2000). In the same period, some scholars investigated the relationship between laws, policies of water resources allocation, and basin hydrology from the perspective of water rights and water market methods. And they built models through flow, water shortage degree, consumption rights, and irrigation efficiency to realize the allocation of water resources in different regions (Green & Hamilton 2000).

Research on water resources allocation started relatively late in China. Scholars mainly carried out qualitative research and theoretical discussion on the principles, objectives, and ideas of allocation, and gradually formed the allocation practice with fairness as the main principle, and the classification weight method (Yang et al. 2006), analytic hierarchy process (Li & Niu 2021) and quota prediction method (Chen et al. 2012) as the main methods. However, AHP is greatly influenced by subjective factors. The Classification weight method reflects absolute fairness, without distinguishing the relative importance of indicators. The quota prediction method is greatly influenced by data length and prediction accuracy. Therefore, based on analyzing the limitations of major allocation methods in China, Cheng (2012) proposed the application of a multi-factor comprehensive analysis model and multi-level semi-structural multi-objective fuzzy optimization model to allocate water resources in the basin, to solve the water resources allocation problem in southern China. However, the determination of allocation weights is still subjective in this study, and the randomness and uncertainty of water resources interannually are not considered. Therefore, the allocation schemes lack dynamic adaptability. In recent years, game theory and bankruptcy theory have been gradually applied to the study of water resources allocation. These two theories can better take into account the fairness, efficiency, and sustainability of allocation, and at the same time consider the socio-economic and environmental differences of allocation objects, to balance the conflicts of interests of all parties (Fu et al. 2018; Liang et al. 2019). However, adaptation to the dynamic characteristics of water resources is still a scientific problem that researchers need to solve.

Therefore, it is not difficult to summarize the main problems of water resources allocation in China. 1. The allocation theory is immature and lacks objective quantitative research (Cheng 2012); 2. The allocation method is relatively single and has limitations (Cheng 2012); 3. The traditional water resources allocation scheme is static and single. It is difficult to adapt to the dynamic change of the amounts of allocable water resources, and the applicability of the allocation scheme is not strong. 4 From the perspective of decision-making requirements, a single scheme does not have the selection space for decision-making as multiple schemes (Ge et al. 2013). Moreover, the common allocation practice is that water utilization objects report their annual water utilization plans, and the water resources management departments make empirical adjustments following the principle of increasing allocatable water resources when there is more water and decreasing allocatable water resources when there is less water. This allocation practice is not conducive to improving the utilization efficiency of water resources and conceals the problems existing in the current water resources allocation.

To address the above issues, this study relies on a comprehensive knowledge visualization integration platform developed by service-oriented architecture and applies visual knowledge map technology to build a simulation environment for water resources allocation. Component technology is applied to structurally encapsulate multi-type allocation models. Web Service technology is applied to invoke and customize component services. The simulation functions are achieved through the coupling of the knowledge map with the customized components and the flow of data between the knowledge map nodes. Therefore, a multi-method integrated simulation system for water resources allocation can be developed. The system solves the problem of lacking allocation methods through multi-model integration applications and solves the problems of static traditional allocation schemes unable to adapt to changes in water resources conditions through the dynamic simulation of allocation scheme sets. The multi-scheme dynamic simulation also provides good decision support for water resources allocation.

Considering that water resources allocation involves multiple interests, this research conceals the specific information of the real study area. The study area consists of one mainstream and two main tributaries. The basin covers an area of 67,108 square kilometers. And the water utilization objects in the basin include seven administrative objects and five irrigation district objects, such as city A, city B, city C, city D, city E, city F, city G, irrigation district A, irrigation district B, irrigation district C, irrigation district D, and irrigation district E. The characteristics of the research area are shown in Figure 1.

To achieve water resources allocation calculation, it is essential to collect correlative data. In this research, two types of data were collected. They are indicator data and water utilization data. To construct the subjective weighting model and objective weighting model, the research collected eight kinds of fairness indicators, five kinds of efficiency indicators, and six kinds of sustainability indicators. To construct the statistical analysis model and multi-objective optimization model, the research collected seven administrative allocation objects' and five irrigation district objects' long series of water utilization data. All of these data are from the statistical yearbook published by the government (Shaanxi Provincial Bureau of Statistics 2019).

