**Coupling Coordination Assessment on Sponge City Construction and Its Spatial Pattern in Henan Province, China**

#### **Kun Wang 1 , Lijun Zhang 1, \* , Lulu Zhang <sup>2</sup> and Shujuan Cheng 1**


Received: 15 October 2020; Accepted: 7 December 2020; Published: 11 December 2020

**Abstract:** Coordinating the "green" and "gray" infrastructure construction and the socioeconomic development is essential to sponge city construction. Most previous research has investigated the structural and non-structural approach for urban water management, such as operational practice, engineered measures, technical solutions, or planning management. However, there is a shortage of strategic management approaches to identify pilot sponge cities, which is essential to cities in developing countries under huge financial pressures. Hence, this paper proposed a coupling coordination evaluation index system to assess the coordination degree between economic development and infrastructure construction in Henan Province in central China. Then, the paper analyzed the differences of the coordination level and its spatial statistical pattern of the coupled and coordinated development of sponge city construction in Henan Province. The results show that: (1) from the perspective of comprehensive level, the problems of inadequate and unbalanced development of infrastructure construction and economic development level are prominent; (2) from the perspective of coordinated development level, the level of coupling and coordination development in Henan Province increased during the sample period, but the level of coupling and coordination development in each region was small; (3) from the perspective of relative development, Zhengzhou City is lagging behind in infrastructure, indicating that economic growth is faster than infrastructure construction, and other regions are lagging economic development, indicating that infrastructure construction is faster than economic growth; and (4) from the spatial statistical analysis, there is spatial positive correlation, that is, the area with high coupling degree of infrastructure construction and economic development level tends to be significantly concentrated in space. Studies have shown that Henan Province should focus on strengthening the construction of "green" infrastructure and increasing the infiltration of the underlying surface to counter the precipitation in urban areas in extreme climates.

**Keywords:** sponge city; coupling coordination degree; spatial pattern

#### **1. Introduction**

In recent years, the acceleration of urbanization in China has caused to the change of underlying surface. Due to the increasement of extreme climate and the backwardness of urban rainwater management system, the increasing of surface runoff is the inevitable consequence of frequent urban flooding. The water problem has become a common urban problem in China, posing a serious threat to the safety of life and property of urban residents, and has resulted in a huge impact on the environment and social economy [1]. For instance, on 21 July 2012, a serious waterlogging disaster occurred in Beijing, causing 79 deaths and 11.64 billion CNY in economic losses [2]. The 2015 Environmental Status Report conducted water quality assessments of groundwater monitoring sites in 202 cities. Among them, 9.1% had good water quality, and 61.3% had poor water quality [3]. Other studies have also shown that life-saving threats and economic losses caused by urban floods are on the rise [4]. In order to solve the urban water disaster and water environment problems, in 2013, China's central urbanization work conference officially proposed the construction of the sponge city [2].

Sponge city means that the city can be like a "sponge". It has good "elasticity" in adapting to environmental changes and coping with natural disasters. When it rains, it absorbs water, stores water, seeps water, and cleans water. When necessary, it releases stored water [5]. In addition to improving the absorption and storage of rainwater, the sponge city combines "green" infrastructure to enhance ecological functions, enhance urban aesthetic value, and create additional comfort spaces [6]. In addition, the construction of China's sponge city also draws on the experience of some developed countries in urban rainwater drainage facilities, such as the United States' Low Impact Development (LID) [7]; the UK's Sustainable Urban Drainage System (SuDs) [8]; Australia's Water Sensitive Urban Design (WSUD) [9]; and New Zealand's Low Impact Urban Development Design (LIDUD) [10]. Chinese scholars have made a lot of progress in the exploration of sponge city construction, combined with local favorable conditions to shape the experience of surrounding landscape system and rainwater system management, and extract the surrounding landscape (river and lake) from the macro level by using modern technology (Geographic Information System, Remote Sensing, etc.). The system, from the microscopic level to the landscape design (green roof, etc.), combined with water landscape reconstruction, can achieve the sustainable development of human-water relationship [2,11–14].

However, in the face of unprecedented climate change and urbanization, it is more important to strengthen the elasticity of urban infrastructure [15–17]. But most of the research is limited to describing the flood recovery capacity of urban drainage systems [18,19]. Changes in climatic conditions, such as increased precipitation intensity, changes in precipitation patterns, and more extreme weather events, have caused urban drainage systems to be frequently hit by heavy rains [20,21]; urbanization has increased the concentration of population and economic activity, plus the burden of existing urban social development systems on sewage, urban runoff and water pollutant types [22]. A sustainable urban infrastructure system should be effective and adaptable in an uncertain future, which will contribute to the flexibility of green and grey infrastructure in the future, and to carry out sustainable development assessments of urban drainage systems [23]. Numerous studies have evaluated the role of different types of green infrastructure in stormwater management and carbon emission control and compared the performance of grey infrastructure, which is considered to be mitigating and adapting to climate change and urbanization. Interferences contribute to the effectiveness of sustainable development [23–25], but grey infrastructure is also a necessary condition for dealing with extreme rains [26]. The mutual development and application of grey infrastructure and green infrastructure has been used in many countries to mitigate urban floods [27,28]. Green infrastructure is an important measure in many common stormwater management strategies, such as low-impact development, best management practices, and water-sensitive urban design [29]. The study of China's sponge cities also evaluated the effects of green infrastructure [30] and evaluated the performance of low-impact development practices as a means of reducing water flooding in urban small watersheds [31]. However, the investment and income mechanism of Sponge City is currently in the exploration stage [32], and the proportion of gray infrastructure and green infrastructure investment is also rarely considered. Therefore, this requires urban managers to plan the rainwater management infrastructure from an economic perspective and an adaptation path [33].

Sponge city construction, as a comprehensive ecological project, should be evaluated from the aspects of cost, social, and economic benefits, hydrology, and water quality. Lack of adequate financial support and effective market incentives is one of the main obstacles to sustainable storm water management [34]. To solve this problem, the United States introduced the Stormwater Utility Fees (SUF) in the 1970s, as user fees; SUF has increasingly been used by local governments as an alternative source of income for implementing sustainable rainwater projects [35]. In China, some researchers are also exploring revenue sources for sustainable stormwater management, and the research results show that 76% of the respondents agreed to pay for life-cycle maintenance of sponge city facilities, and the median amount of willingness to pay was 16.57 CNY (2.53 USD) per month [36]. In addition, under the guidance of ecological civilization construction, the green infrastructure in China's sponge city construction should be incorporated into the ecosystem service value, consider cities as a coupled nature-human system, for such a system, different urban ecosystem structures, such as lakes, wetlands, rivers, and parks, are an integral part, providing important services for the landscape, including water conservation and runoff regulation, and also helping urban residents to support social welfare [37]. The recycling of rainwater for the gray infrastructure of sponge city construction provides services for the community and green infrastructure. On the one hand, it promotes the efficiency of recycled water use in the community. On the other hand, it supports the water demand for green infrastructure.

However, these studies are still obviously insufficient for the construction of sponge cities. Chinese scholars rarely conduct research on economic development level and infrastructure, and the construction of sponge city is a comprehensive process, establishing a perfect "gray" infrastructure and "green" infrastructure system, actively promoting the fact that the pilot work of sponge city construction is the key to solving the problem of domestic embarrassment. To this end, this paper attempts to analyze the coupling and coordination degree of infrastructure and economic development in 18 prefecture-level cities in Henan Province, as well as provide a new research idea for the construction of the sponge city and a theoretical basis for the future development of the sponge city.

#### **2. Materials and Methods**

#### *2.1. Study Area*

Henan Province is located in the transition zone between central and eastern China, the southern region and the northern region. It has a large span from north to south. From the west to the east, the Yellow River runs through. There are many rivers in Henan Province, which provide a surrounding landscape (rivers and lakes) for the construction of the sponge city. Wetland and green space provide good carriers for the construction of sponge city. Henan Province has diverse terrain, most of which is dominated by mountains and plains. The "green" infrastructure construction in different regions is also very different. Most of Henan's climate is dominated by temperate monsoon climate, and some regions are transition from subtropical monsoon climate to temperate monsoon climate. In the zone, affected by the monsoon climate, the impact of the formation of the mountain microclimate is more obvious, and the differences between the regions are greater. Therefore, the construction of sponge cities in different regions varies from region to region; Henan Province has more latitudes, and most of the region belongs to the "northern region (dividing line between subtropical and temperate monsoons climate, northern region belongs to the temperate monsoon climate)"; however, Xinyang and Nanyang cities in the southern part of Henan province have subtropical climates, because they are near the Qinling Mountains-Huaihe River line (the boundary between temperate and subtropical monsoon climates). The difference between the south and the north of the south is obvious. The problems faced by various regions in Henan Province are similar to those faced by China at this stage of development. They are called "the epitome" of China (Figure 1); the construction of sponge cities in Henan Province can be studied. The commonality of the sponge city provides practical experience for the construction of China's sponge city. The research area is the smallest scale of 18 prefecture-level cities in Henan Province. The differences in climate, precipitation, and urban development between the regions are more obvious, but precipitation is mainly concentrated in the summer and autumn, more often with short-term heavy rain, annual average the temperature is about 16 degrees Celsius, and the annual precipitation is about 650 mm (Figure 2). In addition, most of the terrain in the southeast of Henan province is plain, with monsoon climate in summer and short-term precipitation, the precipitation

is relatively large, and the terrain is low and flat, with slow surface runoff and infiltration, where it is easy to cause waterlogging. Due to the influence of climate and topography, the precipitation in the study area decreases from southeast to northwest. However, in recent years, due to unreasonable development and utilization, as well as the impact of extreme climate, some rivers and lakes are seriously deficient in water. In the summer and autumn, the urban drainage capacity is insufficient, and the urban shackles occur, which also hinders the development of the city to a certain extent.

**Figure 1.** Research area.

#### *2.2. Data Source*

The construction of the sponge city not only requires the support of science and technology to change the landscape of the underlying surface but also requires a large amount of funds to build a "gray" infrastructure system and a "green" infrastructure system. The use of "green" and "gray" infrastructure can enhance urban resilience, and "green" infrastructure can provide greater adaptability and resistance to the unpredictability of future climate forecasts. In the face of unsustainable urban drainage practices, a good strategy is to integrate green and gray infrastructure into a cyclic utilization control system, which will bring more advantages and reduce problems, while improving existing elasticity of urban drainage systems [23]. In view of this, this study selected the road area, drainage pipe length, road length, sewage treatment rate, and number of bridges in the "gray" infrastructure system; the green area in the "green" infrastructure system, the per capita park green area, the park area; and the main indicators, such as water reuse rate and green coverage area, economic development level of GDP, fixed investment in garden green space, fixed drainage investment, urban population, and urbanization rate, to be analyzed (Table 1). The missing data of some indicators were simulated by linear interpolation. In this paper, the missing indicators are the fixed investment in Kaifeng garden green space and drainage in 2016, as well as the fixed investment in Luohe garden green space and Xinxiang Drainage in 2014. In this paper, linear interpolation is used, that is, a function method that is used to calculate an unknown quantity between two endpoints of a line, because the interpolation accuracy on nodes can be guaranteed to be higher, and it is more convenient than other interpolation methods, such as parabolic interpolation. The data mainly came from China Urban Construction Statistical Yearbook 2012–2016 and Henan Statistical Yearbook 2013–2017.

**Figure 2.** Annual average temperature and annual precipitation in the study area.


#### *2.3. Coupling Coordination Mechanism*

Coupling refers to the phenomenon that two or more systems interact and influence each other, while coupling degree is used to describe the degree of mutual influence of multiple systems. In general, the degree and quality of coupling action, using coordination to judge. Coordination is a benign embodiment based on coupling effect, and the coordination degree is used to measure the coordination degree of multiple systems in the coupling process. Both have a connection already and have distinction. The degree of coupling reflects the degree of system interaction, and the degree of coordination reflects the degree of coupling coordination [38].

The construction of sponge city is a composite system composed of green infrastructure, gray infrastructure, and economic development level, which interact and form a symbiotic coupling relationship. In China, with ecological civilization, green infrastructure is the leading driving force for the construction and development of sponge cities, and this driving force can improve the material structure of the underlying surface and increase the infiltration of rainwater, which, to a certain extent, is conducive to dealing with urban waterlogging caused by extreme climate. In addition, the construction of green infrastructure can provide ecosystem services value to surrounding communities. The construction of grey infrastructure restricts the improvement of the underlying surface but promotes the management of rainwater, domestic water, and sewage, which is conducive to the economical utilization of water resources and provides water demand for green infrastructure. The level of economic development is the material guarantee for the construction of green infrastructure and gray infrastructure, and it can provide financial support for the construction of green infrastructure and gray infrastructure through the investment effect (PPP, Public-Private Partnership model or willingness to pay) and the value of ecosystem services. Therefore, the three form a multiple correlation interaction coupling effect, which has both positive and negative effects. The sustainable development of sponge city can be realized only when the three cooperate and coordinate with each other.

Green Infrastructure construction in a sponge city references LID (low-impact development), an ecologically-based planning and engineering design approach to managing stormwater runoff and stormwater treatment technologies. In the practice of SuDS (Sensitive urban drainage system design) in Western Europe, SuDS sustainable stormwater management measures mitigate and adapt to climate change through carbon sequestration and urban cooling, with multiple ecological and environmental benefits, based on such a concept, as much as possible to restore the natural and pre-development drainage system. The construction of grey infrastructure in a sponge city is mainly used for sewage treatment and watercourse pipe network construction. In stormwater management, management systems link non-structural approaches to structural deployment for pollution prevention and drainage. Basically, similar to Best Management Practices (BMPs) in the United States and Canada. WSUD (Water-Sensitive Urban Design) is mainly to protect and strengthen the natural water system of urban development, better integrate rainwater into the landscape, minimize impermeable water surface, reduce the peak flow brought by urban development, and protect water quality, reduce the development costs of drainage infrastructure, while adding value at the same time, and all urban water systems should be better integrated in urban design [29]. And sponge city is a systematic framework with methods to improve urban water problems. Therefore, based on the coupling coordination mechanism, this paper provides a basis for decision-making and management of sponge city construction.

#### *2.4. Evaluation Index System*

The sponge city construction evaluation is a diagnostic analysis of the drainage capacity of an area and the environmental impact of the underlying surface. The sponge city construction evaluation is based on the construction of the infrastructure system and economic development level model. Among them, gray infrastructure is a traditional municipal infrastructure dominated by single-function municipal engineering, consisting of roads, bridges, pipelines, and other networks that ensure the proper functioning of the industrial economy [39]. However, the construction of road infrastructure

changes the underlying surface structure, reduces the infiltration amount of surface runoff, and is also the infrastructure that has a great impact on urban residents' travel. In general, drainage pipeline network is under road infrastructure. In the construction of sponge cities, gray infrastructure facilities provide municipal entities and residents with municipal infrastructure services, such as flood protection, stormwater drainage, and wastewater treatment [39]. The "gray" infrastructure system reflects the traditional drainage system and the foundation of a good operation of the city. Green infrastructure refers to a green space network consisting of natural areas and other interconnected open spaces, including natural areas, public protected areas, and productive land with conservation values; it also represents a protected open system network that protects the value of natural resources and maintains the survival functions of humans, animals, and plants [40,41]. The development of green infrastructure is the result of joint promotion of parks, park systems, open spaces, greenways, ecological networks, biological corridors, and storm-water management [39]. The "green" infrastructure system reflects the sustainable development model of improving the underlying surface and is the basis of a good ecological environment of the city. The level of economic development reflects a city's strong support for the construction of sponge cities. Based on these evaluation methods and models, combined with the individual indicators that can reflect the construction of sponge cities in Henan Province, the index system for sponge city construction was initially determined (Table 1).

