A Gateway to Successful River Restorations: A Pre-Assessment Framework on the River Ecosystem in Northeast China

Abstract

Natural rivers have been disturbed for hundreds of years by human activities. Previous water conservancy projects in the form of dams, reservoirs, dykes, and irrigation infrastructure focused on the social and economic benefits and disregarded the adverse effects on the physical, chemical, and biological characteristics of the affected rivers. Since the 2000s, the comprehension of river remolding has transformed so decisions are more socially and ecologically beneficial. However, restoration actions are often implemented aimlessly, without a detailed plan or sufficient communication, leading to the failure of accomplishing objectives for a variety of ecologic, financial, and social reasons. Thus, a pre-assessment framework is proposed in this paper, to determine river restoration priorities, emphasizing both social and ecological aspects. The vague notion of river health is evaluated using the Variable Fuzzy Assessment Model (VFAM) and expressed by modified Nightingale Rose Diagrams (NRDs). The river social ecosystem was subsequently analysed using this framework in the Ashihe River near Harbin City, Northeast China. The application of VFAM demonstrated that the health status of the upper, middle, and lower sections of the river could be classified as sub-healthy, degraded, or sick in terms of ecosystem structures, and sub-healthy, degraded, or degraded in terms of social functions, respectively. The health status of the lower section was the poorest and should be restored first. Using NRDs, we found that water quality deterioration and irrigation works are the two key factors in river degradation, which must be improved throughout the entire watershed. Aesthetics and recreation should also be given priority to restore the lower section due to the demands of nearby residents. Several measures are also suggested for decision makers who need a more detailed design to implement. This framework potentially assists with communicating with stakeholders, avoids aimless restoration actions, and contributes to comparing with the measuring after restorations.

1. Introduction

Humans have settled along rivers for daily life, irrigation, and industry for thousands of years. Due to the purposeful reformation of anthropogenic activities, approximately 60% of natural rivers in residential areas have been dammed to irrigate, supply water, and generate power, and embanked to control flooding [1,2]. Various water conservancy projects have resulted in many social and economic benefits. However, little attention was paid to the subsequent damage to the river habitats of organisms, from bacteria to fish. Natural rivers were threatened by both exploitation and various sources of pollution. Many problems, such as habitat shrinkage, water quality deterioration, discontinuous flow, and other degradation of the river ecosystem, are becoming increasingly severe, arousing more concern from researchers and the public alike.

River restoration is intricate work involving ecological, technical, and socio-economic factors [3], thus recognizing rivers as social-ecological systems is crucial [4]. Although the practice of river restoration has grown exponentially since the 1990s [5], river health and habitat condition have been estimated with various methods since the 2000s [6]. Most of the previously-completed restoration, however, was often pointless due to the lack of theoretical direction [7]. For example, some river restoration emphasized that working to achieve a stable channel may in some cases help protect infrastructure through hard engineering such as rock walls and rip-rap. Similarly, some riparian enhancement projects planted nonnative vegetation instead of native bosk and tussock grass in urban rivers for aesthetic appearance. Such activities were often costly and less effective ecologically [8,9]. The efficient restoration of river ecosystems is still a significant and intricate issue for river ecologists and decision makers. Furthermore, balancing ecological protection and the development of rivers is also a pressing issue, especially in developing countries, for the growing economy, increasing environmental awareness, and improving the standard of living. The deficiencies and priority analysis of river ecosystems could help determine the most efficient river restoration measures [3]. Therefore, pre-evaluating the ecological status and the exploitation conditions prior to implementing river rehabilitation actions are necessary.

Given the diversity and complexity of the understanding and various types of rivers in different regions, similar indicator systems have been proposed to assess the four components of river health: hydrology, geomorphology, water quality, and aquatic organisms [6,10,11]. Two major viewpoints about river health exist. One is that original rivers are healthy rivers. Several indicator systems were developed at the end of 20th century, such as the Index of Stream Condition (ISC) in Australia [12], Rapid Bioassessment Protocols (RBPs) in the U.S. [13], and the River Habitat Survey (RHS) in the U.K. [14]. Although the scoring differs among these methods, almost all of these indicator systems emphasize the status of aquatic organism habitats. The other viewpoint is that rivers should also meet the functional requirement of humans. The issue of structure vs. function in assessing river systems has been discussed at length, and the importance of assessing both has been discussed previously [8,15]. The Water Framework Directive (WFD) highlighted the sustainable development of river basins in the E.U. [16]. An indicator system for river health assessment, including hydrology, water quality, river geomorphology, biology characteristics, and socioeconomic status, was established in China [6]. Dialectically, the system is one-sided, either to immoderately exploit rivers without concerning ecological continuity in conventional ways, or to cancel or avoid most of the artificial facilities with little consideration of the river ecosystem functions, yielding to the pressure of environmental over-protection in residential areas [17]. Realizing that rivers will never return to natural status, even if resources are devoted to an urban stream, is essential. It is inconceivable that we could introduce rare species, a complex food web, or any other ecological benefit into a city [18]. In our opinion, a healthy river is not only healthy in the ecosystem structures but must also have sustainable social functions. Hence, establishing a pre-restoration assessment indicator system with comprehensive consideration of both the ecosystem integrity and ecosystem services of rivers is necessary.

The health of the river ecosystem structure and the sustainability of river social functions are both vague terms because they depend on human demands and self-identities [19,20,21]. Moreover, indicators for river health assessment are often uncertain and nonlinear in river social-ecological systems [22]. Therefore, scientists have attempted to predict and quantify the uncertainty of river systems with different kinds of models, such as conventional scoring systems, grey systematic analyses, set pair analyses, rough set analyses, and fuzzy comprehensive assessment models [22,23,24,25]. In this study, the river health indicator system was holistically evaluated based on both social and ecological aspects using a modified fuzzy comprehensive assessment model called the Variable Fuzzy Assessment Model (VFAM).