Water resources allocation involves the social and economic development of parties. Fairness, efficiency, and sustainability are the basic principles of water resources allocation. Therefore, industrial water consumption, agricultural water consumption, domestic water consumption, populations, and gross domestic product can be used to measure allocation criteria of fairness. Per capita water consumption, output value per cubic meter of water, and per capita gross domestic product can be used to measure allocation criteria of efficiency. Natural population growth rate, forest coverage rate, industrial wastewater treatment volume, and daily wastewater treatment capacity can be used to describe allocation criteria of sustainability. Meanwhile, the selection of indicators refers to Li B's (Li et al. 2018) research. The hierarchical structure of water resources allocation in administrative districts is shown in Table 1.

Allocation criteria outlined above also need to be considered in water resources allocation in irrigation districts. Grain yield, irrigation water consumption, and area of effective irrigation are considered as the common indicators to describe allocation criteria of fairness. Grain yield per unit area and utilization coefficient of irrigation water can be applied to measure the efficiency of water-using in irrigation districts. The integrity of irrigation facilities and water quality of irrigation return flow are the main indicators to measure the sustainability of irrigation districts' development. The hierarchical structure of water resources allocation in irrigation districts is shown in Table 2.

In the construction process of this system, four types of allocation models will be integrated, which are the subjective weighting model, the objective weighting model, the statistical analysis model, and the multi-objective optimization model. The following text briefly introduces the four models.

(1) Subjective weighting model (SW model). The analytic hierarchy process (AHP) is a decision analysis method of hierarchical weight (Feng et al. 2008; Sun et al. 2020), which was proposed by American operations researcher Satie in the 1970s. Water resources allocation needs to take into account the development interests of different allocation objects (Roa-García 2014), and the analytic hierarchy process provides methodological support for balancing the interests of all objects in the allocation process. The subjective weighting model of water resources allocation is constructed as follows: 1. The construction of the hierarchical structure of water resources allocation; 2. The construction of judgments matrix; 3. Hierarchical single ranking of indicators, and consistency test. 4. Hierarchical total ranking and consistency test; 5. Calculation of indicator weights. According to the weight of the criterion layer relative to the target layer and the weight of the indicator layer relative to the criterion layer, the influence weight of each indicator on the target is calculated from top to bottom; 6. The calculation of the comprehensive allocation weight of each object. Based on the proportional allocation of the same indicator among the different objects, the weights of the indicator are allocated to each object, and then the comprehensive weights of all indicators allocated to the same object are obtained.

(4) Multi-objective optimization model (MOO model). Based on the demand for social and economic development, the water resources allocation objectives of administrative units are as follows: 1. Maximizing comprehensive economic benefits (Tian 2020). In other words, each administrative unit maximizes the benefits of living, industry, agriculture, and ecological water outside the river. 2. Minimizing the cost of water utilization. That is, the total cost of water utilization of all administrative units is the lowest. The model constraints are: domestic water utilization constraint, off-river ecological water utilization constraint, allocable water resources constraint, and non-negative variable constraint. Extraction of water resources allocation objectives of irrigation districts (Wu et al. 2019) is maximizing irrigation benefits (Liu 2020) and minimizing the cost of irrigation water. The model constraints are allocable water resources constraint, non-negative variable constraint.

According to the two kinds of weighted model of water resources allocation, the respective weight of twelve kinds of indicators can be calculated firstly. Considering that the indicator weights change with different data and indicator selection results, the indicator weights based on current data and all indicators are shown in Table 3.

Furthermore, the respective weight of seven administrative objects and five irrigation objects can be calculated by four kinds of allocation models. The weights of allocation objects are shown in Table 4. It is worth noting that the allocation weight given here is only one of the cases of dynamic weight.

It should be pointed out that the summation of allocation weights of all allocation objects is 1 in the statistical analysis model. In the rest of the models, the summation of allocation weights of all administrative districts is 1 and of all irrigation districts is 1.

The integration platform is developed based on service-oriented architecture, which can provide the underlying support for the development of a water resources allocation simulation system. Relying on this platform, knowledge map technology (Xie & Luo 2010) is applied to build the business process and system interface for water resources allocation. Component technology (Qiu 2003) is used to standardize the encapsulation of water resources allocation models. Web Service technology (Lv 2010) is applied to achieve dynamic simulation of water resources allocation. The methods of building a water resources allocation simulation system can be divided into the construction of the simulation environment and the integration of simulation models.