#### *2.5. Research Methods*

#### 2.5.1. Determination of Entropy Weight

Entropy was first introduced into the information theory by Shannon. It has been widely used in engineering, social, and economic fields. It hypothesizes that the quantity or quality of information is an important factor in determining the reliability and accuracy of decision-making [42]. Entropy is often used to measure the amount of useful information provided by the dataset itself and is therefore considered to be a suitable indicator for use in various evaluation cases. Weights can be determined based on the data itself, thereby reducing decision bias and increasing the objectivity of the decision process [43]. The entropy weight method is also to calculate the comprehensive index by the size of the selected index information. The index weight is determined by the judgment matrix composed of the evaluation indicators. Since the evaluation system has positive and negative indicators, the sample matrix needs to be dimensionless [44]. The main indicators selected in this paper are studied through positive and negative indicators.

#### (1) Data standardization processing

Assume that there are n research objects (mainly cities) in the study area, including m evaluation indicators, and the definition P is the original data matrix, expressed as:

$$\mathbf{P}\_{\rm nm} = \begin{bmatrix} \mathbf{P}\_{\rm 11} & \mathbf{P}\_{\rm 22} & \mathbf{P}\_{\rm 13} & \cdots & \cdot & \cdot & \cdot & \cdot & \cdot\\ \cdot & & & & & \cdot\\ \cdot & & & & & \cdot\\ \cdot & & & & & \cdot\\ \cdot & & & & & \cdot\\ \cdot & & & & & \cdot\\ \cdot & & & & & \cdot\\ \cdot & \mathbf{P}\_{\rm n1} & \mathbf{P}\_{\rm n2} & \mathbf{P}\_{\rm n3} & \cdot & \cdot & \cdot & \cdot & \cdot\\ \end{bmatrix} \tag{1}$$

Among them, Pnm is expressed as the m item of the n city (m = 1, 2, 3, 4, 5, 6· · · · · · ; n = 1, 2, 3, 4, 5, 6, 7, 8 9, 10, 11· · · · · · ). Because the dimension of each index coefficient is not uniform, the index coefficient is standardized by the method of extreme difference. The main indicators selected in this paper are studied by the forward index and the reverse index. When the positive index is larger than the index value, the better the index. The normalization method is: Ynm = (Xnm − minxm)/(maxx<sup>m</sup> − minxm);

the inverse index, that is, shows that, the smaller the index value, the better the index, and the normalization method is: Ynm = (maxx<sup>m</sup> − Xnm)/(maxx<sup>m</sup> − minxm). In the normalized method formula, Xnm is expressed as a specific value, minx<sup>m</sup> is represented as the minimum value of the m index, and maxx<sup>m</sup> is expressed as the maximum value of the m index. The value is between Ynm∈[0, 1] after normalization [45,46].

#### (2) Determination of indicator weight

In information theory, the larger the entropy value, the smaller the difference between the values of the evaluation indicators, and the smaller the weight of the index; the smaller the entropy value is, and vice versa [46], to calculate the information entropy of each index, assuming that E<sup>m</sup> represents the m. The information entropy under the indicator is calculated as:

$$\mathbf{E\_m} = -\mathbf{k} \sum\_{\mathbf{n}=1}^{\mathbf{h}} \mathbf{f\_{nm}} \text{lnf\_{nm}} \tag{2}$$

where k = 1/lnh, fnm = Ynm/ P<sup>h</sup> n=1 ynm, if fnm = 0, then define

$$\lim\_{\mathbf{f}\_{\rm fnm}\to 0} \mathbf{f}\_{\rm nm} \mathbf{l} \mathbf{n} \mathbf{f}\_{\rm nm} = \mathbf{0} \tag{3}$$

Among them, the m indicator of the n city of Ynm is a specific value.

According to the calculation formula of information entropy, the information entropy E1, E2, . . . , Em of each index is calculated. The weight of each index is calculated by information entropy. Wm is the entropy weight of the m evaluation index, and then the weight of the index is calculated. The method is [45,46]:

$$\mathbf{W\_m} = \frac{1 - \mathbf{E\_m}}{\sum\_{\mathbf{n}=1}^{\mathbf{h}} \mathbf{E\_m}} \tag{4}$$

where W<sup>m</sup> <sup>∈</sup> [0, 1], <sup>P</sup><sup>h</sup> <sup>n</sup>=<sup>1</sup> W<sup>m</sup> = 1.

#### 2.5.2. Coupling Coordinated Development Model

(1) Comprehensive evaluation model

The comprehensive evaluation model is used to measure the level of infrastructure and economic development. The calculation method is:

$$\mathbf{S} = \sum\_{\mathbf{i}=1}^{n} (\mathbf{W}\_{\mathbf{m}} \times \mathbf{Y}\_{\mathbf{k}}) \tag{5}$$

Among them, S represents the comprehensive index of infrastructure construction or economic development; W<sup>m</sup> represents the weight of each index within the system; Y<sup>k</sup> represents the evaluation value of each indicator.

#### (2) Coupling degree model

Coupling refers to the phenomenon in which two or more systems or forms of motion interact and interact with each other through some means. This paper establishes a coupling model of infrastructure and economic development. The calculation method is:

$$\mathbf{C} = \left\{ \frac{(\mathbf{S}\_1 \times \mathbf{S}\_2)}{\left[ \frac{(\mathbf{S}\_1 + \mathbf{S}\_2)}{2} \right]} \right\}^2 \tag{6}$$

Among them, C is the coupling degree between infrastructure construction and economic development, and the value range is [0, 1]. The larger C, the stronger the interaction between infrastructure construction and economic development; S1 and S2 are infrastructure construction and economy, respectively. The comprehensive index of development, k as the adjustment factor, in practice, should make k ≥ 2; this paper takes k = 2.

#### (3) Coupling coordination degree model

The coupling degree model can only indicate the existence of interaction between systems and cannot reflect the level of coupling coordination between systems. Therefore, this paper further constructs a coupling coordination model of infrastructure construction and economic development. The calculation method is:

$$\begin{cases} \mathbf{D} = \sqrt{\mathbf{C} \times \mathbf{T}} \\ \mathbf{T} = \mathbf{a} \times \mathbf{S}\_1 + \boldsymbol{\beta} \times \mathbf{S}\_2 \end{cases} \tag{7}$$

where D is the coupling coordination degree; T is the inter-system comprehensive coordination index; α and β are undetermined coefficients, and α + β = 1. This paper assumes that infrastructure construction and economic development interact, so take α = β = 0.5.

Based on the relevant classification criteria proposed in the existing research [47,48], the median segmentation method is used to divide the D value into four stages. The classification criteria are shown in Table 2.


**Table 2.** Coordination level of coupling coordination.

#### 2.5.3. Spatial Statistical Methods

This paper introduces Moran's I to analyze the imbalance and spatial autocorrelation of the coupling and development of infrastructure construction and economic development between adjacent regions. The calculation formula of Moran's I is as follows:

$$\mathbf{I} = \frac{\sum\_{\mathbf{i}=1}^{n} \sum\_{\mathbf{j}=1}^{n} \mathbf{w}\_{\overline{\mathbf{i}}} (\mathbf{Y}\_{\mathbf{i}} - \overline{\mathbf{Y}}) (\mathbf{Y}\_{\mathbf{j}} - \overline{\mathbf{Y}})}{\mathbf{S}^2 \sum\_{\mathbf{i}=1}^{n} \sum\_{\mathbf{j}=1}^{n} \mathbf{W}\_{\overline{\mathbf{i}}}} \tag{8}$$

$$\mathbf{I}\_{\mathbf{i}} = \frac{\left(\mathbf{Y}\_{\mathbf{i}} - \overline{\mathbf{Y}}\right)}{\mathbf{S}^2} \sum\_{\mathbf{j}=1}^{\mathrm{n}} \mathrm{w}\_{\overline{\mathbf{j}}} \left(\mathbf{Y}\_{\mathbf{j}} - \overline{\mathbf{Y}}\right) \tag{9}$$

where I represents the overall degree of correlation between regions, S <sup>2</sup> = <sup>1</sup> n P<sup>n</sup> i=1 Y<sup>i</sup> − Y 2 ; Y = <sup>1</sup> n P<sup>n</sup> <sup>i</sup>=<sup>1</sup> Y<sup>i</sup> ; Y<sup>i</sup> represents the degree of coupling coordination of the i region; n is the number of regions, and wij represents the element of the spatial weight matrix W. I<sup>i</sup> indicates the degree of correlation between the coordination degree of the i region and the surrounding area, and the local spatial features are displayed by using the Moran scattergram.

#### **3. Results**

#### *3.1. Analysis of Coordination Degree between Regions*

According to the coupling evaluation model of infrastructure construction and economic development, the comprehensive index of infrastructure construction (S1) and the comprehensive index of economic development (S2), and the coupling coordination degree (D) of the two are calculated. This paper will be in the process of empirical analysis. Infrastructure construction and economic development are regarded as two subsystems of equal importance. Therefore, the undetermined coefficients of the two are all 0.5, that is, α = β = 0.5. Therefore, the comprehensive harmonic index of the two is T = 0.5 × S\_1 + 0.5 × S\_2, combined with the coupling coordination degree model for calculation; the obtained empirical results are shown in Table 3.

**Table 3.** Calculation results of the coupling degree of regional infrastructure construction and economic development.


Comparing the coupling and coordination of infrastructure and economic development in each city in 2013 and 2017, we can find that the average system coupling coordination degree of 18 cities in Henan Province in 2016 is 0.6105, which is slightly higher than the average of 0.6036 in 2013. The infrastructure construction of each city and overall average level of economic development level coupling coordination is in the low and medium coupling coordination stage. Among them, the coordination degree of Zhengzhou City, Luoyang City, Jiaozuo City, Xuchang City, Nanyang City, and Zhumadian City is higher than the average level of Henan Province. The coupling coordination degree of most regions is rising continuously. Among them, the coupling coordination degree of Pingdingshan City, Nanyang City, and Jiyuan City shows a downward trend.

In view of the different development speeds of different regions, there are obvious differences in the coupling and coordination degree between infrastructure construction and economic development in various regions. Figure 3 shows that the overall evolution of regional infrastructure construction and economic development level coordination can be divided into two types in 2013–2017: the first category is the area where the coupling coordination degree is between 0.8 and 1, namely Zhengzhou City. Zhengzhou City has been at a relatively high level of coupling and coordination in 2013–2017, indicating that the infrastructure construction and economic development level are at an effective coupling development stage. In recent years, Zhengzhou Zheng Dong New Area has been built and developed according to the concepts and ideas of sustainable "ecological city", "metabolic city", and "sponge city". Among them, the completed water area covers 18 square kilometers, and the green area covers 39 square kilometers, and the green coverage rate in the urban core area is nearly 50%. Drawing on the experience of modern urban construction in the west, the sustainable water cycle is realized through tracking and integrating low impact development (LID) and rainwater utilization. According to the climate characteristics of the Zhengzhou Zheng Dong New Area, the PP module storage device is used to collect and purify rainwater for green irrigation, creating a sustainable green landscape [49]. The construction of Zhengzhou Zheng Dong New Area is inseparable from the strong economic support. The second category is the area where the coupling coordination degree is between 0.5 and 0.8. Although these areas are in a highly coupled and coordinated development stage, the infrastructure construction and economic development level are relatively insufficient, and the infrastructure construction and economic development level are relatively weak. The two have not yet formed a benign interactive coupling development model, and there is still much room for improvement in the coordinated and coordinated development. Government departments in various regions should formulate corresponding infrastructure construction and economic development level strategies according to local actual conditions, as well as upgrade infrastructure and economic development as soon as possible.

**Figure 3.** Evolution of the coupling degree of regional infrastructure construction and economic development level.

#### *3.2. Evolution of Spatial Pattern of Coordinated Development*

#### 3.2.1. Evolution of Regional Differences

In order to better analyze the spatial pattern and dynamic evolution of the coupling and coordinated development of inter-regional infrastructure construction and economic development level, this paper uses 2014 and 2017 as time nodes, combined with Table 3, through ArcGIS 10.2 software, respectively, for an inter-regional 2014, 2017 spatial visualization of the system coupling coordination level for the year (Figure 4).

**Figure 4.** Spatial pattern evolution of the coupling degree of infrastructure construction and economic development in 18 cities in Henan Province.

Combined with Table 3 and Figure 4, it can be seen that the spatial pattern of infrastructure construction and economic development level is obviously different. The coupling and coordination distribution of Henan Province is basically consistent with the spatial location of economic development, and the degree of system coupling coordination is relatively high in regions with relatively developed economy in areas with low levels of economic development. During the study period, the coordination degree of Zhengzhou City, Nanyang City, Luoyang City, and Xuchang City has been maintained at a high level, while the coupling coordination degree of Hebi City, Zhumadian City, Zhoukou City, and Shangqiu City is in a rising trend state. The degree of economic development of the region is closely related. With the development of the economy, the investment in infrastructure construction is also increasing; the coupling coordination degree of Puyang City, Xinxiang City, Kaifeng City, and Pingdingshan City is declining, which is related to the development of the region and policy. Overall, the infrastructure construction and economic development level of Henan Province presents a spatial pattern of "high west and low east". However, the classification shows that the coupling degree between infrastructure construction and economic development level in Henan Province has not changed, but each prefecture-level city fluctuates in the grade interval.

#### 3.2.2. Statistical Analysis of Local Space

Through the global Moran's I statistic, it can be seen that the coupling and coordinated development of infrastructure construction and economic development level in Henan Province has significant spatial agglomeration, as well as the correlation and concentration between regions, the spatial correlation with the surrounding areas, the degree of spatial difference. The distribution of the spatial pattern, on whether there is heterogeneity, in this paper, is analyzed by Moran scatter plot (Figure 5).

It can be observed from Figure 5 that the level of coupling and coordination between infrastructure construction and economic development level in most regions in the past five years is in a stable upward and downward fluctuation state. Compared with 2014, the first quadrant of 2017 Moran's I is reduced, and the fourth quadrant is increased. And the Moran's I index is greater than 0, indicating that there is spatial positive correlation, that is, the area where the coupling degree of infrastructure construction and economic development level is higher (or lower) tends to be significantly concentrated in space, and the correlation is stronger; if Moran's I = 0, it means that the space is not correlated, and the distribution is in a random state. Table 4 reports the relative development degree and relative development type. The results show that Henan Province is generally synchronous development, Zhengzhou City is lagging behind infrastructure, indicating that economic growth is faster than infrastructure construction; other regions are economic development. The lag type indicates that the economic growth is slower than the infrastructure construction, and the construction of infrastructure is more advanced than consumption. Therefore, to promote economic growth, it is necessary to

increase investment in land resource conservation and intensiveness in order to realize infrastructure construction and economy. Coordinated development levels and simultaneous development are important ways to achieve coordinated and coordinated development.