This study aimed to establish a multiple target-oriented framework to determine the most efficient decision for river restoration. Priority determination is an important factor in making the most effective decision for multi-objective restoration actions [26], such as water quality enhancement, habitat improvement, dam removal, etc. It seems that the weakest link should be the priority, however, river restoration prioritization often depends on legal mandates, total cost, technical feasibility, stakeholders’ preferences, and other factors [27]. According to data from the National River Restoration Science Synthesis (NRRSS), most small-scale accomplished and ongoing river restoration enhance water quality and manage riparian zones with low costs and easy implementation [28]. Despite these efforts, previous projects were only implemented depending on what was possible rather than what was the proper course of action. To determine the indicators of the most impaired habitats, we used the Nightingale Rose Diagram (NRD) to visually display the VFAM results for stakeholders. The analysis of riverine deficiency could indicate the cause of river ecosystem degradation, what should be restored, and what should be priority in terms of restoration [29]. Hence, the primary criterion for objective priority determination is to preferentially improve the most impaired areas depending on ecologic, financial, and social conditions.

In addition, the VFAM framework was applied to determine the most degraded area and restoration priority, and to propose an effective restoration strategy for the Ashihe River near Harbin City, Northeast China, which has been severely disturbed by human activities in a rural and urban catchment. According to the results of this pre-assessment framework, several preliminary actions are recommended to restore the impaired aspects of the Ashihe River. This framework shows potential for communication with stakeholders, avoids aimless restoration actions, and allows the comparison of the measurements after restoration actions have been completed.

2. Materials and Methods

2.1. Study Area

The Ashihe River is one of the main branches of the Songhua River in Harbin City, Heilongjiang Province, Northeast China. The river is about 213 km long, with a catchment area of 3581 km², and is frozen for 130–140 days every winter. The river originates from Mountain Zhangguangcailing, meanders across Acheng City, and flows into the Songhua River in urban Harbin City. A vast reservoir is located upstream of the Ashihe River, with a storage capacity of about 480 million cubic meters. The total population in the Ashihe River basin is about 700,000, and the urban population accounts for 33.5%. A dam and six irrigation systems exist along the main river (Figure 1), which leads to an inconsistent flow in the dry season.

The Ashihe River is divided into three sections based on land use and water use types. The upper section is protected as drinking water sources from the origin to Xiquanyan Dam (Figure 1, Point 1) with an average gradient of 0.34%. The middle section is mainly diverted for irrigation from Xiquanyan Dam to Acheng hydrological station (Figure 1, Point 2) with an average gradient of 0.09%. The lower section accepts domestic sewage and industrial effluent from Harbin City and Acheng City, from the Acheng hydrological station to the estuary, with an average gradient of 0.06%. The water is very limpid from the origin to Xiquanyan Reservoir until the river flows through urban Acheng City. Due to the emission of domestic sewage and industrial effluent from the urban area of Acheng City and Harbin City into the river, parts of the tributaries of the Ashihe River are so severely polluted that they turn black and odorous every late spring and early summer.

2.2. Pre-Assessment Framework

Scientists proposed five success criteria for previous river restoration projects, emphasizing the significance of both the pre- and post-project restoration appraisals [9]. New projects should be forecasted with experience drawn from past efforts [30], so more efficient restoration projects can be implemented with fewer errors and readjustments. The first part of the Proceeding Chain of Restoration [3], a useful approach to improve river restoration, is the pre-restoration process used to determine the targets and scope of the research. More details about the pre-restoration process (Figure 2) can be compatibly applied to small- and medium-sized rivers. The seven steps of our framework include: (1) field survey and historical data collection; (2) integrated metrics establishment in consideration of ecosystem structures and social functions; (3) river health assessment using VFAM; (4) potential problem analysis using NRD; (5) determination of objective priority; (6) preliminary restoration recommendation; and (7) feasibility study.

All assessment steps should be communicated to stakeholders, restoration practitioners, scientists, and decision makers due to the complexity of river restoration [31,32,33]. Each group of stakeholders plays a role in the different steps given contrasting interests, the limitation of understanding, and sense of responsibility [34]. The participation of stakeholders could enhance the acceptance of river restoration projects [35]. Scientists involved in hydrology, geomorphology, ecology, and management quantitatively engage in evaluating the river ecosystem structures and social functions to assist stakeholders in predicting the response to specifically designed actions. Restoration practitioners have more experience in judging the technical feasibility, ecological applicability, and social-economic availability of preliminary measures [36]. The river restoration projects are often small-scale given relatively limited budget from the local government, especially in developing regions [28,37,38]. Thus, preliminary recommendations and their feasibility should be discussed by each group to avoid inefficient actions.

2.3. Methodology

2.3.1. Variable Fuzzy Sets and Variable Fuzzy Assessment Model

Variable Fuzzy Sets (VFS) is an improved fuzzy set theory for resolving the dynamic property of fuzzy concepts and the inapplicability of the maximum membership degree principle. VFS can also normalize both qualitative and quantitative indicators into the same assessment criteria. The basic concept of VFS is to derive a measure, and more details can be found in previous reports [39,40].

Definition 1.

$\mathit{u}\in \mathit{U}$, where U is a domain of a pair of opposite fuzzy concept, and $\underset{˜}{\mathit{A}}$ and ${\underset{˜}{\mathit{A}}}^{\mathit{c}}$, and ${\mu }_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)$ and ${\mu }_{{\underset{˜}{\mathit{A}}}^{\mathit{c}}}\left(\mathit{u}\right)$ are their relative membership degree, respectively.

Definition 2.