Draw a knowledge map of water resources allocation. The integration platform comes with knowledge map development tools such as the point element drawing tool, line element drawing tool, background, and font setting tool. Through the application of a combination of point elements, line elements, and surface elements, the knowledge map for the allocation business is drawn. And a business environment to support water resources allocation simulation is built. The schematic diagram of the knowledge map is shown in Figure 2.

(1) Componentized encapsulation of the water resources allocation models. 1. Developing water resources allocation model components. Based on eclipse development software, the subjective weighting model, objective weighting model, statistical analysis model, and multi-objective optimization model are packaged component-based by using the java programming language. 2. Packaging of the water resources allocation models components package. After developing all the business components, the Axis2 Service Archiver plug-in is used to complete the encapsulation of the model components package. 3. Upload the water resources allocation models' components package. Login to the Axis2 website to upload the file with the .aar suffix to the server where the system will be deployed. 4. Service registration. Login to the UDDI Centre to register the web service and point the service address to the location where the allocation models' components package will be located after upload.

(2) The coupling of the water resources allocation knowledge map with the business components. When the models' components package upload and complete the service registration, the service is invoked by Web Service technology. The invoked Web service is customized, and the customized components are coupled with the corresponding knowledge map nodes. Business functions based on the coupling of allocation models components and the knowledge map are achieved through the flow of data between knowledge map nodes. A schematic diagram of the coupling of the knowledge map to business components is shown in Figure 3.

The simulation system is built by applying the system construction methods and water resources allocation models described in the previous section, and the developed system interface is shown in Figure 4.

To achieve the visual generation of water resources allocation schemes, it is necessary to set the amounts of allocatable water resources in the system. On this basis, the subjective weighting allocation model and subjective weighting indicators are selected. Based on the selected results, the system passes the allocatable water resources, model selection, and weighting indicators to the subjective weighting model components by the data stream. The calculation of water resources allocation is achieved by calling the parameter passed from the components. The calculated allocation scheme is generated visually in the form of tables and histograms. The allocation scheme based on the subjective weighting model is shown in Figure 5.

Similarly, the process of applying the objective weighting model to generate an allocation scheme is as follows: Firstly, setting the amounts of allocable water resources. Secondly, the objective weighting allocation model is selected. On this basis, the indicators for objective weighting are selected. According to the results of the amounts of allocable water resources, the selection of allocation model, and the selection of objective weighting indicators, they are transmitted to the objective weighting model components through the data flow. And then the allocation scheme calculation based on the model is achieved. The allocation scheme based on the objective weighting model is shown in Figure 6.

When applying a statistical analysis model to generate an allocation scheme, the amount of water available for allocation is first set. And secondly, the statistical analysis model is selected. Data flow is used to transmit the amounts of allocatable water resources, model selection results, and long series of actual water data of each object to the statistical analysis model components. The allocation scheme table and statistical diagram generated by model calculation are shown in Figure 7.

When the multi-objective optimization model is used to generate the allocation scheme, the setting results of the amounts of allocatable water resources, the selection results of the model, and the parameters of the multi-objective optimization model are transmitted to the multi-objective optimization model components. The allocation scheme table and the statistical diagram of the allocation scheme generated by calculation components are shown in Figure 8.

Compared with the traditional allocation scheme, which has the characteristics of fixed and static, the biggest feature and advantage of the simulation system is that it integrates multi-type allocation models, supports multi-model dynamic simulation, makes the simulation results ‘dynamically change’, and highlights the business nature of the simulation system.

(1) Dynamic simulation of a single model scheme.

The dynamics of the dynamic simulation of the single model scheme are reflected in the following aspects: 1. The dynamics of the weights in the allocation model. With the continuous updating of the calculation data stored in the database, the parameters of each model remain updated, and the allocation weights are dynamic. Moreover, the weighting indicators in this simulation system are optional, and the change of weighting indicators also makes the allocation of weights dynamic. 2. The amounts of allocatable water resources can change dynamically. The system designed a human-computer interaction window for the amounts of allocatable water resources to achieve the allocation simulation of water resources with different levels of abundance and drought. The dynamic simulation of the allocation scheme under the change of the amounts of allocatable water resources and the weighting indicators is shown in Figure 9.