**Figure 5.** Moran scatter plot of the coupling degree of infrastructure construction and economic development level in Henan Province. (**a**): Moran's I index in 2014. (**b**) Moran's I index in 2017.


**Table 4.** Types of relative development of infrastructure construction and economic development level in 2013–2017.

#### **4. Conclusions and Recommendations**

Based on the coupling coordination degree model, relative development degree model, and spatial statistical analysis, this research work studied the difference of horizontal and spatial statistical analysis of the coupled and coordinated development of sponge city construction in Henan Province. The results show that: (1) From the perspective of comprehensive level, the problems of inadequate and unbalanced development of infrastructure construction and economic development level are prominent. (2) From the perspective of coordinated development level, the level of coupling and coordination development in Henan Province increased during the sample period, but the level of coupling and coordination development in each region was small. (3) From the perspective of relative development, Zhengzhou City is lagging behind in infrastructure, indicating that economic growth is faster than infrastructure construction; other regions are lagging economic development, indicating that economic growth is faster than infrastructure construction. Slowly, (4) from the spatial statistical analysis, the Moran's I index is greater than 0, indicating that there is spatial positive correlation, that is, the area with high coupling degree of infrastructure construction and economic development level tends to be significantly concentrated in space.

According to the above conclusions, due to the different natural foundations, economic reserves, location advantages, historical background, social influence, and policy conditions of various regions, the coordinated development of sponge city construction in Henan Province requires differentiated regional infrastructure construction and economy development policy. In areas with better economic development, it is necessary to increase the proportion of investment in "green" infrastructure and "gray" infrastructure, as well as appropriately increase the proportion of investment in "green" infrastructure; in areas with insufficient economic development level, it is necessary to adapt to local conditions. In the built-up area, the proportion of "green" infrastructure and "gray" infrastructure will be created to reduce the frequent flooding in the city and to continuously improve the renewal and construction of the drainage system. At the same time, sponge buildings and residential areas should be promoted, and measures, such as roof greening, rainwater storage, collection and utilization, and micro-topography, should be taken according to local conditions to improve the rainwater storage and retention capacity of buildings and residential areas. Rainwater collection and recycling, on the one hand, can provide water for the vegetation of green infrastructure, and, on the other hand, they can provide the use of reclaimed water for the community. The construction of green infrastructure in sponge cities increases the vegetation cover and water area of cities, effectively weakens the urban heat island effect, and thus affects the precipitation process. Therefore, the construction of green infrastructure in sponge cities can conserve water resources, regulate runoff, purify water quality, save water resources, improve the carrying capacity of regional water resources, and enhance the capacity of natural water storage and drainage. Sustainable sponge city construction needs to coordinate the coordination and development of green infrastructure, grey infrastructure, and economic development.

**Author Contributions:** Conceptualization, L.Z. (Lijun Zhang); methodology, K.W. and L.Z. (Lijun Zhang); software, K.W.; validation, K.W., L.Z. (Lulu Zhang); Formal Analysis, K.W.; Investigation, K.W., S.C., and L.Z. (Lulu Zhang); Resources, L.Z. (Lijun Zhang); writing—original draft preparation, K.W.; writing—review & editing, K.W. and L.Z. (Lijun Zhang); visualization, K.W.; project administration, L.Z. (Lijun Zhang). All authors have read and agreed to the published version of the manuscript.

**Funding:** The present study is supported by the Key Scientific Research Projects of Institutions of Higher Learning in Henan Province in 2017 (Grant No. 17A170006).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


49. Zheng Dong New District. Available online: http://www.zhengdong.gov.cn/sitesources/zhengdong/page\_pc/ index.html (accessed on 25 November 2020).

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Delving into the Divisive Waters of River Basin Planning in Bolivia: A Case Study in the Cochabamba Valley**

**Nilo Lima-Quispe 1 , Cláudia Coleoni 1 , Wilford Rincón 2 , Zulema Gutierrez 3 , Freddy Zubieta 3 , Sergio Nuñez 3 , Jorge Iriarte <sup>4</sup> , Cecilia Saldías 3 , David Purkey 1 , Marisa Escobar <sup>5</sup> and Héctor Angarita 1, \***


**Abstract:** River basin planning in Bolivia is a relatively new endeavor that is primed for innovation and learning. One important learning opportunity relates to connecting watershed planning to processes within other planning units (e.g., municipalities) that have water management implications. A second opportunity relates to integrating watershed management, with a focus on land-based interventions, and water resources management, with a focus on the use and control of surface and groundwater resources. Bolivia's River Basin Policy and its primary planning instrument, the River Basin Master Plan (PDC in Spanish), provide the relevant innovation and learning context. Official guidance related to PDC development lacks explicit instructions related to the use of analytical tools, the definition of spatially and temporally dis-aggregated indicators to evaluate specific watershed and water management interventions, and a description of the exact way stakeholders engage in the evaluation process. This paper describes an effort to adapt the tenets of a novel planning support practice, Robust Decision Support (RDS), to the official guidelines of PDC development. The work enabled stakeholders to discern positive and negative interactions among water management interventions related to overall system performance, hydrologic risk management, and ecosystem functions; use indicators across varying spatial and temporal reference frames; and identify management strategies to improve outcomes and mitigate cross-regional or inter-sectorial conflicts.

**Keywords:** water resources systems; participatory modeling; river basin planning; watershed management; water scarcity; water conflicts; Robust Decision Support; WEAP; Integrated Water Resources Management; Bolivia

#### **1. Introduction**

Water resources managers worldwide face high levels of natural and human-induced hydrologic variability accompanied by climate change projections [1] suggesting increased risks of water scarcity [2]. Defining future changes in hydrologic variability is highly uncertain, hindering the prediction of extreme events such as floods and droughts [3]. In addition to hydrologic variability, water managers deal with growing long-term demands for water from rapidly expanding urban areas and increased consumption across sectors such as agriculture and energy [4]. The intensification of human–water interactions reaffirms the need for 'good governance' to improve water management [5,6]. As the UN World Water Development Report [7] stated, "the world's water crisis is one of water governance, essentially caused by the ways we mismanage water". Under scenarios of deep uncertainty, water governance approaches should support water-related decision making [1]. The recognition of the interconnected nature of the biophysical and socioeconomic factors

**Citation:** Lima-Quispe, N.; Coleoni, C.; Rincón, W.; Gutierrez, Z.; Zubieta, F.; Nuñez, S.; Iriarte, J.; Saldías, C.; Purkey, D.; Escobar, M.; et al. Delving into the Divisive Waters of River Basin Planning in Bolivia: A Case Study in the Cochabamba Valley. *Water* **2021**, *13*, 190. https:// doi.org/10.3390/w13020190

Received: 31 October 2020 Accepted: 5 January 2021 Published: 14 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

that converge in water management has resulted in the progressive adoption of integrative governance approaches, such as the Integrated Water Resources Management (IWRM) [8,9]. The IWRM framework has been widely adopted as a template for water governance [10,11] and was recently included as an implementation target for the Sustainable Development Goal (SDG) 6 'Water and Sanitation' [12]. Numerous countries, including Bolivia, have attempted to adapt IWRM to their contexts and realities in the hope of improving outcomes in river basins characterized by scarcity and conflict. At its core, IWRM relies on placing biophysical and technical knowledge that is comprehensively developed at the scale of a river basin planning unit at the center of stakeholder interactions designed to identify integrated and coherent river basin scale solutions [8,9]. However, efforts to evaluate the river basin scale performance of specific interventions does not necessarily align with an individual stakeholder's more parochial objectives and interests. Nonetheless, within existing water policy frameworks, such as the European Water Framework Directive and many others, the river basin is commonly defined as the predominant spatial domain for water management, superseding other territorial and administrative boundaries [13,14]. The river basin unit also drives water management governance, generating new regulatory frameworks and institutional arrangements such as river basin organizations to promote the participation of multiple water users and civil society [15–17]. However, the river basin perspective may lead to potential conflicts or ambiguities with other existing levels of territorial governance more connected to individual stakeholder interests or frames of reference, such as municipalities or targeted water management entities (e.g., water utilities, irrigation districts). Although river basins certainly do define useful 'natural' boundaries [18,19], river basin management is also a result of 'political' processes and choices based on values and preferences often defined at sub-watershed scales, based on administrative boundaries [20,21].

Although the IWRM framework is often used interchangeably with the concept of water governance, Lautze et al. [22] warn that "setting pre-determined goals or outcomes associated with IWRM circumscribes a major role of water governance—that of determining goals". In practice, the attempt to uniformly apply IWRM principles leads to 'poor' rather than 'good' water governance since local conditions, preferences, and values would remain largely misrepresented [22]. Ensuring engaged and effective participation from across sub-basin jurisdictions is key to address collective action dilemmas arising from diverse and potentially conflicting users' interests over Common Pool Resources such as a shared river basin [23]. Human–water interactions are often characterized by mismanagement (e.g., depleted, overused water) and ecosystem impairment [24], threating the water commons. For instance, in river basins, collective action dilemmas can lead to unsustainable water use and insufficient or unsafe water supply for downstream users [25]. This is surely contrary to the interests of some sub-basin stakeholder interests. A common approach toward representing sub-basin interests within a comprehensive river basin scale planning process involves developing analytical tools sufficiently disaggregated to capture the diversity of conditions and interest within the river basin planning unit. Many approaches and case studies pertaining to the use of models to simulate water resources system are available in the literature, which are addressed and discussed by Mashaly and Fernald [26].

Mirchi et al. [27] define three modeling approaches in support of water resources planning and decision making: predictive simulation, integrated descriptive models, and participatory models. The first case predicts the future behavior of a particular subsystem, such as hydrological conditions [27], based on output from a calibrated and validated historical model [26]. The second case adopts a more holistic approach allowing for feedback between two or more disparate subsystems such as hydrological, social, ecological, economic, and political based on historical patterns of interaction [27]. The third approach promotes the participation of decision makers and stakeholders in the modeling process [26,27] where they can express their interests [28]. This approach allows for the performance evaluation of interventions in the face of changing climate or evolving so-

cietal prerogatives [29–32]. This requires an understanding of the relationships between climate, water resources, and user expectations, which can be addressed using integrated hydrological/water management modeling tools [33] that quantify potential impacts at varying spatial and temporal scales [34]. For instance, the Water Evaluation and Planning system (WEAP) [35] has been applied in several river basins worldwide as a simulation model for decision support while considering multiple objectives [36–38]. The Sacramento Water Allocation Model, known as SacWAM, is an example of a WEAP-based hydrologic and system model that illustrates the complex water system operation, besides showing how its water would flow if there were no dams, diversions, or infrastructure [39]. The use of modeling tools allows water resources managers and stakeholders to identify the operational, institutional, or infrastructure interventions that meet the disparate goals set out for a water system [4]. Although participatory frameworks for decision making under uncertainty have been increasingly applied to water resources management [40,41], major challenges remain, such as dealing with trade-offs between multiple goals and addressing various sources of uncertainty in contested political processes [4]. There are many barriers to the implementation of participatory decision support approaches—limited capacities in local institutions, associated costs, implementation time, the challenge of designing an adequate methodological framework, and the sustainability of the process in the long term [28]. In addition to these barriers, the knowledge and methods to tackle uncertainties should be considered in learning and decision-making processes [42].

If the goal is to avoid collective action dilemmas, the analytical tools used to evaluate potential water management interventions must, beyond basin-level consideration, align with river basin sub-jurisdictions and sectorial interest [15]. The complexity of this challenge is multi-fold:


The Robust Decision Support (RDS) framework proposed by Purkey et al. [46] consists of an iterative bottom–up process with active stakeholder participation where decisions are supported by the use of water resources models, which are accompanied by a strong process of local capacity building. Case studies for RDS include urban water management in the metropolitan region of La Paz/El Alto, Bolivia [47], evaluation of climate change impacts in a large basin in northern Patagonia, Argentina [36], and implementation of an IWRM planning process in the Yuba River basin, California [46,47]. This paper presents innovations in the RDS participatory framework as a contribution to Bolivia's National Watershed Policy, specifically in the formulation of the Rocha River Basin Master Plan (PDC in Spanish), which is located in the Cochabamba Valley, Bolivia. This effort took place within Bolivia's unique historical and political context related to water management. Our methodological approach consists of combining the RDS framework for water resources management [46] to extend Bolivia's guiding framework for the formulation of river basin master plans. We address the following questions: How does the water resources system model respond to water-related decision-making processes and institutional governance design at a range of scales within a river basin? How does the water resources system model contribute to the development of effective water planning instruments?

#### **2. Materials and Methods**

#### *2.1. Study Area*

The Rocha River Basin is located in the department of Cochabamba in the Plurinational State of Bolivia and contains all or part of 25 rural and urban municipalities (Figure 1). The basin has an area of approximately 3699 km<sup>2</sup> and a population of almost 1,300,000 people (13% of the country). From the hydrological point of view, it is comprised of three sub-basins: Rocha, Maylanco, and Sulty (also known as Valle Alto). The Rocha and Maylanco sub-basins—for decades predominantly agricultural areas—are rapidly urbanizing, including cities such as Cochabamba, Sacaba, Colcapirhua, Vinto, Tiquipaya, and Sipe Sipe, which make up the greater Cochabamba metropolitan area. In contrast, the Sulty sub-basin is mainly rural, with agriculture as its primary economic activity. Climatically, 80% of the annual precipitation is concentrated between the months of December and March (rainy season), 2% between the months of May and August (dry season), and the remaining precipitation in the months of the transition seasons (April, September to November) [48]. Annual rainfall varies between 300 and 900 mm, wetter to the north and the east, where most of the water supply reservoirs and current and potential sources of inter-basin transfer are located (Figure 1).

**Figure 1.** The Rocha River Basin and its geographical environment related to the urban area, agriculture, and the different hydrographic elements.

Challenging climatic conditions are likely to worsen under projected climate conditions. Previous studies have developed climate change scenarios on a basin scale for the 2020–2050 horizon, based on Intergovernmental Panel on Climate Change (IPCC)- 5th Phase of the Coupled Model Intercomparison Project (CMIP5) climate models prioritized based on their capacity to capture long-term historical (observed) climate variability attributes in the period 1981–2005 [48]. According to the Institut Pierre-Simon Laplace Climate Model 5A Low-Resolution (IPSL-CM5A-LR) [49] model for Representative Concentration Pathway 8.5 (RCP8.5) [50], the annual precipitation could be reduced by 7%. In some months of the rainy season (February and March), precipitation could be reduced by up to 30%. In the remaining months of the rainy season (December and January), it could increase by up to 11%. The dry season could be even drier with reductions of up to 77%. In the transition season, particularly between September and November, reductions in precipitation of between 10 and 20% are also expected. The average temperature could increase by 1.1 ◦C.