${\mathit{D}}_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)$ is the relative difference degree of $\mathit{u}$ to $\underset{˜}{\mathit{A}}$.

$${\mathit{D}}_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)={\mu }_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)-{\mu }_{{\underset{˜}{\mathit{A}}}^{\mathit{c}}}\left(\mathit{u}\right)$$

(1)

$${\mu }_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)+{\mu }_{{\underset{˜}{\mathit{A}}}^{\mathit{c}}}\left(\mathit{u}\right)=1$$

(2)

Definition 3.

Mapping.

$$\begin{array}{l}{\mathit{D}}_{\underset{˜}{\mathit{A}}}:\mathit{U}\to \left[-1,1\right]\hfill \\ \mathit{u}|\to {\mathit{D}}_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)\in \left[-1,1\right]\hfill \end{array}$$

(3)

where Equation (3) is the relative difference function of $\mathit{u}$ to $\underset{˜}{\mathit{A}}$.

Definition 4.

$\underset{˜}{\mathit{V}}\subseteq \mathit{U}$ is a variable fuzzy set.

$$\underset{˜}{\mathit{V}}=\left\{\left(\mathit{u},\mathit{D}\right)|\mathit{u}\in \mathit{U},{\mathit{D}}_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)={\mu }_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)-{\mu }_{{\underset{˜}{\mathit{A}}}^{\mathit{c}}}\left(\mathit{u}\right),\mathit{D}\in \left[-1,1\right]\right\}$$

(4)

$${\mathit{A}}_{+}=\left\{\mathit{u}|\mathit{u}\in \mathit{U},{\mu }_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)>{\mu }_{{\underset{˜}{\mathit{A}}}^{\mathit{c}}}\left(\mathit{u}\right)\right\}$$

(5)

$${\mathit{A}}_{-}=\left\{\mathit{u}|\mathit{u}\in \mathit{U},{\mu }_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)<{\mu }_{{\underset{˜}{\mathit{A}}}^{\mathit{c}}}\left(\mathit{u}\right)\right\}$$

(6)

$${\mathit{A}}_0=\left\{\mathit{u}|\mathit{u}\in \mathit{U},{\mu }_{\underset{˜}{\mathit{A}}}\left(\mathit{u}\right)={\mu }_{{\underset{˜}{\mathit{A}}}^{\mathit{c}}}\left(\mathit{u}\right)\right\}$$

(7)

where ${\mathit{A}}_{+}$, ${\mathit{A}}_{-}$, and ${\mathit{A}}_0$ are the attraction domain, repulsion domain, and balance boundary, respectively.

Suppose ${\mathit{Y}}_{\mathit{h}}=\left[{\mathit{y}}_{\mathit{h}},{\mathit{y}}_{\mathit{h}+1}\right]$ is the attraction domain of ${\underset{˜}{\mathit{V}}}_{\mathit{h}}·\mathit{Y}=\left[{\mathit{y}}_{\mathit{h}-1},{\mathit{y}}_{\mathit{h}+2}\right]$ is an interval containing ${\mathit{Y}}_{\mathit{h}}({\mathit{Y}}_{\mathit{h}}\subset \mathit{Y})$. Both intervals $\left[\mathit{c},\mathit{a}\right]$ and $\left[\mathit{b},\mathit{d}\right]$ are the repulsion domain of $\underset{˜}{\mathit{V}}$. Suppose $\mathit{M}\in \left[\mathit{a},\mathit{b}\right]$, then ${\mathit{D}}_{\mathit{h}}\left({\mathit{M}}_{\mathit{h}}\right)=1$.

Definition 5.

$\forall \mathit{x}\in \mathit{Y}$, if $\mathit{x}\in \left[{\mathit{y}}_{\mathit{h}-1},{\mathit{M}}_{\mathit{h}}\right]$, the relative difference function is:

$$\{\begin{array}{c}{\mathit{D}}_{\mathit{h}}\left(\mathit{x}\right)=\left(\mathit{x}-{\mathit{y}}_{\mathit{h}}\right)/\left({\mathit{M}}_{\mathit{h}}-\mathit{y}\right),\mathit{x}\in \left[{\mathit{y}}_{\mathit{h}},{\mathit{M}}_{\mathit{h}}\right]\\ {\mathit{D}}_{\mathit{h}}\left(\mathit{x}\right)=-\left(\mathit{x}-{\mathit{y}}_{\mathit{h}}\right)/\left({\mathit{y}}_{\mathit{h}-1}-{\mathit{y}}_{\mathit{h}}\right),\mathit{x}\in \left[{\mathit{y}}_{\mathit{h}-1},{\mathit{y}}_{\mathit{h}}\right]\end{array}$$

(8)

If $\mathit{x}\in \left[{\mathit{M}}_{\mathit{h}},{\mathit{y}}_{\mathit{h}+2}\right]$, the relative difference function is:

$$\{\begin{array}{c}{\mathit{D}}_{\mathit{h}}\left(\mathit{x}\right)=\left(\left(\mathit{x}-{\mathit{y}}_{\mathit{h}+1}\right)/\left({\mathit{M}}_{\mathit{h}}-{\mathit{y}}_{\mathit{h}+1}\right)\right),\mathit{x}\in \left[{\mathit{M}}_{\mathit{h}},{\mathit{y}}_{\mathit{h}+1}\right]\\ {\mathit{D}}_{\mathit{h}}\left(\mathit{x}\right)=-\left(\left(\mathit{x}-{\mathit{y}}_{\mathit{h}+1}\right)/\left({\mathit{y}}_{\mathit{h}+2}-{\mathit{y}}_{\mathit{h}+1}\right)\right),\mathit{x}\in \left[{\mathit{y}}_{\mathit{h}+1},{\mathit{y}}_{\mathit{h}+2}\right]\end{array}$$

(9)

$\mathit{M}$ could be set to a specific value or as the mid-value of the interval $\left[\mathit{a},\mathit{b}\right]$, as shown in Figure 3.

