A Dual-Scale Assessment System for Urban River Networks Based on the URBAN Framework

Abstract

Urban river networks face significant ecological challenges due to intensive urbanization. Traditional assessment methods focus mainly on individual rivers and overlook cross-scale connections. To fill this research gap, the study refined the Urban Riverscape Conditions-based Assessment for Management Needs (URBAN) framework and developed a dual-scale assessment system covering the entire river network and individual rivers. It evaluates hydrology, geomorphology, ecology, and the waterfront public service dimension. Taking the Qingxi area of Shanghai as a case study, this study integrated multi-source data and adopted field investigations, the analytic hierarchy process (AHP) and principal component analysis (PCA) to collect field data, calculate indicator weights, and extract dominant functional factors. The results show that the overall comprehensive health score of the study area is 59.39, classified as average; the river network scale scores 58.34, and the 21 monitored rivers achieve an average score of 61.80. The assessment identifies clear advantages in hydrological and geomorphological conditions, whereas waterfront public services and river morphological diversity are still deficient. Overall, this system demonstrates good operability and scientific validity, providing practical technical approaches for sustainable urban river network management and supporting refined watershed governance.

1. Introduction

Rivers are the source of life and the foundation of human activities, carrying crucial ecological functions and socio-economic values [1]. Water resources constitute the indispensable basis for human survival and development and represent core elements underpinning the sustainable progress of socio-economic systems [2,3]. Consequently, the efficient utilization and scientifically grounded management of water resources are critical to advancing long-term socio-economic development [4]. A river network refers to the interconnected drainage pattern formed by rivers, lakes, artificial channels, and other water bodies within a basin [5]. These networks function as hydrological continua that sustain species dynamics and a wide array of ecological processes. A well-connected river network promotes biological exchange, energy flow, and nutrient cycling, thereby maintaining ecological stability amid environmental changes [6,7,8,9].

However, rapid urbanization and intensive exploitation of water systems have profoundly altered river networks worldwide, transforming many natural systems in urban areas into complex natural-artificial hybrid networks [10]. Hydrological connectivity is influenced by spatial organization, and the dynamic behavior of river networks reflects the spatial-temporal heterogeneity of hydrological processes [11]. Large-scale human interventions have become the dominant driver reshaping connectivity and causing widespread degradation of freshwater ecosystems [12,13,14]. This has triggered systemic aquatic ecological crises, including deteriorating water quality, riparian habitat degradation, impaired river network connectivity, and shrinking aquatic habitats [15,16,17]. In the context of global Sustainable Development Goals (SDG 6) and advancing urban ecological civilization, scientific assessment of urban river network landscapes has become essential for balancing high-intensity urban development with ecological conservation. Such assessments are of great significance for promoting the coordinated evolution of human settlements and aquatic ecosystems. This is particularly true for densely distributed river networks, such as those on the Shanghai Plain, which play a central role in regional disaster prevention, mitigation and ecological maintenance [18].

To evaluate river health, scholars worldwide have developed various monitoring and assessment frameworks, such as the United States’ Rapid Bioassessment Protocols (RBPs) [19], the United Kingdom’s River Habitat Survey (RHS) [20], the Index of Biotic Integrity (IBI) [21], and several modified assessment systems developed in China [22,23,24,25]. Nevertheless, these frameworks predominantly focus on individual rivers or localized habitats, often neglecting the intrinsic interconnections between natural and cultural attributes of urban river networks at a systematic, basin-wide scale. In response to this gap, Murphy et al. [26] proposed the Urban Riverscape Conditions-based Assessment for Management Needs (URBAN) framework. This approach integrates hydrological, geomorphological, and socio-ecological dimensions. It emphasizes the continuity, hierarchy, and spatial diversity of river landscapes as the ecological and cultural link between nature and cities, thereby providing a robust methodological foundation for the systematic management of urban rivers. The present study adopts the URBAN framework because of its holistic perspective, which aligns closely with the need to treat urban river networks as integrated socio-ecological systems.

Previous studies on river networks have typically been conducted either at the single-river scale or at the broader basin scale, without simultaneously addressing multi-scale interactions between the overall river-network pattern and the characteristics of individual river segments. As a result, they have been unable to fully capture the structural and functional coupling mechanisms that operate across scales. Assessment of such regions therefore requires a dual-scale perspective that simultaneously encompasses the macro-level river network pattern and the micro-level characteristics of individual river segments. This study transcends the limitations of conventional single-river assessment models by developing a multi-scale comprehensive evaluation system for urban river networks. It spans four key dimensions, including hydrology, geomorphology, ecology, and waterfront public service, thereby addressing an important research gap while offering an innovative approach to understanding the multi-scale and multi-dimensional dynamics of urban river landscapes.

The urban river network of Shanghai’s Qingxi area is a typical case for this type of assessment. With a distinct urbanization gradient across its territory, the Qingxi area provides an ideal heterogeneous sample for multi-scale “river network-river” landscape monitoring in the context of urban river networks. Its dense, hybrid natural-artificial configuration, combined with its critical functions in flood control, water supply, and ecological support, makes it highly representative of rapidly urbanizing plains in China and elsewhere. Building upon the URBAN framework and refining it to suit the distinctive structural and functional traits of urban river networks, this study examines the overall supply characteristics of the water system’s structure and function alongside localized response patterns in the Qingxi area. It elucidates the multi-scale and multi-dimensional coupling mechanisms of the urban river network and explores their comprehensive influence on regional river landscape conditions. In doing so, it aims to provide scientific support for the sustainable development of urban water ecosystems and the harmonious coordination of regional ecological security with human well-being.