Given these challenging climatic conditions, it is not surprising that the Rocha River has historically been affected by water supply problems. Limited water availability due to the prevailing semi-arid climate as well as long-standing conflicts over access, governance, and environmental degradation contribute to the basin's water-related challenges. Cochabamba, Bolivia's third largest city, has experienced conflicts over the expansion of water access, particularly in the rural areas, highlighting the complexity of rural–urban hydrosocial relations [51] where established water users confront emerging water use communities. The Cochabamba Water War (*la Guerra del Agua*) emerged as a conflict between a centralized, foreign and private water company and urban residents over water tariffs that increased by as much as 200% [52,53], internationally recognized as a "David versus Goliath success" [53,54]. Empowered by the success of the Water War, the peri-urban water committees (*comités de agua*) moved toward more decentralized, small-scale, and autonomous water management led by community-owned supply systems [52]. However, this shift led to large disparities in water access in many extensive areas of the basin where small-scale options based on the water availability in the immediate environment were no longer compatible with the current levels of water demands. In response, new centralized water transfer projects from neighboring basins [51] with favorable quantity and quality conditions emerged as options [55]. The construction and expansion of some water transfers is currently underway; yet, local water managers also seek short-term solutions [52] such as drilling wells to extract water from aquifers. In addition to these problems, the region is exposed to hazards such as floods [56], landslides, debris flow, and droughts. The quality of surface water [57] and groundwater is degraded by the direct discharge of domestic and industrial wastewater, which is later re-used in irrigation [58–60]. In addition, the absence of effective land planning policies has been causing unregulated urban growth to the detriment of agricultural areas, aquifer recharge zones, and national parks [61]. This is hardly the setting for successful river basin-scale decision making, more so in the absence of sufficient comprehensive basin-scale information.

#### *2.2. Bolivia's Basin Plan in the Context of the River Rocha Basin Planning*

Nonetheless, similar to other countries, Bolivia has progressively adopted the river basin as the spatial domain for water management. However, it is worth noting that Bolivia's original Integrated River Basin Management (IRBM) approach adopted more of a terrestrial focus rather than a focus on water resources. Small-scale actions such as soil conservation, forest conservation, afforestation and reforestation, flood control, and bank protection civil works were the focus. For instance, the 1991 Cochabamba's Integrated River Basin Program (*Programa de Manejo Integral de Cuenca*, PROMIC) [62] was a national IRBM reference in the implementation of projects aimed at reducing local damage from flood events in prioritized river basins. This created an expectation that large-scale river basin plans would justify small-scale, community-level interventions.

This expectation aligned with customary practices in the Andean region of Bolivia known as customs and habits (*usos y costumbres*) that establish a relationship between water management and existing local governance in communal territories [63,64]. In this context, "water belongs to the territory and the territory belongs to the community" [63,65]. However, by the 2000s, Bolivia expanded the initial terrestrial-oriented scope to incorporate the social, environmental, and sectoral dimensions in water resources management [66]. The combination of the IWRM and IRBM approaches attempt to represent both the socioeconomic dimension and the natural resources conditions of the river basin, broadening the national concept of water governance beyond strictly local imperatives. However, the IWRM and IRBM approaches may not offer an appropriate representation of water management for complex community-managed water supply systems [67] upon which large-scale water resources management intervention are superimposed.

Bolivia's Ministry of the Environment and Water (MMAyA for its Spanish initials) has developed the conceptual framework and the national policy for IWRM and IRBM through the National Basin Plan (PNC in Spanish), which was promulgated in 2006. The PNC framework seeks to deliver solutions to integrated land and water-related problems with intervention justified within a PDC. The PDC is a planning instrument aimed at establishing intergovernmental and intersectoral coordination to develop water resource governance [68]. An early effort at PDC development occurred within the Rocha River Basin before systematic learning from other implementation experiences became available [69]. The Cochabamba Departmental Government originally formulated strongly IRBM-oriented and less so IWRM-oriented guidelines for the Rocha River Basin in 2014 [70], hampering the implementation of the proposed plan, as there was little buy-in among the 25 basin municipalities to implement large-scale water management interventions. In the Cochabamba Valley, competing sectoral and regional (i.e., upstream vs. downstream) interests such as household water users, rural communities, and civic organizations in urban areas make intersectoral coordination extremely difficult to achieve. As a result, the MMAyA and the Cochabamba Departmental Government decided to update the planning instrument by formulating a package of medium- and long-term actions based on three main factors:


#### *2.3. Proposed Approach*

The RDS framework developed by Purkey et al. [23] and Bolivia's guiding framework of the PDCs [11] provided our methodological approach for the formulation of a river basin master plan supported by participatory water resources systems modeling (Figure 2). The RDS framework has two phases: (i) preparation and formulation and (ii) evaluation and agreement. The first phase has six steps to identify the current and future vulnerability of the system. The second phase is a three-step participatory-driven process for the assessment of different management options leading to the identification of robust actions (actions that can satisfy disparate objectives under the assumed uncertainties). An essential step of this framework is the formulation of the problem using the XLRM matrix [71], where (X) stands for the uncertainties, (L) stands for the management options, (R) stands for the analytical tools that relate the (X) and (L), leading to performance measures, and (M) is used to evaluate the potential options. Uncertainties (X) are generally not contentious, i.e., all interest groups can agree that climate change or demographic growth are uncertainties that have the potential to impact outcomes related to water management. In contrast, preferences related to sectoral or strategic actions (L) are contentious, as stakeholders can oppose strategies offered by others. Metrics of performance (M) are identified for each sector to evaluate the outcome of each strategy identified, the stakeholder's preferred strategy, as well as those offered by others. These metrics are independent of any strategy, enabling a sector-specific strategy to improve outcomes (M) defined by another sector.

This is the basis of trade-off analysis and compromise, which serves for the formulation of a participatory-driven analysis in the decision process.

**Figure 2.** Framework to develop a Basin Master Plan based on model-driven participatory-based decisions. The new approach (center panel) allows grounding and implementation of the National Basin Policy guidelines (left panel) [68] and the robust decision support approach (right panel) [46].

> The guiding framework of the PDC contemplates three stages: formulation, implementation, and evaluation and monitoring. Each of these stages has a step-by-step process with general guidelines (Figure 2). All stages rely on public participation of the basin's institutions and actors as an essential element to promote environmental governance. The formulation stage comprises an integral participatory appraisal that allows for the identification and prioritization of the main objectives of basin and water management, albeit without the benefit of detailed modeling and analysis. This stage also considers the identification, construction, and validation of actions to achieve the proposed objectives. Stakeholder participation is a key element for the PDC guidelines and the RDS framework. The institutional approach and the mapping of key actors are considered at the beginning of the process for both cases.

> The comparison of PNC guidelines with the RDS framework reveals several gaps within the PDC guidelines. These gaps reduce the capacity of the PDC guidelines to inform water management in Bolivia's river basins, including:


• The strategic actions lack quantitative performance indicators or measures (M) to help forecast progress toward medium- and long-term goals and objectives, to which models have a great potential to contribute effectively.

However, the RDS general template to approach decision processes requires adapting to the local particulars of water governance and institutional design. To use RDS effectively to support the development of a new Rocha River basin PDC, we oriented the framework toward the development of a participatory decision-making process. A novel component of the RDS process was the introduction of 'hard coupled' decision interfaces to the WEAP model to support participatory forums. This allowed diverse institutions, interests, and organizations—that benefit or are affected by the decisions of the basin's intervention to interactively explore the medium- and long-term implications of water management options, recognize disparities between their own and others' frames of reference, and share ideas to build a Master Plan. The implemented framework, shown in Figure 2, is detailed in the list below:


scope, reference costs, and social and environmental conflicts that may make their implementation unfeasible.

	- a. It allows the consideration of action proposals from the different territorial levels. It provides an overview of the coherent actions' prioritization and implementation, recognizing the present biophysical, sectoral, and territorial interconnections that make the basin an indivisible planning unit.
	- b. The indicators (M) that consider a system of hierarchical analysis units, which allows a nested measurement between working scales, and where the basin domain is represented by sub-units of different scales, starting from indicators in basic modeling units (i.e., micro-basin, irrigation zone, urban demand unit).
	- c. The specific objectives for the prioritized problems must be articulated with the general principles established in the national policy and planning instruments such as the PDC.
	- d. Their application is not limited only to the formulation stage but also to the other stages such as monitoring or follow-up. In the formulation stage, it allows a comparative and participatory analysis to be made over time, making it possible to differentiate between medium- and long-term horizons. In the monitoring and follow-up stage, it will have the capacity to review the fulfillment of goals and potential strategic adjustments and redirection.
	- e. The type and scope of intervention actions that become relevant in the performance of objectives and indicators at the intervention scales of a basin plan. Through the PDM, it is possible to evaluate the multiple effects (positive or negative) of the interaction of the various interventions.
	- f. With respect to both frameworks, the set of quantitative indicators that can operate at various scales and the use of an interactive decision panel connected to a water resources system model were important innovations. Co-designed with stakeholders, these innovations create a common language and interactive feedback of model runs to specific requirements.

At the implementation stage, intervention actions can be reviewed and adjusted as new redirections are needed. The water resources system model can be updated with new information to help improve its performance. To make this effective, it is necessary to build local capacity in the management of all the tools that constitute the PDM.

#### **3. Implementation of the Proposed Approach**

The results of the implementation of the proposed approach in the formulation of the Rocha River PDC are presented below.

#### *3.1. Strategic Modeling of the Basin's Water Resources*

The MMAyA and the Development Bank of Latin America (CAF) developed a previous instance of the Rocha River basin model using WEAP [48]. The modeling process included the following in the collection and systematization of available data: participatory co-development of the model, fieldwork, definition of scenarios, and discussions about the benefits of using models in support of river basin planning. The WEAP model of the basin includes hydrology, water demand and supply, rules of operation, use rights, and water quality in terms of organic contamination of the main rivers. The model has a monthly time step for a historical horizon of 1980–2015 and a prospective period of 2020–2050. Hydrology was implemented using WEAP's Soil Moisture Model (SMM) [35], which is a one-dimensional model based on the notion of water transfer between two buckets: an upper bucket representing the root zone and a lower bucket representing deep storage or regional aquifers (when applicable). These two buckets represent the dynamics between evapotranspiration, surface runoff, interflow, and percolation for each basic modeling unit (catchments). The model allows dividing the basin in catchments in a semi-distributed way. The SMM is forced with climate data such as precipitation, temperature, relative humidity, wind speed, and insolation. Climate data from the Bolivian Water Balance [72] were used as input. The SMM is also used to calculate irrigation demands in agricultural areas, which are obtained as deficits of water required to maintain soil moisture within desired boundaries following crop type and calendars and other technical restrictions such as distribution or application efficiency and water allocation priority.

The catchments were delimited based on the location of reservoirs, water use points, transfer basins, wastewater discharge, calibration points, and topographic transition zones between mountain and valley areas (Figure 3). To parameterize the hydrological model, vegetation coverage, slope, and geology were considered. The vegetation cover was classified with a Landsat 8 image (January 2018) for the following categories: agriculture, forest, dispersed vegetation, temporary flood areas, urban area, scrubland, badland, and water bodies. The slope was obtained from a HydroSHEDS digital elevation model (DEM) [73] for the following ranges: sloping (<10%), strongly sloping (10–15%), moderately steep (15–30%), steep (30–60%), and very steep (>60%). Using the 1:100,000 scale geological map from Bolivia's Geological Mining Service, it was possible to differentiate the Quaternary deposits. We used this information to identify the zones of recharge and the three aquifers (Sacaba Valley, Cochabamba Valley, and Alto Valley) with high probability of groundwater occurrence.

For the modeling, the 48 existing reservoirs were considered (Figure 3), of which 90% have a storage capacity of less than 1 Mm<sup>3</sup> and are used for small-scale irrigation. In the rainy season, storage is prioritized, and in the transition and dry seasons, water volumes are released for irrigation according to the agricultural calendar. The reservoirs for human consumption are operated according to water demand. Information regarding physical characteristics and location was collected from the national inventory [74] and potential investment project database.

Household water demand was represented in WEAP as a function of population, per capita consumption, and average losses reported by the water distribution systems of each municipality. In the Rocha and Maylanco sub-basins, there are two types of service providers: Water and Sanitation Service Providers (EPSA) and Small-scale Local Operators (OLPE) [75]. The EPSA is an entity that depends on the municipal government and whose service area is limited to the urban area, and it is regulated by the corresponding authorities. The OLPE are autonomous community organizations or cooperatives that provide water service to their areas of immediate territorial occupation. For the modeling, a distinction has been made between the coverage of EPSA and OLPE. A total of 26 water demand nodes have been represented (Figure 3). Water quality in the main section of the Rocha River was represented in the model as the organic contamination constituents that determine biological oxygen demand (BOD) using the Streeter–Phelps model [76] and water temperature of the river. The modeled river section and assessment points are shown

in Figure 3. The assessment points are located downstream of the wastewater treatment plant (WWTP) discharge.

**Figure 3.** Key components of the Rocha River basin WEAP model. The numbers in parentheses in the legend refer to the number of objects simulated in WEAP.

To determine the irrigation demand and allocation, we calculated the water balance using the Food and Agriculture Organization of the United Nations (FAO) approach, which is based on reference potential evapotranspiration and crop coefficients (Kc) [77]. We defined the irrigation area with the digitalization of high spatial resolution images. We also assimilated information from previous studies to characterize the crop schedule, agricultural calendar, and extent and type of irrigation systems [78–80]. The water balance was implemented at an irrigation zone scale that can include one or several irrigation systems. The catchments that have irrigation are shown in Figure 3. Irrigated zones can cover more than one catchment; for this reason, the total number of modeled zones is 74, while the number of irrigated catchments is 70.

For the model calibration, we used the flows measured in Misicuni, Taquiña, and Puente Cajón stations, and the storage levels in La Angostura reservoir (Figure 3). In the drainage area of these stations, the flows are modified by the extractions and storage in reservoirs. Therefore, model calibration is not only limited to optimizing SMM parameters but also to the operation rules and extraction volume. The initial parameters were defined with the reference values [81] and then adjusted until an acceptable performance was achieved. At the Misicuni station, a satisfactory Nash–Sutcliffe index of 0.67 and a very good Percent Bias (PBIAS) of −4.99% were obtained [82]. The determination coefficient (R<sup>2</sup> ) obtained between the modeled and measured storage volumes at La Angostura is 0.76, which indicates a good performance. The flows measured at Taquiña and Puente Cajón stations are limited. These stations were used to validate the results obtained in La Angostura and Misicuni.

Climate change scenarios in terms of precipitation and temperature for the 2020–2050 future horizon were generated from General Climate Models (GCMs) with downscaling using the non-parametric K-nearest neighbor (K-nn)-Bootstrap statistical method [83]. This modeling process also identified other future uncertainties such as land-use changes and population growth.

#### *3.2. Participatory Process of Identification and Prioritization of Problems*

The establishment of a formal Inter-Institutional Platform in the Rocha River Basin was a slow process plagued by many challenges. This is due to the large number of municipal jurisdictions in the basin (25 in total), the complexity of the problems, and the interests of each territorial level. The Board of Directors and the Technical Council were legally constituted only at the time of identifying intervention actions. The previous steps of the proposed approach involved representatives of municipalities, universities, service providers, and irrigation associations, some of whom became members of the Technical Council. To identify the problems, 15 workshops were organized, which were followed by the collection and assimilation of additional available information on biophysical and sociocultural characteristics, land occupation, hydrological risks, water management, management of life systems, and the institutional framework. The results were presented at a workshop where stakeholders prioritized 11 strategic problems in the basin (Table 1).