Thus, ${\mu }_{\underset{˜}{\mathit{A}}}\left({\mathit{x}}_{\mathit{i}\mathit{h}}\right)$ could be derived from Equations (1), (2), (8), and (9).

$${\mu }_{\underset{˜}{\mathit{A}}}\left({\mathit{x}}_{\mathit{i}\mathit{h}}\right)=\frac{1}{2}({\mathit{D}}_{\mathit{h}}({\mathit{x}}_{\mathit{i}})-1)$$

(10)

Definition 6.

Suppose there are $\mathit{m}$ indicators, and each indicator is graded into $\mathit{h}$ ( $\mathit{h}=\begin{array}{ccccc}1& 2& 3& \cdots & \mathit{c}\end{array}$), the relative membership degree of the grade $\mathit{h}$ is derived as follows:

$${\mathit{u}}_{\mathit{i}\mathit{h}}^{\prime }={\left\{1+{\left[\frac{{{\displaystyle \sum _{\mathit{i}=1}^{\mathit{m}}\left[{\mathit{w}}_{\mathit{i}}(1-{\mu }_{\underset{˜}{\mathit{A}}}({\mathit{x}}_{\mathit{i}\mathit{h}})\right]}}^{\mathit{p}}}{{{\displaystyle \sum _{\mathit{i}=1}^{\mathit{m}}\left[{\mathit{w}}_{\mathit{i}}{\mu }_{\underset{˜}{\mathit{A}}}({\mathit{x}}_{\mathit{i}\mathit{h}})\right]}}^{\mathit{p}}}\right]}^{\raisebox{1ex}{$\alpha $}\!\left/ \!\raisebox{-1ex}{$\mathit{p}$}\right.}\right\}}^{-1}$$

(11)

where ${\mathit{u}}_{\mathit{i}\mathit{h}}^{\prime }$ is the non-normalized relative membership degree, ${\mathit{w}}_{\mathit{i}}$ is the weight of the indicator, ${\mathit{w}}_{\mathit{i}}$ is a variable parameter determined by actual conditions in VFS, α is the parameter of the optimized criterion, α = 1 and α = 2 represent the least absolute method and the least-squares method, respectively, p is the distance parameter, and p = 1 and p = 2 represent the Hamming and Euclidean distance, respectively.

The normalized relative membership degree is defined as follows:

$${\mathit{u}}_{\mathit{i}\mathit{h}}={\mathit{u}}_{\mathit{i}\mathit{h}}^{\prime }/{\displaystyle \sum _{\mathit{h}=1}^{\mathit{c}}{\mathit{u}}_{\mathit{i}\mathit{h}}^{\prime }}$$

(12)

The eigenvalue of the grade variable could be derived as:

$$\mathit{H}={\displaystyle \sum _{\mathit{h}=1}^{\mathit{c}}{\mathit{u}}_{\mathit{i}\mathit{h}}}\mathit{h},\mathit{H}\in \left[1,\mathit{c}\right]$$

(13)

The VFAM was established with the four combinations of α and p in Equation (11). The VFAM comprehensive assessment criterion is as follows:

$$\{\begin{array}{l}\mathit{h}=1\mathit{H}\in \left[1,1.5\right]\hfill \\ \mathit{h}=\mathit{h}\mathit{H}\in \left[\mathit{h}-0.5,\mathit{h}+0.5\right]\begin{array}{c}\left(\mathit{h}=23\dots \mathit{c}-1\right)\end{array}\hfill \\ \mathit{h}=\mathit{c}\mathit{H}\in \left[\mathit{c}-0.5,\mathit{c}\right]\hfill \end{array}$$

(14)

2.3.2. Modified Nightingale Rose Diagram

NRD, a form of pie chart, was developed by Nightingale to illustrate seasonal sources of patient mortality in a military field hospital [41]. The comparison of different indicators could be visually shown in a modified NRD regardless of audience education level.

The modified NRD was drawn as shown in Figure 4. Indicators are marked with colored sectors in a polar coordinate. Lengths $\mathit{l}$ and degrees θ were determined as follows:

Suppose that there are $\mathit{m}$ indicators in the system, $\forall {\mathit{X}}_{\mathit{i}}(\mathit{i}=\begin{array}{cccc}1,& 2,& \dots ,& \mathit{m})\end{array}$,

$${\mathit{l}}_{\mathit{i}}={\displaystyle \sum _{\mathit{h}=1}^{\mathit{c}}{\mu }_{\underset{˜}{\mathit{A}}}({\mathit{x}}_{\mathit{i}\mathit{h}})}\mathit{h},(\mathit{i}=\begin{array}{cccc}1,& 2,& \dots ,& \mathit{m})\end{array}$$

(15)

where ${\mu }_{\underset{˜}{\mathit{A}}}({\mathit{x}}_{\mathit{i}\mathit{h}})$ is the normalized relative membership degree. The weight ${\mathit{w}}_{\mathit{i}}$ is transformed into an angle ${\theta }_{\mathit{i}}$ by the following equation:

$${\theta }_{\mathit{i}}=360{\mathit{w}}_{\mathit{i}}$$

(16)

More information can be shown in the modified NRD. The summed area of the colored sectors demonstrates the health of the riverine social ecosystem. The larger uncolored area of either the red or yellow sector shows the weaker aspect of the river ecosystem structure or social function. For example, as shown in Figure 4, X₈ is the weakest aspect and should be highest priority restoration objective.