(2) Simultaneous dynamic simulation of multi-model schemes.

The system is designed with a dynamic adjustment window for allocatable water resources, which can meet the dynamic changes in allocatable water resources with an interannual variation. According to the change of the amounts of allocatable water resources, the years can be divided into the wet year, normal year, and dry year, etc. The system supports the application of multiple allocation models for dynamic simulation of allocation schemes at the same time. Scenario 1: It is predicted that the current year will be a wet year. Multiple groups of allocatable water resources can be designed, and two to four allocation models can be used to perform dynamic simulations of the allocation schemes. Relying on the visual display interface of multiple schemes, decision-makers can make comprehensive decisions more efficiently. Scenario 2: The year is predicted to be a normal year. Multiple predicted water resource values for a normal year are set up via an interactive window and multiple allocation models are applied for dynamic simulation. Scenario 3: The year is predicted to be a dry year. Setting up multiple groups of allocatable water resources in a dry year, and carrying out dynamic simulation through multiple models for decision-makers to analyze and choose. The dynamic simulation of the multi-model schemes is shown in Figure 10.

The selection of multi-model dynamic simulation results is recommended as follows: 1. When the decision preference takes into account the principles of fairness, efficiency, and sustainability, the allocation scheme based on subjective and objective weighting model is recommended. The model approach takes better account of the mutual importance of multiple allocation principles by assigning weights to multiple impact indicators in the calculation process. 2. When the decision preference is to take into account the current water utilization situation of each target, it is recommended to apply a solution based on the statistical analysis model. By analyzing the law of water utilization, the model takes the proportion of current water utilization of each object as the weight to better represent the current water utilization pattern of each object. 3. The allocation scheme based on multi-objective optimization is recommended when the decision preference is the maximum comprehensive water utilization benefit. By optimizing the benefit and cost of water resources utilization, the allocation scheme of the maximum benefit of water resources utilization is obtained.

In summary, the dynamic simulation of multi-model allocation schemes based on the simulation system overcomes the shortcomings of traditional allocation schemes that do not adapt to changes, increases the optional space for decision optimization of allocation schemes, and makes the generation and application of allocation schemes more efficient and flexible.

Owing to the immaturity of water resources allocation theory, the disconnect between allocation theory and practical application, and the difficulty of adapting traditional allocation schemes to dynamic changes of water resources, this research applies computer technology to standardize and encapsulate traditional allocation models and builds a multi-method integrated simulation system to achieve dynamic simulation and integrated application of the sets of allocation models and schemes. The main conclusions of this research are as follows:

• 1.

  Relying on the comprehensive integration platform and based on components, knowledge map, and other development technologies, this research builds a multi-method integrated simulation system for water resources allocation. The simulation system integrates multiple traditional allocation models and solves the scientific problem of less traditional allocation methods by integrating multiple models.

• 2.

  This simulation system supports the visual generation and dynamic simulation of multi-model allocation schemes, which solves the problem that traditional allocation schemes are difficult to adapt to the dynamic changes of allocatable water resources. The allocation scheme generation and dynamic simulation based on this system have the advantages of process visualization, results in credibility, and supporting the decision. At the same time, the idea of dynamic simulation of water resources allocation through the application of method sets and scheme sets of the simulation system has opened up a new research path for water resources allocation research.

The adaptability of allocation schemes is improved through the dynamic simulation system. In the future, models that can simultaneously take into account multiple allocation principles will be the main research direction. At the same time, how to select the scheme set formed by using the simulation system still needs further research.

This work was supported by the Natural Science Basic Research Program of Shaanxi Province (Grant No. 2019JLZ-15, 2019JLZ-16) and the Science and Technology Program of Shaanxi Province (Grant No. 2018slkj-4, 2020slkj-16) and State Key Laboratory of Eco-hydraulics in Northwest Arid Region, Xi'an University of Technology (Grant No. 2019KJCXTD-5). The authors thank the editor for their comments and suggestions.

All relevant data are available from an online repository or repositories at http://tjj.shaanxi.gov.cn/upload/2020/pro/3sxtjnj/zk/indexch.htm.