**Table 1.** Problems prioritized by stakeholders in the Rocha River Basin and incorporated in the scoping in WEAP model (shaded in gray).


The participatory spaces and the new information helped to scope the development of the WEAP model. For problems related to territorial and terrestrial environmental aspects, spatial analysis was also carried out using Geographic Information System (GIS) and remote sensing methods. The resulting modeling tools contributed to the quantitative characterization of eight of the problems for the historical and future condition representing identified uncertainties in the strategic modeling process [48] (Table 1).

#### *3.3. Identification of Intervention Actions*

The starting point was the review of the implementation status of current sector plans (e.g., Water and Sanitation Master Plan in the Metropolitan Area), the inventory of Sulty

pre-investment projects of territorial entities, and the coordination with other planning instruments in the process of formulation (e.g., Valle Alto Water Master Plan). In addition, through participatory processes (workshops and meetings with local institutions), new intervention actions were identified in response to the prioritized problems. Table 2 shows the summary of intervention actions to improve conditions associated with water quantity and quality, which can be implemented in the WEAP model. Each has one or more conceptual design options according to its spatial scope, type of technology, and planning level (e.g., pre-feasibility, final design).

Siches reservoir Infrastructure 1 Canllamayu reservoir Infrastructure 1 Kangani reservoir Infrastructure 1 Pucara Mayu reservoir Infrastructure 1 Revitalization of irrigation systems Water demand management 15 Optimization of irrigation systems Water demand management 15


All Source priority Water management

**Table 2.** Summary of intervention actions (L) identified to improve water management in the Rocha River Basin.

The type of actions identified to solve the problem of water scarcity is focused on new water transfers, new reservoirs, and efficiency measures. The water transfers have a multipurpose approach (Misicuni, Cordillera Norte and Khomer Khocha) to supply drinking water, irrigation, and hydropower generation. Due to the complexity of these actions, their implementation is conceived in phases and with different options for their implementation. The number of phases or implementation options are shown in Table 2. In the case of actions referred to the management of irrigation demand, revitalization, and technification, the options refer to the sites of implementation. In the Rocha and Maylanco sub-basin, the six irrigation zones are the options, while in Sulty, there are 15 options. The revitalization consists of improving the current traditional irrigation systems, which includes modernizing the infrastructure (e.g., canals) and irrigation management (operation and maintenance). Technification refers to reducing the demand for water by using technology for the application of irrigation. In terms of water quality management, the options refer to the number of WWTP (11) considered for construction or improvement. In addition, each WWTP has two technology options for a total of 22 options.

Allocation priority Water management

#### *3.4. Design of a Participatory Decision-Making Model (PDM)*

The Participatory Decision-Making Model (PDM) of the basin integrates the WEAP model incorporating uncertainties and intervention actions, and an interactive dashboard that provides user-friendly visualizations to create decision packages and to navigate the resulting performance indicators across the different objectives and levels of disaggregation (Figure 4). The uncertainties implemented in the WEAP model were the climate change scenarios for the IPSL-CM5A-LR model, population growth, and land-use change. The possible intervention actions are summarized in Table 2. The control panel designed in Microsoft Excel facilitated the interaction of the WEAP model with stakeholders. In the

panel, they could select the intervention actions, uncertainties, water allocation, and the planning horizon (medium—2025, and long term—2040). The selected decisions are sent to the WEAP model through a Microsoft VBA script and then, the model is run. The stakeholders were able to visualize the performance of the intervention measures by means of multi-scale indicators quantified at a range of scales from the basin, sub-basins, microbasins, aquifers, irrigation zones, urban demand units, and river reach. The platform allowed for the visualization of these indicators at the precise scale of interest of each actor. The indicators used are shown in Table 3 where each has the unit of measurement, basic unit of spatial analysis, the upper-level spatial scale, and the function of aggregation. For example, for drinking water, the basic unit of spatial analysis is the urban demand unit, the next level is the sub-basin, and the subsequent level is the basin. The aggregation function depends on the unit of measurement; in the case of volume, the function is the sum. A cost-efficiency indicator was also considered. Multiscale indicators enabled a diverse audience of stakeholders to explore the positive and negative interactions of intervention actions, identify disparities in action performance across scales, and interactively compare different actions that help identify and mitigate emerging regional or sectoral conflicts.

**Figure 4.** Conceptual design of the participatory decision-making model for the Rocha River Basin.



**Table 3.** *Cont.*

<sup>1</sup> The numbers in parentheses refer to the quantity of basic units. There are 26 urban demand units, 74 irrigation zones, 3 aquifers, and 9 river reaches. <sup>2</sup> The numbers in parentheses indicate that there is a basin, 3 sub-basins, and a main river.

#### *3.5. Evaluation of Intervention Actions*

First, each intervention was evaluated individually using the participatory decisionmaking model considering efficiency indicators for different uncertainty scenarios and their performance in the medium (2025) and long term (2040). Second, stakeholders were divided into two groups that through an iterative process identified and prioritized action packages for the time horizons considered and the collection of stakeholder-specific goals (Figure 5). Public institutions such as the MMAyA and different operational institutions of the departmental government (Departmental Watershed Service, and Directorate of Water Management and Basic Services) participated in this process. These institutions work on pre-investment and investment planning related to water (Figure 5). During the group sessions, participants offered four main insights:


During the exercise, not only technical and financial indicators were analyzed but also potential social and environmental conflicts that may make implementation unfeasible. Aspects such as water distribution and allocation priorities were also addressed.

#### *3.6. Proposal and Approval of the Strategic and Programmatic Framework*

The prioritized actions were organized and structured in lines of action and strategic lines that focused on institutional, sectorial, and community interventions, always considering the principle of the basin approach. The PDC gives action plans that must be pursued in the medium and long term to make substantial progress in the well-being of the inhabitants and their way of life. It also provides an integrated framework to strengthen

financing processes by coordinating the investments that the territorial authorities make in water management issues, which are currently carried out in a fragmented manner and generate environmental conflicts. Coordination allows for the expansion of the scope of benefits through the coherent use of resources. The PDC document was agreed to by all key stakeholders and approved by the Technical Council. Moreover, the PDC has been declared a departmental law (*Decreto Departamental* No 4544, 18th of September 2020) [84]. A legal framing means the PDC actions are now binding and must be fulfilled by the sub-basin jurisdictions, hence leveraging financial resources for action implementation and increasing the PDC's visibility across planning scales. Table 4 shows the plan in detail, which has five strategic lines and 82 interventions. The total investment for the period 2020–2040 amounts to USD 1.5 billion; for the current population, this is equivalent to approximately USD 58/person/year. Most of the total investment cost of the plan (98%) corresponds to the strategic line of water management, which has an implementation horizon of 20 years. For the first five years (planning horizon of the PDC), priority was given to the construction and improvement of the WWTP, the construction of water conduction infrastructure of the Misicuni, starting with detailed pre-investment studies for new water transfers (Cordillera Norte and Khomer Khocha), water demand management, and preservation of the drainage area of current and potential transfer basins (water reserve zone). The 2026–2040 horizon of the plan contemplates the construction and putting in operation of the new water transfers, the extension of the water distribution network and sanitary sewerage, and the new WWTP. With this package of actions in the medium term (2025), progress will be made in improving the quality of water in the Rocha River and reducing unsatisfied demand in the Rocha and Maylanco sub-basins. The solution to water scarcity problems in the entire basin depends mostly on new water transfers, which will require decades to implement. The other strategic lines have an implementation horizon in accordance with the times established in the national basin policy. These lines seek to improve the sustainable management of micro-basins, institutional strengthening for water resources management, improving information and knowledge, and water culture. The baseline and plan indicators agreed upon for the 2020 and 2040 horizons are described in detail below. The baseline is the current system of water resource management projected for different future time horizons.

**Figure 5.** Photograph of the workshop for the evaluation of medium- (2025) and long-term (2040) intervention actions using the participatory decision-making model.


**Table 4.** Summary of the Rocha River Basin master plan agreed with the Inter-Institutional Platform.

> 1 Conversion rate: 1 USD = 6.96 BOB (reference year 2019).

#### 3.6.1. Baseline Indicators

Projections until 2025 and 2040 indicate that the basin would have a population of 1,518,502 and 2,228,632 people, respectively. The irrigable area is approximately 40,002 ha, which was assumed to be constant for future time horizons. Performance indicators up to 2025 (Table 5) indicate that the current supply system could supply only 66% of the population and optimally irrigate 19,443 ha. The annual volume of unsatisfied demand in the basin would be 340.3 Mm<sup>3</sup> . The water quality in the Rocha River modeled by BOD would be 153 mg/L, which is approximately four times higher than the limits established in national legislation. The Sulty sub-basin would have an unsustainable exploitation of groundwater. By 2040, these indicators could worsen (see Table 5); for example, the unsatisfied demand could reach 362.7 Mm<sup>3</sup> . The current supply system would only have the capacity to supply 59% of the population, and irrigation conditions would remain similar to those encountered in the medium term.

#### 3.6.2. Indicators of the Agreed Plan

In this paper, we present the aggregated results at the basin and sub-basin scale; however, in the PDM, the results can be visualized at the basic service area unit scale of modeling. Table 5 shows the expected level of improvement in the main performance indicators up to 2025 (medium term). The plan expects to reduce the annual unmet demand in the basin by 64.9 Mm<sup>3</sup> . The population benefited by the package of actions is 403,647 people. In irrigation, it is expected that the deficit will be reduced by 58.1 Mm<sup>3</sup> and the optimal irrigated area will increase by 1768 ha. In water quality, a reduction of BOD by 101 mg/L is expected—that is, a reduction of more than 70% of the current levels of contamination, and at the same time, an increase of approximately 20% in the flow during the low water season to improve the environmental functionality and the assimilation capacity of the Rocha River.

Table 5 shows the indicators expected in the long term (2040). The package of actions will reduce unsatisfied water demand by 227 Mm<sup>3</sup> in the basin, thus increasing access to safe water by 834,049 people and increasing the area under optimal irrigation by 10,497 ha. The increase in the supply of surface water will make it possible to achieve the sustainable use of all the aquifers in the basin—that is, by exploiting them within the limits of their natural recharge. In terms of water quality, a reduction in BOD of 125 mg/L is expected as well as an increase in the flow of the River Rocha in the months of low water by 359 L/s.

In the Maylanco sub-basin, indicators presented negative changes up until the year 2025. For example, drinking water coverage reduced by 3759 people, and unmet demand increased by 0.4 Mm<sup>3</sup> (Table 5). This is mainly due to the expansion of the water distribution system from Misicuni to the municipalities of the Rocha sub-basin. In the baseline scenario, the Maylanco sub-basin would be benefitted from higher water volumes; however, the proposed plan will reduce the water volumes given the expansion of the water distribution system. These changes are also reflected in the increase of the groundwater use rate by 52%. In the Pucara and Abra sections (Maylanco sub-basin), there is a small BOD increase due to the reduction in volumes delivered from Misicuni, which also reduces the river's self-purification capacity.


#### **Table 5.** Performance of the proposed actions (2025 and 2040).

Cost-effectiveness analysis was part of the performance indicators used in the process of evaluating intervention actions. Some sector interventions identified before the formulation of the PDC presented unfavorable values in terms of cost-effectiveness, especially water transfers such as Khomer Khocha. The intervention assessment process explored options to make interventions feasible within the framework of the basin planning. The multipurpose approach of the actions has been an option that allowed obtaining more encouraging indicators of economic efficiency. Table 6 shows the cost-effectiveness indicators of the agreed plan for the different time horizons. In the case of drinking water, the highest value is related to the implementation of the Misicuni additional water transfers. In irrigation, the indicators are well within the Bolivian viability threshold of 10,000 USD/ha. The highest value in the horizon 2035 is related to the implementation of the Khomer Khocha water transfer that benefits mainly irrigation in the Sulty.

**Table 6.** Cost-effectiveness indicators of the agreed plan for water supply actions for drinking water and irrigation.


#### **4. Discussion**

*4.1. How Does the Water Resources System Model Respond to Water-Related Decision-Making Processes and Institutional Governance Design at a Range of Scales within a River Basin?*

In the Rocha River Basin, all available water is in use, which means that there is no free water for new uses [85], and management decisions are fundamentally a redistribution between existing users. According to the modeling, the demand for water is potentially three times the natural availability of the basin; hence, there is extensive unsafe reuse and a growing dependence on external sources. The implementation or intervention actions that consider expanding the use of the basin's supply sources generate conflicts. In this context, the formulation of the basin plan faced many challenges. In the decisionmaking process, analytical modeling tools helped to recognize, quantify, and differentiate the local and regional impacts of water reallocation. Our multiscale approach allowed the diverse frames of reference of basin stakeholders such as small OLPES, individual irrigation zones, and municipal jurisdictions to identify their interdependence with other sector-specific goals or regions, including those operating at different scales, and engage in productive negotiations.

An example to illustrate this is the Siches Reservoir (Table 7), which in the Sulty sub-basin generates an incremental irrigation area of 797 ha, but in the Rocha sub-basin, the optimal irrigation area is reduced by 222 ha. The drainage area of this potential reservoir is one of the main tributaries to the La Angostura reservoir, which in turn supplies irrigation areas in the Rocha sub-basin. To compensate for the conflicts identified, iterative exercises (as shown in Figure 3) were carried out. These exercises contributed to the construction of a package of actions to achieve the full spectrum of objectives and goals (Table 7). In addition, it was possible to incorporate climate change into decision making with the model. In the process of building the action set, stakeholders were able to visualize results of how climate change in the long term could lead to problems in the reliability of water supply systems. This is how actions to build resilience in water supply systems were assessed. Based on this exercise, the new Cordillera Norte water transfer action was formulated as part of the plan.


**Table 7.** Performance in irrigation of the intervention action of Siches Reservoir.

Likewise, the results of the model showed that water-use efficiency actions and expansion of current supply sources in the basin, while providing important gains in performance, cannot compensate for existing conflicts on their own or fulfill the medium- and long-term PDC main goals, such as universal and equitable access to safe water. Therefore, solutions to scarcity problems will also require several strategic actions that increase the volume of water transfers. Due to the multipurpose nature and regional scope, it will also allow substantial progress in the sustainable use of groundwater by supplementing current wells with transferred surface water. By improving the environmental conditions of the rivers, it will also provide greater capacity for assimilation by increasing the flow in months of low water from the return flows.

The analytical tool for decision-making in the basin makes it possible to co-develop strategies in participatory spaces with the interested parties. The use of models in conflict management is recognized in the scientific literature [86,87]. However, formalizing their use in decision-making within the framework of the basin plan had many challenges related to limited data, the complexity of traditional water management, the spatial and temporal scale of water supply systems, and some resistance to the use of models. At the beginning of the process, the main challenge was the lack of credibility for the use of the model in formulating the plan, as stakeholders indicated that the available data were insufficient to develop the watershed WEAP model. In addition, incorporating the model into planning meant adjusting the PDC's methodological guidelines, which also generated resistance in certain government institutions. To address these challenges, partnerships were formed with local universities to maximize the use of available data and advances, intensive fieldwork to collect irrigation data, discussion meetings with local experts, and workshops to share the inputs and results. The participatory modeling approach generated opportunities to iterate the process while interacting with local experts. By involving stakeholders early in the process, we were able to make adjustments and continuous improvement in the formulation of the river basin plan. As mentioned by Loucks et al. [45], the credibility of a model is very subjective, and the modeling, besides trying to represent reality, is also an attempt to formalize and guide those perceptions.