2.4. Riverine Social Ecosystem Indicators

A healthy river is not only healthy in terms of the ecosystem structures but is also sustainable in its social functions. Field survey and historical data collection are essential for obtaining first-hand information of a study area, including hydrology, climate, precipitation, vegetation, hydraulic structure, water quality, population, economic level, and aquatic life data. All these data were categorized to establish a comprehensive indicator system. Indicators were divided into structural indicators (S) and functional indicators (F), as shown in

Table 1. Summary of the indicators for river social ecosystem assessment in the Ashihe River.

Category	Indicator, Unit (Xi), and Description	Note
Ecosystem Structures
Watershed	Mean runoff depth (mm; X1): the average runoff of a square kilometre catchment indicates the abundance of water resources.	Y
	Exploitation rate of groundwater (%; X2): the percentage of exploited water to available groundwater.	Y
	Fraction of vegetation cover (%; X3): the fraction of ground covered by vegetation.	Y
Reach Corridor	Bank stability (X4): the eroded and vegetative condition of both sides of the river banks.	N
	Riparian buffer zone width (m; X5): the width of natural vegetation from the edge of the stream bank out through the riparian zone.	Y
	Channel connectivity (X6): the number of barrages per hundred kilometres indicates the habitat status of fishes.	Y
	Sinuosity (X7): an increase in the stream length longer than in a straight line indicates the diversity of habitat and fauna.	Y
Flow Regime	Velocity/Depth (X8): patterns of velocity and depth indicate habitat diversity.	N
	Deviation of runoff distribution (X9): the difference in runoff distribution before and after water projects indicates the influence of aquatic organisms.	Y
	Satisfaction of ecological flow (%; X10): the days exceeding ecological water demand measured by Tennant method indicates the habitat status.	Y
Water Quality	Water quality level (X11): water quality classified by comprehensive assessment.	Y
	Eutrophication index (X12): characteristics of eutrophication including TP, TN, Chl-α, CODMn, and SD.	Y
	Surplus pollutant (X13): the total pollution load accounts for Total Maximum Daily Loads. We used COD as a calculated parameter.	Y
Aquatic Life	Species loss (%; X14): the loss in recorded species of fishes compared to the past.	Y
	Index of biodiversity (X15): the biodiversity measured by Shannon index.	Y
	Index of biotic integrity (X16): the relationship between anthropologic activities and aquatic organisms, such as fish or benthic invertebrates.	Y
Social Functions
Water Supply	Water resources per capita (m3; X17): the ratio of total available water resources to the population of the catchment.	Y
	Water consumption per GDP (Ұ ×104; X18): lower water consumption indicates higher water use efficiency.	Y
	Water resources development degree (%; X19): the ratio of exploitation and use amount to total available water resources.	Y
	Lack rate of water consumption (%; X20): the ratio of the difference between water requirement and water supply.	Y
Irrigation	Scale of efficient irrigation (%; X21): the ratio of water-efficient irrigation projects to total ploughland.	Y
	Irrigation efficiency (%; X22): the ratio of water output to water input.	Y
Water Purification	Water quality reaching standard (%; X23): the frequency of water analysis results meeting the functional target of the water function zone.	Y
	Wastewater treatment (%; X24): the ratio of treated wastewater to total wastewater in urban areas.	Y
Flood Regulation	Reservoir storage (X25): the ratio of flood regulation capacity of existing reservoirs to the average upstream runoff.	Y
	Levee status (X26): whether or not the designed high water levels of levees meet the requirements.	N
Aquatic Production/Navigation/Hydropower	Total pisciculture area (X27): the total fishing culture area of the catchment.	Y
	Navigation capacity (X28): the frequency of days deeper than the allowed water depth.	Y
	Hydropower exploitation rate (X29): the portion of hydropower generation under environmental safety.	Y
Aesthetics/Recreation	Comfortable degree of the landscape (X30): the landscape structure of the river includes trash absence, vegetative conditions, and other aesthetic indicators.	N
	Satisfaction of closeness to water (X31): whether the existing entertainment near rivers such as wetland parks, platforms, and other water-loving facilities are sufficient.	N

Note: The binary variable (Y/N) indicates if the factor is a quantitative indicator.

. The integrity or deficiency of these aspects indicate the health of a river ecosystem. Ecosystem services of a riverine system refer to the benefit obtained by anthropogenic activities. The structural indicators of a river ecosystem were further classified into five categories: watershed, river corridor, flow regime, water quality, and aquatic life, with different scales. The functional indicator categories depend on the preferences and concerns of stakeholders who fund the project and/or live nearby. Indicators of river ecosystem structures and ecosystem services functions are not entirely separate. The degradation of structural indicators may be related to the decrease or loss of functional indicators, whereas the functional development of a river system may also lead to the deterioration of the river ecosystem structure.

We obtained original information for these 31 indicators from three sources: (1) hydrological information and water resources materials from Harbin the Water Resources Bureau (HWRB) for indicators X₁–X₃, X₆, X₇, X₉–X₂₇, and X₂₉); (2) the eutrophication index [42], historical records of fish species, and benthic macroinvertebrate [13] data from published articles [43,44,45,46,47] for indicators X₁₂–X₁₆; and (3) field measurement and scoring according to the study of Barbour et al. [13] for indicators X₄, X₅, and X₈, and [6] (containing the indicators of X₃₀, X₃₁) in June 2014.

Structural and functional indicators were selected from Table 1. X₂₈ was not calculated because there was no demand from stakeholders. X₁₂ was used rather than X₁₁ for the eutrophic risk of the Xiquanyan Reservoir in the upper section. X₁₅ was used instead of X₁₄ given the lack of historical data in the upper section. X₉ was used instead of X₁₀ to indicate the influence of reservoirs, headworks, and other water conservancy works in the middle section.

To determine if the river has become degraded, the integrity of ecosystem structures and sustainability of social functions were both pre-evaluated by VFAM. We ranked the river health status into four grades: sick (I); degraded (II); sub-healthy (III); and healthy (IV). If the river ecosystem was not better than degraded (II), additional artificial techniques should be applied to the river restoration. Otherwise, the river would be restored by itself if the external disturbances are removed.

3. Results

3.1. Weight

The health of the three parts of the river (Figure 1) was estimated both on the ecosystem structures (S) and social functions (F) using VFAM in Table 1. The weights of the indicators (w_i) in Equations (10) and (16) were investigated and discussed by stakeholders, decision-makers, and scientists in August 2014, including officials from the agricultural and water sectors, sewage treatment engineers, region leaders, and nearby residents.

The weights of the structural (w_s) and functional (w_f) indicators were determined using the Analytic Hierarchy Process (AHP), which passed the consistency check. In this paper, because AHP has been so widely used, the details on the method will not be described here. More details and formal definitions can be found in the cited literature [48].

The distinction between economic development and the functional regionalization may lead to different emphases on river social functions. For instance, aesthetics and recreation will be emphasized in reaches that flow through urban areas, whereas irrigation is more important in rural rivers banked by farmland on both sides. Hence, we used different weights to evaluate the functional indicators. The weights of indicators in each section, determined by collective basin stakeholders, are shown in

Table 2. Weights of indicators for river social ecosystem assessment in the Ashihe River.

${\mathit{w}}_{\mathit{S}}$	X1	X2	X3	X4	X5	X6	X7	X8	X9	X10	X11	X12	X13	X14	X15	X16
U	0.04	0.02	0.04	0.08	0.09	0.03	0.02	0.15	-	0.07	-	0.15	0.07	-	0.12	0.12
M	0.04	0.02	0.04	0.08	0.09	0.03	0.02	0.15	0.07	-	0.15	-	0.07	0.12	-	0.12
L	0.04	0.02	0.04	0.08	0.09	0.03	0.02	0.15	-	0.07	0.15	-	0.07	0.12	-	0.12
${\mathit{w}}_{\mathit{F}}$	X17	X18	X19	X20	X21	X22	X23	X24	X25	X26	X27	X28	X29	X30	X31
U	0.05	0.07	0.1	0.03	0.03	0.12	0.08	0.15	0.01	0.09	0.04	-	0.01	0.17	0.06
M	0.09	0.04	0.04	0.04	0.08	0.23	0.13	0.04	0.03	0.13	0.03	-	0.01	0.06	0.06
L	0.05	0.04	0.06	0.08	0.03	0.03	0.19	0.06	0.02	0.12	0.02	-	0.01	0.07	0.21

Note: “-”means no data or not calculated. w_s and w_f represent the weights of the structural and functional indicators, respectively, the in upper (U), middle (M), and lower (L) section.

.

3.2. River Social Ecosystem Assessment

Grade eigenvalues were calculated using Equations (8)–(14). Four VFAMs were applied by varying the $\mathsf{\alpha }$ and $p$ parameters of Equation (11). Thus, four eigenvalues for the three parts were obtained. The mean and standard deviation for each section are shown in Figure 5. The S and F grade eigenvalues were 3.261 and 2.583 in the upper section, respectively, which means that the riverine social ecosystem of the upper section is sub-healthy. The S and F grade eigenvalues were 2.334 and 1.939 in the middle section, respectively, which means that the middle section is degraded, and the S and F grade eigenvalues of the lower section were 1.437 and 1.768, respectively, which means that the integrity of the ecosystem structure in the lower section is sick and degraded in the social functions of the ecosystem services.

These results show that the river is increasingly damaged from the source to the estuary, both in terms of the ecosystem structure and social functions. Moreover, the different degrees of disturbance due to anthropogenic activities were responsible for the health condition scoring result. The lower section should receive more attention than the other parts because the structure is severely sick and the functions cannot meet the requirement of the public. The degraded middle section should also be remolded and further analyzed. Although the upper section is barely satisfactory, ecosystem services functions are closer to degraded, so the risk of the ecosystem services decreasing should be noted. Thus, the restoration strategies for the different segments should not be identical, according to the assessment results. Additional engineering measures should be applied to the lower and middle sections to improve the ecosystem structures. Meanwhile, the upper section is more likely independently return to health if the external influences are removed by watershed management rather than incurring unnecessary project expenses.

4. Discussion

4.1. Deficiencies Analysis in Different Sections

Using Equations (15) and (16), we drew the NRDs for the three sections in S and F (Figure 6). We used four colors to indicate the condition of each indicator: sick is in red, degraded in yellow, sub-healthy in green, and healthy in blue. All of evaluated indicators were visually shown their conditions compared with each other in Figure 6. Deficiencies were separately analyzed from the NRDs by comparing the three sections.

4.1.1. Upper Section

The NRDs show that the poor water quality and water purification are highly-weighted deficiencies in the structures and functions of the upper river section. The Xiquanyan Reservoir is a reserve drinking water source for Harbin City, so the water quality requirement for this reach is higher than Grade II of Chinese environmental quality standards for surface water [49]. The reservoir is vulnerable to pollution from agriculture and village drainage with little wastewater treatment, which results in the inferior water quality. In addition, the irrigation efficiency is low in the upper section because of extensive irrigation. Although the level of flood control is low in this segment, enhancing the embankment is less urgent due to the lower population and ecological protection.