It is important to recognize that there are conflicts around intervention actions that cannot be negotiated through the support of models because they require arrangements in traditional, sectoral, institutional, and legal use rights. For example, water transfer actions generate conflicts in the territory of new uses [55]. Based on experiences of the Misicuni water transfer, this type of action generates the displacement of rural community settlements, the loss of productive land, and the drying up of wetlands and springs [51] with unjust compensation [88,89]. With these precedents, the transfer actions proposed in the watershed plan could face similar conflicts.

#### *4.2. How Does the Water Resources System Model Contribute to the Development of Effective Water Planning Instruments?*

Formulating and implementing river basin master plans in Bolivia is a learning process built on experiences from strategic basins of varied geographic contexts. For instance, the first master plan of the Rocha River Basin showed progress in elaborating the diagnostic study and the strategic guidelines. It also made progress in the implementation of IRBM actions in priority micro-basins and flood mitigation infrastructure. Its water management strategy focused on water access and use. However, it was up to sectorial plans (e.g., the Water Master Plan) to identify the actions needed to fulfill the master plan. The initial strategy lacked quantitative indicators and goals in its implementation horizon, which would have enabled monitoring and follow-up processes as well as connections with other plans at the national scale. For instance, the Bolivian Economic and Social Development Plan could have been a potential connection, since 'achieving universal access to water' is also one of its objectives. Since the formulation of the first basin plan, the Rocha and Maylanco sub-basins made significant progress with their Water and Sanitation Master Plan. However, the proposed actions were not aligned with other water uses, such as irrigation, since the Water and Sanitation Master Plan focused on water for human consumption. Lastly, the Sulty sub-basin made progress in managing financial sources for the preparation of the Water Master Plan.

The updating of the Basin Plan under the proposed approach was an opportunity to coordinate these sectoral interventions and the actions promoted from the different territorial entities that are not necessarily connected to objectives at the regional- and basin-level. In addition, the updated plan has a sequence of short-, medium-, and long-term actions, thus recognizing the potential effects of climate change. Its strategic and multisectoral approach made it possible to identify a wide range of co-benefits and compensation measures. The flexibility of the analytical decision-making tool will allow monitoring and follow-up as the plan is implemented, and if necessary, the goals could be adjusted according to new orientations in policy or new stakeholder priorities.

#### **5. Conclusions**

Our integrated approach lays the foundation to apply the RDS framework in water resources management, building upon and expanding on Bolivia's basin policy. The formulation of the River Rocha PDC led to formalizing the use of modeling tools in the context of river basin planning in Bolivia. However, we faced many challenges arising from the planning process to identify strategic actions, particularly those related to expanding the use of the basin's supply sources. We connected sector- and territorial-based actions across the river basin while engaging municipalities and the regional and national governments in the planning process. Our analytical modeling tools helped to identify and quantify the trade-offs between local and regional impacts that a new water use could generate, hence facilitating the decision-making process. Once we identified positive and negative effects of the proposed actions, the participants could negotiate and propose compensation measures within the PDC. We formulated a PDC considering short-, medium-, and long-term climate change scenarios, setting out measurable goals for each time horizon and establishing implementation phases for the proposed actions. Our iterative RDS process allowed for re-arrangements or changes within the water resources system modeling according to the stakeholders' inputs and needs.

The PDC covers a wide range of topics—from flood control measures to determining a financial plan for the implementation of strategic actions. As previously mentioned, the analytical tools are key to the river basin planning process. However, not all territorialbased actions are integrated in the water resources system model. Further research is needed to integrate territorial-based actions and water resources modeling for participatory decision-making process, hence minimizing potential conflicts in planning instruments such as the PDC. In addition, water resources system models alone may not recognize all water-related conflicts. Water resources modeling tools may be limited in their capacity to propose sound socioeconomic alternatives and/or compensations to people directly or indirectly affected by water-related projects. Although our water resources system model accommodated water volumes according to each implemented action in the river basin, our work has not directly addressed legal and institutional conflicts behind those actions. In addition to indicating the water availability through a detailed water budget, anticipating conflicts that might emerge in the strategy implementation phase could help with conflict management under scenarios of uncertainty. Moreover, the Andean region of Bolivia widely adopts *usos y costumbres* to ensure water rights in communal territories. The Cochabamba Valley is no exception to these historical customary practices. As properly observed by Hendriks [65], "water belongs to the territory and the territory belongs to the community". Further research is needed to intentionally intersect planning at the river basin level with water management in communal territories, addressing any mismatching policies that may escalate potentially divisive water governance approaches.

The RDS approach applied in the Bolivian context may facilitate IWRM implementation at the river basin scale by providing both rigorous water resources system modeling and effective stakeholder participation. Our proposed framework creates an opportunity for stakeholders to engage in the water management process and highlights the need of ensuring participatory processes to legitimate planning instruments for the river basin. We are comfortable that the approach taken addresses the general challenges facing IWRM and the specific context of water management in Bolivia. The key was allowing for cross-scale evaluation of the performance of different actions and a direct integration of watershed and water resources management interventions.

**Author Contributions:** Conceptualization, N.L.-Q., C.C. and H.A.; methodology, N.L.-Q., W.R., Z.G. and H.A.; software, N.L.-Q. and H.A.; validation, N.L.-Q., C.C. and H.A.; formal analysis, N.L.-Q., Z.G., F.Z., S.N., J.I. and H.A.; investigation, N.L.-Q., W.R., Z.G., F.Z., S.N., J.I., C.S. and H.A.; resources, N.L.-Q., D.P., M.E. and H.A.; data curation, N.L.-Q., F.Z., S.N., J.I. and C.S.; writing original draft preparation, N.L.-Q., C.C., D.P. and H.A.; writing—review and editing, C.C., D.P. and H.A.; visualization, N.L.-Q., J.I. and H.A.; supervision, H.A.; project administration, D.P.; funding acquisition, D.P. and M.E. All authors have read and agreed to the published version of the manuscript.

**Funding:** The "Pilot Program for Climate Resilience—Ministry of Environment and Water, Bolivia, grant number 01/2018" funded the formulation of the River Rocha PDC and the "Stockholm Environment Institute's Regional Engagement Funds" financed the preparation of the manuscript.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We gratefully acknowledge the support of the Pilot Program for Climate Resilience—Ministry of Environment and Water (PPCR-MMAyA), the World Bank office in La Paz, and the *Secretaría Departamental de los Derechos de la Madre Tierra* (including the *Servicio Departamental de Cuencas* and the *Dirección de Gestión de Agua y Servicios Básicos*). We would also like to acknowledge the support of some SEI consultants and the PPCR supervision team who assisted in the implementation of the proposed approach.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Risk Assessment of China's Water-Saving Contract Projects**

#### **Qian Li 1,2 , Ziheng Shangguan 3, \*, Mark Yaolin Wang 4,5 , Dengcai Yan 3 , Ruizhi Zhai <sup>6</sup> and Chuanhao Wen 1,7, \***


Received: 31 August 2020; Accepted: 24 September 2020; Published: 25 September 2020

**Abstract:** In order to alleviate the problem of water shortage, the Ministry of Water Resources of China proposed a Water-Saving Contract (WSC) project management model in 2014, which is similar to the Energy Performance Contract (EPC). In this context, this research aims to explore the applicability of China's WSC projects by risk assessment, and to help promote WSC projects in China. Different from traditional risk assessment, this paper takes into account the uncertainty of the EPC project's risks, and adopts the multielement connection degree set pair analysis to evaluate both the level and trend of the risks. The results show: (1) the overall risk of China's WSC projects is low, so WSC projects are very suitable for promotion in China. However, the overall risk shows a trend of decelerated ascent, which shows that there are some potential high-risk factors in China's WSC projects; (2) among the many risks of the WSC projects, audit risk, financing risk, and payment risk are at a high-risk level; market competition risk is at a medium-risk level; the remaining risks are at a low-risk level; (3) among the medium and high risks, audit risk, financing risk, and market competition risk have a trend of accelerated ascent, while payment risk has a trend of decelerated decline; in low risks, inflation risk has a trend of decelerated ascent, while the remaining risks have a trend of accelerated decline.

**Keywords:** risk assessment; water-saving; set pair analysis; China

#### **1. Introduction**

Since the reform and opening up, China's population has continued to grow. The process of industrialization and urbanization has accelerated, which has led to a gradual increase in water consumption. With global warming and the pollution of water resources caused by industrial development, the problem of insufficient regional water supply in China has become increasingly prominent. Insufficient water resource carrying capacity has become the main constraint in China's sustainable development [1]. Normally, the way to solve water shortage can be divided into the increasing water supply and saving water. For a long time, China has been relying on the construction of a transbasin water diversion project to solve the problem of insufficient water supply in the north, which is a typical supply-oriented solution while throttling has been relatively ignored [2]. Therefore, in order to achieve the purpose of saving water, the Ministry of Water Resources of China (MWRC) began

to vigorously implement Water-Saving Contract (WSC) projects in 2016 [3]. At present, China's WSC project is still in the initial stage, and it faces many uncertain risks. In order to promote the development of the WSC projects in China, it is of great significance to assess the risks of the projects and formulate relevant policies to reduce them.

#### *1.1. Profit Model of the WSC Project and Its Stakeholders*

The WSC contains a specific water-saving target which is beneficial to the water user. By providing advanced and applicable water-saving technologies, the water-saving service operators carry out a technological transformation, establish long-term management mechanisms, and eventually pay the full cost by water-saving benefits, while water users can also share the benefits. This is a market-based water-saving management model [4]. The profit model is shown in Figure 1. The project involves three subjects, namely: the government, water-saving service operators, and water users. The government mainly assumes the role of policy guidance. Water-saving service operators provide services to water users through technological innovation to alleviate the pressure on water consumption. Water users are the ultimate beneficiaries. Here, we mainly discuss the risks encountered by water-saving service operators.

**Figure 1.** Profit model of water-saving contract projects.

#### *1.2. Enlightenment of the Energy Performance Contract's (EPC) Risk Assessment to WSC*

At present, there are few studies on the risk assessment of the WSC project, and there is no uniform analytical framework and research method for systematically assessing the risks. However, in essence, the WSC proposed by the MWRC is similar to the Energy Performance Contract (EPC); therefore, when analyzing the risks of China's WSC projects, the research results of EPC projects can be used for reference.

In the research of the EPC project, Mills et al. identified the inherent risks of EPC projects and divided them into five categories, namely, economic risks, environmental risks, technical risks, operational risks, and measurement and verification (M&V) risks [5]. On the basis of Mills et al.'s research, Lee et al. added financial risk, project design risk, and installation risk to the research of risk assessment in the EPC project, and identified the key risks of EPC projects through questionnaires. They believe that the key risks to energy service companies are possible payment defaults of hosts after installation, the uncertainty of baseline measurement, and the increase in installation costs in EPC projects [6]. Hu and Zhou further refined the risks of the EPC project in China. They considered that EPC project risks include political and legal risks, market risks, technical risks, management risks, financial risks, project quality risks, and customer risks [7]. Duan et al. constructed a life cycle analysis framework of EPC projects containing four stages, namely the contract signing stage, investment stage, implementation stage, and benefit-sharing stage [8]. Based on the framework proposed by Duan, Wu et al. used an improved analytic hierarchy process (AHP) to determine the weight of various risk indicators of the EPC project in China, and established a risk evaluation model using a fuzzy comprehensive evaluation method [9]; Huang et al. combined AHP and gray evaluation theory to construct a gray multilayer evaluation model of the EPC project's risks, and analyzed the high-risk factors of China's EPC project [10]. Garbuzova-Schlifter and Madlener systematically studied the

common risk factors and causes of risk associated with EPC projects executed in three Russian sectors: (1) industrial; (2) housing and communal services; (3) public. They also conducted a quantitative assessment of risks based on an AHP approach, and proposed a widely applicable risk management framework for Russian EPC projects [11]. Valipour et al. divided EPC project risks in Iran into six categories: design risk, market risk, political risk, environmental risk, construction risk, and political risk, and assessed the level and occurrence probability of the EPC project's risks using Analytic Network Process (ANP) method. Their results indicate that political risk and design risk are the most significant types of risk in Iran's EPC projects [12].

#### *1.3. The Risk Characteristics of the WSC Project and the Determination of Its Research Methods*

The study of EPC projects provides effective analysis frameworks and research methods for the risk assessment of the WSC projects However, unlike EPC projects, WSC projects have the characteristics of the long project cycle, relatively low return on investment, and weak liquidity. These characteristics may cause the risk level of the WSC projects to change significantly during the whole life cycle of the project [4]. In addition, as WSC projects are still in their infancy in China, the level of the risk is prone to change under the influence of national development strategy, overall economic level, management technique, and market resource allocation [13]. Therefore, simply analyzing the level of risk can no longer meet the needs of risk assessment in the WSC project, and the trend of risk should also be analyzed.

In the previous studies on EPC projects, the trend of risk was hardly considered, and its uncertainty makes it difficult to assess. Set pair analysis (SPA) has strong adaptability in dealing with the interaction between certainty and uncertainty in the system [14]. Cui et al. built an evaluation model to quantitatively evaluate and diagnose the carrying capacity of regional water resources under uncertain conditions by applying set pair analysis [15]. Gao et al. put forward a model based on set pair analysis about information risk evaluation. This model can not only divide the extent of the information risk, but it describes the trend of the information risk. It could describe the information risk from static and dynamic [16]. Zheng et al. used the set pair analysis method to analyze the safety of the tailing pond. Through set pair analysis, the development trend of the safety status of the tailings pond can be judged [17].

Based on the studies above, this article intends to use a literature analytic method to determine the risks existing in the WSC project's life cycle which was proposed by Duan [8], and uses multielement connection number set pair analysis to evaluate the level and trend of the risks. Finally, suggestions are made based on the characteristics of various risks to control and reduce them.

#### **2. Materials and Methods**

#### *2.1. Risk Identification in Water-Saving Contract Projects*

This article divides the life cycle of the WSC project into four stages according to Duan et al. [8]. The four stages are the contract signing stage, investment stage, implementation stage, and benefit-sharing stage. Then, the risks in each stage of the WSC project were sorted out through a literature review, as shown in Table 1.

#### (1) Contract signing stage

The contract signing stage includes the process of water audit, feasibility study, and contract signing. In the process of water consumption audit, if the water-saving service operator cannot accurately obtain the actual water consumption of the water user, the payback period will be lengthened [18]. The focus of the feasibility study is to evaluate and demonstrate the water-saving technique. If the technique obtained by water-saving service operators fails to reach the target, it will cause economic losses and waste of resources [19]. Li et al. consider that WSC projects have not yet formed a mature market-based management mechanism in China, which can easily cause malicious competition in the

industry [20]. As a result, this article determines the risk evaluation indicators of the contract signing stage, such as: information risk, technical risk, and market competition risk.