4.1.2. Middle Section

The inefficient irrigation, with an efficiency below 47% (data from HWRB), is one of the main deficiencies as shown by the NRDs. A large amount of water was pumped to grow rice, which resulted in a water shortage in the river. Six irrigation works were constructed along this section, which led to the degradation of the flow regime and river corridor. This explains the fair water quality but weak aquatic life. Because a large amount of water was diverted to grow rice in May, the decreased instream flow was insufficient for fish migrations, spawning, and other behaviors [50]. The vegetation cover declined due to slashing for cultivation in the watershed, which aggravates soil erosion. The water supply issue may be the factor restricting local development with the high water shortage rate of about 27% (data from HWRB).

4.1.3. Lower Section

The functional NRD of the lower section of the river, which meanders through urban areas, indicates that the current aesthetic and recreation situation is far from meeting the needs of the nearby townspeople. As shown in Figure 1, construction and demolition debris and domestic waste are dumped on flood lands and alongside riverbanks so that most of the hydrophytes and hygrophytes are hard to grow. Additionally, grazing by domestic livestock in riparian areas has destroyed the river corridor conditions. Because too much sewage is poured into the river, the water quality is inferior to Grade V of the Chinese environmental quality standards for surface water [51]. Many aquatic animals and amphibians lost their habitat required for survival, refuge, foraging and breeding. The poor water quality and river corridor led to the loss of biodiversity and landscape degradation. The number of fish species has decreased from 76 in 1995 to 61 in 2010 [45]. In addition, as a result of the high population density and the high level of industry in the area, the high water consumption pressure caused the immoderate exploration of water resources.

4.2. Goals Prioritization

The goal of restoring the river social ecosystem is directly correlated with the deficiencies in the structures and functions. The priority of the project depends on the degradation level and the understanding of the objectives. Generally, the larger uncolored areas of red and yellow pies comparing 6 NRDs in Figure 6 indicate the weaker aspects of the river ecosystem structure or social function, and also the top restoration priorities. Furthermore, we listed several goal categories [27] as shown in

Table 3. Major goal categories and stakeholder interests. Ratings of high (H), moderate (M), and low (L) indicate the priorities of goals based on the deficiencies analysis.

Restoration Goal	Upper Section	Middle Section	Lower Section
Stakeholder Interest	Irrigation improvement Flood protection	Irrigation improvement Water quality management	Aesthetics/Recreation Water quality management Flood protection
Vegetation Improvement	L	H	L
Channel reconnection	L	H	L
Riparian management	L	M	H
Floodplain reconnection	L	H	H
Bank stabilization	L	L	H
Flow modification	L	H	M
Water quality management	H	H	H
Fish passage	L	M	L
Instream species management	M	M	M
Irrigation improvement	H	H	L
Flood protection	H	M	L
Aesthetics/ Recreation	L	M	H

, and rated them into high, middle, and low priorities by the result of the deficiencies analysis and communication with decision makers and stakeholders.

Although several goal categories have been proposed, and given stakeholder demands, stakeholders tend to be interested in obtaining sufficient water to irrigate in the upper section and middle section, and providing natural and wild-like environments in the lower section. According to the results of VFAM, efforts should be focused on the lower section of the Ashihe River. The NRDs show that most of the structural indicators are sick in the lower section, and the water purification, aesthetic and recreation function cannot meet stakeholder desire. Based on the deficiency analysis, the poor water quality and river corridor are the main causes of river ecosystem degradation. So, water quality management and river corridor improvement goals such as riparian management, floodplain reconnection, and bank stabilization should be high priority. Aesthetic and recreation should also be high priority because of the dissatisfaction with aesthetic appearance and recreation as shown in NRDs. Although the river ecosystem does not work well in term of aquatic life in the middle and lower section, it will not turn better unless the water quality has improved. Hence, the instream species management should be middle priority. Because the amount of irrigation water being removed from the river influences the water quality, flow regime, and aquatic life, irrigation must immediately be improved, both in terms of water-saving irrigation and removal of redundant and dilapidated barrage for irrigation. It will contribute to improving the poor flow regime in the middle and lower section. For this reason, the irrigation improvement should be high priority in the upper and middle section and the channel reconnection in the middle section. Besides, the vegetation improvement should also be high priority due to the reduction of soil erosion and non-point source pollution in the middle section.

Many goals within each section and between sections are interrelated and cannot be tackled individually, or restoration efforts will be less efficient if tackled individually [5,8,29]. For example, in the lower section, the aesthetics, recreation, and water quality improvement goals are likely to benefit from riparian remediation actions and bank stabilization, and species management will likely need to consider flow modification and connectivity to be successful. Furthermore, water quality enhancement in the lower section is likely to be highly dependent on restoration activities in upstream sections. Hence, water quality enhancement is the primary goal in all sections of Ashihe River.

4.3. Preliminary Restoration Recommendations

To achieve the goal of restoring the Ashihe River social ecosystem, we recommended some preliminary actions based on the following criteria: (1) the local economy; (2) a comprehensive plan on the watershed scale; (3) landscape and aesthetics; and (4) preferentially selecting the low impact projects.

To achieve the goal of water quality enhancement, it is necessary to reduce the total pollution loads from the catchment. Hence, sewage treatment plants should be improved in cities and towns. The existing works, such as the Xinyi Sewage Work and the Acheng Sewage Work, should increase both volume of sewage treatment from Harbin and Acheng City and the effluent standard from Grade IB to IA [51]. New sewage works should be primarily constructed to collect domestic sewage from every town in the upper section.

To achieve the goal of riparian management, floodplain reconnection, channel connectivity, and bank stabilization, four restoration actions were recommended as follows, which will also contribute to water quality improvement. A comprehensive riparian system aims to filter contaminants from farmland drainage and other non-point source pollution. The system consists of an interception buffer zone, an artificial bosk buffer zone, a grasses filter buffer zone, a hygrophytes buffer zone, and an emergent hydrophytes zone from outside to inside the riverbank. Riparian revegetation should be designed according to the natural structure, soil matrix, and hydrologic conditions. Local plants, such as Populus and Tamarix, should be preferentially chosen for vegetation [52]. This system is suitable for water quality enhancement, aesthetics, riparian remediation, and other objectives throughout the entire river system [53].

River wetlands contribute to the maintenance of regional biodiversity [7]. Many wetland conservation and restoration projects have been undertaken to reconnect the floodplains in North America [54] and Europe [55,56]. Moreover, the importance of floating-leaved, submerged, and emergent macrophytes has been noted in wetland systems [57]. Native plants, such as Phragmites australis, Typha orientalis, Oenanthe javanica, and Nelumbo nucifera, could be selected to revegetate the river and benchland in the middle and lower sections.

An ecological embankment is a composite cross-section construction that addresses flood control, bank stability, floodplain connectivity, aquatic habitats, and aesthetics [58]. The distinctive features of an ecological embankment include plant cover and proximity to water. The ecological gabion system, vegetation geonets, porous concrete, and other new revetment techniques could be applied as urban embankments for the middle and lower sections.

The purpose of dam removal or retrofitting is restoring the channel connectivity and increasing discharge capacity [59]. Superfluous or ruined dams and buildings should be removed from the floodplain. Fish ladders should be installed in conventional dams to provide a passage for fish migration. Revegetation is also necessary for the larger dams and head-works.

The water diversion for irrigation resulted in the poor flow regime in the middle and lower section. Irrigation water-saving projects seek to obtain the most benefit with the lowest water consumption. Alternately submerged and non-submerged systems save water compared with continuous submergence, without affecting rice yield [60]. The government could provide incentives to farmers to grow alternative crops such as vegetables, corn, and other lower water use crops, rather than rice.

Artificial flood has been used to protect valued landscapes along the Colorado River [61]. By optimizing the operation of the Xiquanyan Reservoir Dam, an artificial flood could be used to remove fine sediments downstream and recover the spring flood that is diverted to grow rice in late spring or early summer. As artificial flood improves flow regimes in the channels and raises more nutrients from the sediment, it can also facilitate the riparian remediation and the habitats improvement [62].

Due to the dissatisfaction with aesthetic appearance and recreation, we gave two suggestions to improve it in the lower section. River cleaning includes ecological dredging and trash removal. Ecological dredging emphasizes the removal of polluted sediment for water quality improvement and the safe disposal of dredged sediment [63]. Trash should be dug up or caught from the water and transported to incinerators or landfills. Construction debris and demolition trash could also be used as embankment materials. This action could improve aesthetics and water quality of the river to some extent, especially in the lower section.

Shallow beaches and islands are scattered over the lower section of the Ashihe River, but they are severely disturbed by human activities. Wetland parks are conservation, education, and tourism locations [64]. Construction of a wetland park could create insulation from the communities and factories and provide the nearby residents with a physical exercise and leisure facility in the urban area.

Although the actions mentioned above are generally described and compared in

Table 4. Simple analysis of restoration recommendations in the upper (U), middle (M), and lower (L) river sections.

Action	Total Cost	Response Time	Duration	Suitability
				U	M	L
Riparian comprehensive system	Medium	Slow	Long	Yes	Yes	Yes
Riverine wetland	Medium	Slow	Medium		Yes	Yes
River cleaning	Low	Quick	Short			Yes
Ecological embankment	Medium	Medium	Medium		Yes	Yes
Dam removal or retrofit	Low	Quick	Long		Yes
Sewage treatment	High	Medium	Long	Yes	Yes	Yes
Wetland park	High	Medium	Medium		Yes	Yes
Artificial flood	Low	Quick	Short		Yes
Irrigation water-saving projects	High	Slow	Medium	Yes	Yes

, further discussion is required to ensure their efficacy and efficiency. The cumulative effect of small projects also requires better understanding to determine priority of actions [65]. As detail design is interdisciplinary, involving ecology, hydrology, civil engineering, and landscape architecture, repeated analysis and discussion by relevant scientists and practitioners is required.

5. Conclusions

A river is a complex and fuzzy social ecosystem. River social ecosystem structures and functions interact with each other. An excellent structure is the basis for perfect function, and immoderate functional development will lead to increased structural damage. River remoulding has developed from discovery, conservation, restoration, to effective actions since the 1960s [4]. Hence, effective actions should fully consider whether the structural status is healthy and the functional demands are adequate [8,66].

In this paper, we proposed a pre-assessment framework to determine the priority of the river social-ecosystem restoration objectives. The VFAM normalized the different dimensions of the fuzzy indicators and evaluated the indicators based on four assessment models. Both the structural and functional deficiencies were subsequently analyzed using a modified NRD method. The river deficiencies were visually shown by the modified NRDs. Thus, the objective and action priorities could be readily determined by the decision maker and communicated with stakeholders.

The river was divided into three sections based on the various functional demands. The Ashihe River assessment results show that the lower section should be the highest priority restoration section. Furthermore, the restoration priorities for the three sections are listed in this study. using NRDs, we found that water quality deterioration and irrigation works are the two key river degradation factors that must be improved throughout the entire watershed. Aesthetics and recreation should also be given priority to restore the lower section due to the strong demands of nearby residents. Additionally, some preliminary measures were generally discussed in accordance with the goal priority and four restoration criteria. However, determining the most effective actions is still a complex issue that requires further discussion among scientists, decision makers, and restoration practitioners. Hence, a comprehensive objective decision model should be established in a future study.

This framework enhances the linkage of each step and results of river social ecosystem assessments. The framework also has the potential to assist with communication with stakeholders, focusing on achieving a comprehensive goal. Therefore, the framework would provide a river restoration guideline for planning and designing coordination between social development and river ecosystem protection while considering all affected stakeholders.