#### (2) Investment stage

Sustainable funding is an important guarantee for the implementation of the project. In the investment stage, water-saving service operators need to finance to ensure the progress of the project, and its financing channel mainly comes from commercial banks [21]. At the same time, due to the long payback period of investment in water-saving contract projects, the bank's interest rate may increase during the project, which will also affect the company's financing costs and reduce the final profit of water-saving service operators [7]. Therefore, this article determines the risk indicators of the investment stage, such as: financing risk and interest rate risk.

#### (3) Implementation stage

The implementation stage includes engineering construction, equipment procurement, installation, and commissioning [26]. In this stage, construction safety directly affects the process and construction cycle of the WSC projects [22]. In addition, some external factors also increase the risk of the implementation stage, such as: policy changes [20], force majeure [23], and inflation [24]. Policy changes may directly affect the enthusiasm of water-saving service operators, so as to affect the final quality of the project; the risk of force majeure will directly lead to the termination of the project; the impact of inflation is the same as the increase in bank interest rates, which will reduce the final profit of the project. In summary, this article determines the risk indicators of the implementation stage as: construction risk, policy risk, force majeure risk, and inflation risk.

(4) Benefit-sharing stage

After the implementation stage is over, it enters the benefit-sharing stage. At this stage, water-saving service operators are responsible for project operation and equipment maintenance. At the same time, it recovers investment costs and obtains reasonable profits by sharing water-saving benefits. Li et al. pointed out that the depreciation rate of the equipment would greatly affect the operating costs, so they suggested strict maintenance of the equipment [20]. The payment default of the water user is also one of the important risks in the benefit-sharing stage. If water users have weak credit awareness or cannot reach the predetermined water consumption due to their own economic problems, the investment of the WSC projects will not be recovered [6]. The fluctuation of water prices will affect the payback period of the WSC projects [25]. As mentioned above, this article determines the risk indicators in the benefit-sharing stage as: facility depreciation risk, payment risk, and water price change risk.


#### **Table 1.** Risk index system of WSC project.

#### *2.2. Multielement Connection Degree Set Pair Analysis*

#### 2.2.1. Basic Theory of Set Pair Analysis

The set pair analysis (SPA), proposed by Zhao in 1989, is a modified uncertainty theory considering both certainties and uncertainties as an integrated certain–uncertain system and depicting the certainty and uncertainty systematically from three aspects as identity, discrepancy, and contrary [27]. In set pair analysis, the connection degree is usually expressed as follows:

$$
\mu = \mathbf{a} + \mathbf{b}\mathbf{i} + \mathbf{c}\mathbf{j} \tag{1}
$$

where a is the identity degree, b the discrepancy degree, c the contradictory degree, a + b + c = 1 and ∀a, b, c ∈ [0, 1]; µ is the 3-element connection degree; i is the uncertainty coefficient of discrepancy, which has different values in [−1, 1]; j is the uncertainty coefficient of contradiction, which has value of −1.

#### 2.2.2. Multielement Connection Degree and Partial Connection Degree

In Equation (1), bi is the measurement between identity degree a and contradictory degree c with uncertainty. This item could be often expanded in actual applications. The expanded equation is as follows [28]:

$$\mu = \mathbf{a} + \mathbf{b}\_1 \mathbf{i}\_1 + \mathbf{b}\_2 \mathbf{i}\_2 + \dots + \mathbf{b}\_{n-2} \mathbf{i}\_{n-2} - \mathbf{c} \tag{2}$$

where µ in Equation (2) is the n-element connection degree; a + b<sup>1</sup> + b<sup>2</sup> + · · · + b<sup>n</sup> <sup>−</sup> <sup>2</sup> + c = 1; ∀a, b1, b<sup>2</sup> · · · bn−2, c ∈ [0, 1]; ∀i1, i<sup>2</sup> · · · in−<sup>2</sup> ∈ [−1, 1].

Similar to the concept of derivatives, the partial connection degree could be used to describe the development tendency of the connection degree. The first-order and second-order partial connection degree of the multielement connection degree could be described as follows [29]:

First-order partial connection degree:

$$
\partial \mathfrak{a} = \partial \mathfrak{a} + \mathbf{i}\_1 \partial \mathfrak{b}\_1 + \mathbf{i}\_2 \partial \mathfrak{b}\_2 + \dots + \mathbf{i}\_{n-2} \partial \mathfrak{b}\_{n-2} \tag{3}
$$

where ∂a = <sup>a</sup> a+b<sup>1</sup> , ∂b<sup>1</sup> = b1 b1+b<sup>2</sup> , ∂b<sup>2</sup> = b2 b2+b<sup>3</sup> , . . . , ∂bn−<sup>2</sup> = bn−<sup>2</sup> <sup>b</sup>n−2+<sup>c</sup> . This equation describes the development trend from *c* to *a*. Its essence is the n − 1-element connection degree, which could be used to describe the development trend of Equation (2).

Second-order partial connection degree:

$$
\delta^2 \mathfrak{a} = \partial(\partial \mathfrak{a}) = \partial^2 \mathfrak{a} + \mathrm{i}\_1 \partial^2 \mathfrak{b}\_1 + \mathrm{i}\_2 \partial^2 \mathfrak{b}\_2 + \dots + \mathrm{i}\_{\mathrm{n-3}} \partial^2 \mathfrak{b}\_{\mathrm{n-3}} \tag{4}
$$

Similar to Equation (3), Equation (4) is the n − 2-element connection degree which could use to describe the development trend of Equation (3), where ∂ <sup>2</sup>a = <sup>∂</sup><sup>a</sup> ∂a+∂b<sup>1</sup> , ∂ <sup>2</sup>b<sup>1</sup> = ∂b<sup>1</sup> ∂b1+∂b<sup>2</sup> , ∂ <sup>2</sup>b<sup>2</sup> = ∂b<sup>2</sup> ∂b2+∂b<sup>3</sup> , . . . , ∂ <sup>2</sup>bn−<sup>3</sup> <sup>=</sup> ∂bn−<sup>3</sup> <sup>∂</sup>bn−3+∂bn−<sup>2</sup> .

In practical applications, system risks are often divided into five levels, namely: low, relatively low, medium, relatively high, and high risks. Therefore, the five-element connection degree is often used to analyze the level and trend of risk. Its expression is:

$$\mu = \mathbf{a} + \mathbf{b}\_1 \mathbf{i}\_1 + \mathbf{b}\_2 \mathbf{i}\_2 + \mathbf{b}\_3 \mathbf{i}\_3 - \mathbf{c} \tag{5}$$

If a is taken as the reference set and define it as low risk, then b<sup>1</sup> represents relatively low risk, b<sup>2</sup> medium risk, b<sup>3</sup> relatively high risk, and c high risk.

#### 2.2.3. Set Pair Potential

When c , 0, the ratio a/c is the set pair potential [30], which is expressed as:

$$\text{Shi}\left(\mathbf{x}\right) = \mathbf{a}/\mathbf{c}, \left(\mathbf{c} \neq \mathbf{0}\right) \tag{6}$$

When a/c > 1, Shi (x) is at the same potential, which means that the risk is on the low side. When a/c = 1, Shi (x) is at the equal potential, which means the risk is at a medium size. When a/c < 1, Shi (x) is at opposite potential, which means that the risk is on the high side.

When using the five-element connection degree for risk analysis, the ratio a/c can show the situation of risk, but it fails to classify the level of risk that needs to be determined by the size of b, c, and d. According to the size of b, c, and d, the risks in the situation of the same potential and opposite potential can be divided into 65 levels (as shown in Appendix A). When Shi (x) is at the same potential, the higher the level (Level 1 is the highest level), the lower the risk; when Shi (H) is at the opposite potential, the higher the level, the higher the risk.

The concept of set pair potential can also be applied to first-order and second-order partial connection degree, which is expressed as:

$$\text{Shi}^1\left(\mathbf{x}\right) = \frac{\partial \mathbf{a}}{\partial \mathbf{b}\_{\mathbf{n}-2}}\tag{7}$$

$$\text{Shi}^2\left(\mathbf{x}\right) = \frac{\partial^2 \mathbf{a}}{\partial^2 \mathbf{b}\_{\mathbf{n}-3}}\tag{8}$$

where Shi<sup>1</sup> (x) is the set pair potential of first-order partial connection degree, and Shi<sup>2</sup> (x) is the set pair potential of second-order partial connection degree. Shi<sup>1</sup> (x) and Shi<sup>2</sup> (x) can be used to describe the trend of the risk, as shown in Appendix B.

*2.3. Risk Assessment Process of Water-Saving Contract Project Based on Five-Element Connection Degree*

(1) Calculate the index weight by entropy method

In order to eliminate the subjectivity of experts in evaluating each risk, this article uses the entropy method to calculate the weight of risk indicators. The calculation steps are as follows:

• Build the judgment matrix B, namely:

$$\mathbf{B} = \begin{bmatrix} \mathbf{x\_{11}} & \mathbf{x\_{12}} & \cdots & \mathbf{x\_{1n}} \\ & \mathbf{x\_{21}} & \mathbf{x\_{22}} & \cdots & \mathbf{x\_{2n}} \\ & \vdots & \vdots & \ddots & \vdots \\ & \mathbf{x\_{m1}} & \mathbf{x\_{m2}} & \cdots & \mathbf{x\_{mn}} \end{bmatrix} \tag{9}$$

In Equation (9), n is the number of risk assessment indicators, and m is the number of experts.

• Calculate the entropy of the indicator *j*:

$$\mathbf{H}\_{\mathbf{j}} = -\frac{1}{\ln(\mathbf{m})} \sum\_{\mathbf{i}=1}^{\mathbf{m}} \mathbf{P}\_{\mathbf{ij}} \ln(\mathbf{P}\_{\mathbf{ij}})\_{\mathbf{i}} \ (\mathbf{i} = 1, 2, \cdots, \mathbf{m}; \mathbf{j} = 1, 2, \cdots, \mathbf{n}) \tag{10}$$

$$P\_{\rm ij} = \frac{\mathbf{x\_{ij}}}{\sum\_{j=1}^{n} \mathbf{x\_{ij}}} \tag{11}$$

*Water* **2020**, *12*, 2689

• Calculate the weight of the indicator j:

$$
\omega\_{\rm{j}} = \frac{\mathbf{1} - \mathbf{H}\_{\rm{j}}}{\mathbf{n} - \sum\_{j=1}^{n} \mathbf{H}\_{\rm{j}}} \tag{12}
$$

where 0 <sup>≤</sup> <sup>ω</sup><sup>j</sup> <sup>≤</sup> 1 and <sup>P</sup><sup>n</sup> j=1 ω<sup>j</sup> = 1.

#### (2) Risk assessment based on the five-element connection degree

After determining the index weight ω, the calculation equation of the five-element connection degree can be obtained using Equation (13).

$$\begin{aligned} \boldsymbol{\mu} &= \boldsymbol{\omega} \times \mathbf{R} \times \mathbf{E}^{\mathrm{T}} &= (\boldsymbol{\omega}\_{1}, \boldsymbol{\omega}\_{2}, \cdots, \boldsymbol{\omega}\_{n}) \begin{bmatrix} \mathbf{R}\_{11} & \mathbf{R}\_{12} & \mathbf{R}\_{13} & \mathbf{R}\_{14} & \mathbf{R}\_{15} \\ \mathbf{R}\_{21} & \mathbf{R}\_{22} & \mathbf{R}\_{23} & \mathbf{R}\_{24} & \mathbf{R}\_{25} \\ \vdots & \vdots & \vdots & \vdots & \vdots \\ \mathbf{R}\_{n1} & \mathbf{R}\_{n2} & \mathbf{R}\_{n3} & \mathbf{R}\_{n4} & \mathbf{R}\_{n5} \end{bmatrix} \begin{bmatrix} 1 \\ \vdots \\ \vdots \\ \mathbf{k} \\ 1 \end{bmatrix} \\ &\tag{13} \\ &= \sum\_{\mathbf{i}=1}^{\mathbf{n}} \boldsymbol{\omega}\_{\mathbf{r}} \mathbf{R}\_{\mathbf{r} 1} + \sum\_{\mathbf{i}=1}^{\mathbf{n}} \boldsymbol{\omega}\_{\mathbf{r}} \mathbf{R}\_{\mathbf{r} 2} \mathbf{i} + \sum\_{\mathbf{i}=1}^{\mathbf{n}} \boldsymbol{\omega}\_{\mathbf{r}} \mathbf{R}\_{\mathbf{r} 3} \mathbf{j} + \sum\_{\mathbf{i}=1}^{\mathbf{n}} \boldsymbol{\omega}\_{\mathbf{r}} \mathbf{R}\_{\mathbf{r} 4} \mathbf{k} + \sum\_{\mathbf{i}=1}^{\mathbf{n}} \boldsymbol{\omega}\_{\mathbf{r}} \mathbf{R}\_{\mathbf{r} 5} \mathbf{l} \end{aligned} \tag{13}$$

where R is the occurrence probability matrix of the risk; E T is the coefficient matrix of the five-element connection degree; Rij = Nij/N (*i* = 1, 2, · · · , n; j = 1, 2, · · · , 5; j = 1 means low risk, j = 2 means relatively low risk, j = 3 means medium risk, j = 4 means relatively high risk, j = 5 means high risk). Among them, Nij is the number of experts who determine the risk indicator i as risk level j, and N is the total number of experts. Finally, we can figure out a = Pn i=1 ωrRr1, b = Pn i=1 ωrRr2, c = Pn i=1 ωrRr3,

$$\mathbf{d} = \sum\_{\mathbf{i}=1}^{\mathrm{n}} \omega\_{\mathbf{r}} \mathbf{R}\_{\mathbf{r}4} \text{ and } \mathbf{e} = \sum\_{\mathbf{i}=1}^{\mathrm{n}} \omega\_{\mathbf{r}} \mathbf{R}\_{\mathbf{r}5}.$$

In the actual analysis, it is usually difficult to encounter equal potential. Therefore, this article ranks the risk level into 5 levels according to the five-element degree of the similar potential and the inverse potential, namely: high-risk level (Levels 1–26 of inverse potential), relatively high-risk level (Levels 27–52 of inverse potential), medium-risk level (Levels 53–65 of inverse potential, Levels 53–65 of the same potential), relatively low-risk level (Levels 27–52 of the same potential), and low-risk level (Levels 1–26 of the same potential). At the same time, use the set pair potential of the first-order partial connection degree and the second-order partial connection degree to analyze the trend of risk, which can be divided into 6 types: accelerated decline, decelerated decline, decelerated ascent, accelerated ascent, uniform decline, and uniform ascent.

#### *2.4. Data Collection*

This article adopts the entropy method to determine the weight of the risk indicator. In order to obtain xij in Equation (9), the evaluation of the expert i on the risk indicator j, this article adopts the interval classification method and requires 5 experts to score it. The scoring table is shown in Table 2. Five experts are from Hebei University of Engineering, North China University of Water Resources and Electric Power, Hohai University, Beijing Guotai Water-Saving Development Co., Ltd., and the Yellow River Conservancy Commission of the Ministry of Water Resources. Among them, Hebei University of Engineering and Beijing Guotai Water-Saving Development Co., Ltd. are the participants in China's first water-saving contract project; North China University of Water Resources and Electric Power and Hohai University are specialized universities for China's water conservancy and hydropower research; the Yellow River Conservancy Commission of the Ministry of Water Resources belongs to MWRC.


**Table 2.** The interval classification of the impact degree of the risk indicator.

In order to obtain the risk occurrence probability of Rij in Equation (13), this article uses the Likert five-point scale method to design the questionnaire. In the Likert five-point scale, 1 means low risk, 2 means relatively low risk, 3 means medium risk, 4 means relatively high risk, 5 means high risk. In this article, 300 questionnaires were distributed to staff members of water-saving service operators (including the technical manager, financial manager, procurement manager, and project manager) from 13 water-saving contract pilot projects, and 276 valid questionnaires were finally obtained. The descriptive statistics results are shown in Table 3

**Table 3.** Descriptive statistical results of the questionnaire.


Note: Source: calculated by SPSS 19.

In this paper, Cronbach's Alpha is used for the reliability test. The results show that Cronbach's Alpha of the questionnaire is 0.912 (>0.9), indicating that the questionnaire has good reliability. The results are shown in Table 4


**Table 4.** Reliability Analysis of questionnaire on the risk level.

Note: Source: calculated by SPSS 19.

#### **3. Results**

The weight of each risk and their five- element connection degree are shown in Table 5.


**Table 5.** Calculation table of the five-element connection degree.

Note: "+" means same potential, "−" means opposite potential.

As shown in Table 5, uCCS = 0.1686 + 0.2042i + 0.2250j + 0.2306k + 0.1703l, shi(CCS) = 0.9900 < 1. It is at Level 47 of the opposite potential, which means the risk in the contract signing stage is relatively high. In this stage, the audit risk is at Level 9 of the opposite potential, which is at a high-risk level; technical risk is at Level 25 of the same potential, which is at a low-risk level; the market competition risk is at Level 61 of the opposite potential, which is at a medium-risk level.

uIS = 0.2393 + 0.2164i + 0.1812j + 0.1661k + 0.1957l, shi(INS) = 1.2227 > 1. It is at Level 3 of the same potential, which means the risk of this stage is low. In this stage, financing risk is at Level 25 of the opposite potential, which is at a high-risk level; technical risk is at Level 25 of the same potential, which is at a low-risk level.

uOS = 0.2907 + 0.1824i + 0.1902j + 0.1917k + 0.1449l, shi(IMS) = 2.0062 > 1. It is at Level 25 of the same potential, which means the risk of this stage is low. In this stage, construction risk, policy risk, force majeure risk, and inflation risk are at Levels 7, 21, 19, and 25 of the same potential, respectively, which means they are all at low-risk levels.

uBSS = 0.2343 + 0.1695i + 0.2015j + 0.1736k + 0.2210l, shi(BSS) = 1.0601 > 1. It is at Level 21 of the same potential, which means the risk of this stage is low. In this stage, facility depreciation risk and water price change risk are at Levels 21 and 19 of the same potential, respectively, which means they are at a low-risk level; payment risk is at Level 9 of the opposite potential, which is at a high-risk level.

uWSC = 0.2265 + 0.1982i + 0.2027j + 0.1931k + 0.1839l, shi(WSC) = 1.2316 > 1. It is at Level 19 of the same potential, which means the overall risk of the WSC project is low. Therefore, WSC projects are suitable for development in China.

In each stage of the WSC project, the contract signing stage is at a relatively high level of risk; the investment stage, implementation stage, and benefit-sharing stage are at a low-risk level. Therefore, the contract signing stage is the focus of risk control in the WSC project. From the perspective of each risk, audit risk, financing risk, and payment risk are at a high-risk level, they are the primary concern in risk control; market competition risk is at a medium-risk level, it is the secondary concern in risk control; the remaining risks are at a low-risk level, but it does not mean that these risks can be ignored, because the risk level may be easily affected by factors such as national development strategy, overall economic level, management technology, and market resource allocation as the WSC project is still in its infancy in China. Therefore, it is necessary to analyze Shi<sup>1</sup> (x) and Shi<sup>2</sup> (x) of each risk, as shown in Table 6.


**Table 6.** Calculation of partial connection degree and trend analysis.

Note: "+" means same potential, "−" means opposite potential.

As shown in Table 6, Shi<sup>1</sup> (CSS) is at the opposite potential while Shi<sup>2</sup> (CSS) is at the same potential, which means the risk of contract signing stage has a trend of decelerated ascent according to Appendix B. In this stage, audit risk and market competition risk have a trend of accelerated ascent, and technical risk has a trend of accelerated decline.

Shi<sup>1</sup> (INS) is at the same potential while Shi<sup>2</sup> (INS) is at the opposite potential, which means the risk of investment stage has a trend of decelerated decline. In this stage, financing risk has a trend of accelerated ascent, and interest rate risk has a trend of accelerated decline.

Shi<sup>1</sup> (IMS) and Shi<sup>2</sup> (IMS) are both at the same potential, which means the risk of implementation stage has a trend of accelerated decline. In this stage, construction risk, policy risk, and force majeure risk have a trend of accelerated decline, and inflation risk has a trend of decelerated ascent.

Shi<sup>1</sup> (BSS) and Shi<sup>2</sup> (BSS) are both at the same potential, which means the risk of the benefit-sharing stage has a trend of accelerated decline. In this stage, facility depreciation risk and water price change risk have a trend of accelerated decline, and payment risk has a trend of decelerated decline.

Shi<sup>1</sup> (WSC) is at the opposite potential while Shi<sup>2</sup> (WSC) is at the same potential, which means the overall risk of the WSC project has a trend of decelerated ascent. Although the overall risk of the WSC project is low as we concluded before, it shows a trend of decelerated ascent. This indicates that there are some potential high risks in WSC projects. From the perspective of each risk, audit risk and financing risk are not only at a high-risk level but also show a trend of accelerated ascent, so they are at the highest risk. Although the market competition risk is at a medium-risk level, it shows a trend of accelerated ascent, so it should also be considered as high risk. Payment risk is at a high-risk level, it shows a trend of decelerated decline; while inflation risk is at a low-risk level, it shows a trend of decelerated ascent. For these two risks, based on conservative principles, the former should still be treated as high risk, while the latter should be treated as medium risk. The remaining risks are all at a low-risk level and show a trend of accelerated decline, so their impact can be ignored under normal circumstances.

#### **4. Discussion**

In this section, we focus on discussing the risk level and risk trend of the WSC projects, and then put forward some policy recommendations for the high risks.

#### *4.1. Audit Risk*

Audit risk is not only at a high-risk level but also shows a trend of accelerated ascent. After interviewing the managers of water-saving service operators, it is concluded that the audit risk is at a high-risk level for the following two reasons: (1) For urban permanent residents, water-saving transformation can save costs in the long run. Therefore, these water users will deliberately over-report their water consumption in order to allow water-saving service operators to transform their water supply facilities, which will result in a longer payback period for water-saving service operators; (2) For rural water users who need agricultural irrigation, it is difficult to accurately assess their water consumption due to the influence of climate, environment and market demand, which leads to the uncertainty of investment payback period of water-saving service operators. Obviously, the payback period is the main factor that affects audit risk. The long payback period will increase the debt burden of the water-saving service operators [18]. Moreover, it will also increase the probability of other risks, resulting in some secondary risks. Therefore, audit risk shows a trend of accelerated ascent.

#### *4.2. Financing Risk*

As same as audit risk, financing risk is also at a high-risk level with a trend of accelerated ascent. In terms of financing risks, credit, mortgage, and loan mechanisms have not been established for WSC projects in the bank's financial system, which is the main reason for the high financing risk [31]. First of all, WSC projects are still in their infancy in China, and water-saving service operators have not yet obtained good credit ratings from the credit evaluation departments of financial institutions. Second, assets formed by WSC projects, such as equipment and contract receivables, can only be evaluated at the implementation stage and benefit-sharing stage. Banks and other financial institutions often do not recognize such assets or accept them as credit collateral during the investment stage. Thirdly, the technique and risk of the WSC projects are not well known by commercial banks, which greatly increases the cost of loan examination. According to the above reasons, the loan review of commercial banks will inevitably be stricter, and the requirements for loan guarantees will inevitably increase, making it more difficult for water-saving service operators to obtain financing. Li et al. pointed out that the water-saving service operators in China are generally small- and medium-sized enterprises. Unlike state-owned enterprises, these enterprises have difficulty obtaining financing as the market economy is not developed at a high level and the preferential policies are not strong enough [20]. In addition, commercial banks tend to lend to projects with short cycles and high returns [32]. Therefore, the majority of water-saving service operators will fall into a vicious circle of difficulty in obtaining loans, which leads financing risk to show a trend of accelerated ascent.

#### *4.3. Market Competition Risk*

Market competition risk is at a medium-risk level, but it has a trend of accelerated ascent, so it should also be regarded as a high-risk level. The high market competition risk of China's WSC projects is mainly due to the monopoly of local water-saving service operators and state-owned enterprises. First of all, no clear industry access standard has been established for WSC projects in China, and there is a lack of authoritative evaluation standard for water-saving efficiency and service level, which has led to irregular operation and market monopoly by some local water-saving service operators [33]. Second, state-owned enterprises have monopolized almost all large-scale WSC projects with their unique resource endowments (https://wsmc-china.com/home/main.html), these resource endowments include advanced technology, standardized management, high reliability, and state subsidies [34]. Local monopolies can be eliminated by regulating the market, but the monopoly of state-owned enterprises is difficult to change in China. As the current trend of "guojingmingtui" (the retreat of the private sector and advancement of state-owned enterprises) in China becomes more and more intense [35], the market competition risk caused by the monopoly of state-owned enterprises has a trend of accelerated ascent.

#### *4.4. Payment Risk*

Payment risk is at a high-risk level, it shows a trend of decelerated decline. Based on conservative principles, it should be treated as high risk. After interviewing the managers of water-saving service operators, it is concluded that the reasons for the high payment risk are customer default and business problems. There are two main cases of customer default: (1) Water users do not pay the water-saving benefits belonging to water-saving service operators; (2) As other water-saving operators gave more favorable terms, the water user breached the contract and resigned the contract with other water-saving service operators. The business problem refers to the customer's inability to reach the expected water consumption due to economic and demand pressures. In the final analysis, customer defaults are caused by customers' low moral standards and creditworthiness, and customer's business problems are caused by inaccurate assessments of customers' economic level and water consumption. Wang et al. believe that the application of blockchain and big data can help companies obtain more customer information, so that they can conduct a comprehensive evaluation of customers [36]. At present, blockchain and big data have begun to be applied in China in terms of information sharing, so the payment risk will be reduced in the long run [37].

#### *4.5. Inflation Risk*

Inflation risk is at a low-risk level, but it shows a trend of decelerated ascent, so it should be regarded as a medium level risk. Yadav et al. compared the inflation rates of China, the United States, and India. Their research shows that the inflation rates of China and the United States are basically the same and are at a relatively low level, while the inflation rate of India is much higher than that of China and the United States [38]. However, although the risk of inflation is very low in China, it has a steady upward trend, which is related to China's currency oversupply in recent years [39], and China's inflation rate in recent years has also proved this trend (as shown in Figure 2).

**Figure 2.** China's inflation rate (2014–2019).

#### *4.6. Other Risks*

The remaining risks are all at a low-risk level and show a trend of accelerated decline. Among these risks, technical risks, construction risks, and facility depreciation risks are technical and management risks. Such risks can be reduced significantly by the accumulation of project experience and the rapid development of technologies, so the risks show an accelerated decline.

Interest rate risk, policy risk, force majeure risk, and water price change risk are risks outside of technology and management, which are greatly affected by the fluctuations of the external environment.

In recent years, in order to promote economic development, China has continuously cut interest rates. The benchmark lending rate of China's central bank showed a continuous and large-scale decline (by 1.25%) from 22 November 2014 to 24 October 2015, and remained stable after 24 October 2015 [40], as shown in Figure 3. Therefore, the interest rate risk has a trend of accelerated decline.

**Figure 3.** The benchmark lending rate of the People's Bank of China (2014–2020).

For the policy risk, although the preferential policies for water-saving contract projects are not enough at present [20], related policies are still being introduced [41]. Therefore, policy risk has a trend of accelerated decline.

We have counted the force majeure events encountered in 13 water-saving contract pilots, and the results showed that none of the projects were affected by force majeure. Therefore, it can be seen that the force majeure risk is low. In water-saving contract projects, except for the core technologies of water-saving, infrastructure construction takes up the majority of the projects. With the continuous accumulation of experience in China's infrastructure construction, the impact of force majeure on it has become smaller and smaller [42]. Therefore, Force majeure risk has a trend of accelerated decline.

China's water price has been at a low level due to the government's macrocontrol [43], and the overall water price in China has shown a downward trend since 2014 (http://www.h2o-china.com/price/). Therefore, the risk of water price changes has a trend of accelerated decline.

#### **5. Conclusions**

This paper uses the literature analysis method to determine the risks in the life cycle of the WSC project, and uses the multielement connection degree set pair analysis to evaluate the level and trend of the risks. The results show:


In summary, audit risk, financing risk, market competition risk, payment risk, and inflation risk are the risks that should be focused on in water-saving contract projects. Since inflation risk can only be avoided through financial analysis, and is not controllable, this article proposes the following recommendations for audit risk, financing risk, market competition risk, and payment risk:

(1) Audit risk and payment risk

Audit risk and payment risk can be reduced through effective third-party management mechanisms. Therefore, it is necessary to cultivate a group of qualified third-party management institutions for

WSC projects. These institutions not only supervise the performance of contracts by water users and water-saving service operators, but also coordinate and arbitrate contradictions between both parties.

#### (2) Market competition risk

China has not yet formed a standard market competition environment. In today's economic globalization, China should increase international cooperation, learn from the mature experience and models accumulated by other countries in EPC projects, and actively create a good market-oriented competition environment, so as to ensure the high quality and sustainable development of the WSC projects.

#### (3) Financing risk

The financing difficulty of small- and medium-sized enterprises is a universal problem, which does not only exist in WSC projects. In order to help water-saving service operators obtain financing, special funds for WSC projects in the banking system could be considered; Secondly, in order to solve the loan guarantee problem, water-saving service operators should be allowed to use the improved technology as collateral. Finally, insurance mechanisms can be introduced into ESC projects, that is, taking insurance premiums as a means of financing.

**Author Contributions:** Conceptualization, Q.L. and M.Y.W.; data curation, Z.S., D.Y., and R.Z.; investigation, Z.S. and R.Z.; methodology, Q.L. and Z.S.; software, Z.S.; supervision, M.Y.W., D.Y., and C.W.; writing—original draft preparation, Q.L. and Z.S.; writing—review and editing, M.Y.W. and C.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** Australian Research Council: DP170104138; National Social Science Fund: 18BGL006; the Major Projects of the Key Research Base of Humanities and Social Sciences of the Ministry of Education: 18JJD790018.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **Appendix A**


**Table A1.** The rank of five-element connection degree of similar potential.


**Table A2.** The rank of five-element connection degree of inverse potential.

#### **Appendix B**

**Table A3.** Trend curve of the risk.


–

–

–

–

–

–

–

–

–

–

–

–

China's

China's

China's

China's

China's

China's

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

China's —

China's —

China's —

China's —

China's —

China's —

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