2. Materials and Methods

2.1. Study Area

Shanghai is situated on the front of the Yangtze River Delta in eastern China (30°40′~31°53′ N, 120°52′~122°12′ E). It is a typical urban river network region. The Qingxi area encompasses the towns of Jinze, Liantang and Zhujiajiao, with a total administrative area of 670.14 km². Located in the upper reaches of the Huangpu River, Shanghai’s mother river, the Qingxi area plays a vital role in safeguarding the ecological and hydrological quality of the entire river system. With a distinct urbanization gradient across its territory, the Qingxi area serves as an ideal heterogeneous sample for multi-scale river network and river landscape monitoring in urban settings. The study area features a well-developed river and lake system, with 2185 rivers of different levels stretching for a total of 2503.45 km; its river network density reaches 3.74 km/km², and the water surface ratio stands at 18.8%. Figure 1 presents an overview of the study area.

2.2. Data Acquisition and Processing

This study constructed a targeted data processing chain for the Qingxi area river landscape assessment, featuring multi-source collection, classification analysis, and integrated output as core sequential steps. Multi-source research data were collected from multiple authoritative channels, as detailed in

Table 1. Data sources and characteristics.

Data Name	Year	Data Sources	Data Description
Sluice Operation Data	2022	Water Bureau of Qingpu District, Shanghai, https://www.shqp.gov.cn/water/ (accessed on 26 January 2024)	/
Water Quality Data	2022	Water Bureau of Qingpu District, Shanghai, https://www.shqp.gov.cn/water/ (accessed on 26 January 2024)	Mean values of monthly data.
Benthic Macroinvertebrate Data	2022	Water Bureau of Qingpu District, Shanghai, https://www.shqp.gov.cn/water/ (accessed on 26 January 2024)	Mean values of data collected in March and August.
Fish Data	2022	Water Bureau of Qingpu District, Shanghai, https://www.shqp.gov.cn/water/ (accessed on 26 January 2024)	Mean values of data collected in March and August.
River Network Data	2021	Institute of Geographic Sciences and Natural Resources Research (IGSNRR), Chinese Academy of Sciences (CAS), https://www.cas.cn/ (accessed on 30 November 2023)	Vector river network data with a mapping scale of 1:25,000.
Land Use Data	2020	European Space Agency (ESA), https://esa-worldcover.org/ (accessed on 13 December 2023)	Derived from Sentinel-1 and Sentinel-2 satellite imagery; 10 m spatial resolution. Covers six typical land use types, including cropland, forest, grassland, built-up areas, permanent water, and bare land.
Remote Sensing Image Data	2020	Google Earth Engine (GEE), https://earthengine.google.com/ (accessed on 07 December 2023)	Based on Landsat 8 Surface Reflectance products; 30 m spatial resolution.
POI Data	2020	Amap (AutoNavi), https://ditu.amap.com/ (accessed on 03 December 2023)	Point of Interest (POI) data for scenic spots and tourist attractions.
Population Data	2020	WorldPop, https://hub.worldpop.org/ (accessed on 11 December 2023)	Gridded population data with a spatial resolution of 100 m.
Field Survey Data	2023	Field investigation conducted by trained environmental science professionals	Conducted in accordance with the UK’s Urban River Survey (URS) manual [27]. Each river segment was surveyed for a distance of 1 km with 5 sampling sections set at 200 m intervals. Both banks of each section were comprehensively evaluated, and the average value was used as the final score for the entire river reach. The central location of the surveyed segment is illustrated in Figure 1.

. Quantitative analysis was then implemented with dedicated tools matched to specific research indicators. In the quantitative analysis phase, MATLAB 2021b, SPSS 27, and ArcGIS 10.8were applied to calculate the full set of indicators. This analytical process ultimately integrated all obtained results to establish a fundamental database of river landscape conditions. It encompasses 4 dimensions and 20 key indicators, including 8 for the river network scale and 12 for the river scale.

2.3. River Network Landscape Condition Indicator System

Relying on the improved URBAN framework, this study develops a dual-scale “river network–river” evaluation system. Its relationship matrix is presented in Figure 2.

Indicator selection adheres to scientific validity, representativeness and innovation. Scientific validity reflects the physical and biological characteristics of urban river networks, while representativeness covers the core management-dimension elements of the Yangtze River Delta. Key indicators, including the abundance of waterfront recreational facilities and the accessibility of public riverbanks, are selected to characterize the waterfront public service dimension. Each indicator parameter is selected based on existing river assessment protocols worldwide and validated via field monitoring of the Yangtze River Delta urban rivers. This guarantees the evaluation system’s applicability to the Shanghai urban river networks. The specific calculation methods for the four dimensions are detailed in

Table 2. Calculation methods of various indicators at the river network scale.

Evaluation Level	Evaluation Metrics	Calculation Method
Hydrological elements	RNCON [28]	Measured jointly by node connectivity rate β and water system integration γ. β = m/n, where m represents the number of river links between nodes, and n represents the number of nodes in the water system network; γ = m/3(n − 2), where m represents the number of river links between two nodes, and n represents the number of nodes in the water system network.
	RWSR [29]	Wp = RS/S, where Wp refers to the river water surface ratio, RS represents the total area of all rivers within the study area, and S represents the area of the study area.
Geomorphological elements	RNLR [30]	α = (m − n + 1)/(2n − 5), α refers to the loop rate of the river network, m is the number of river links between two nodes, and n is the number of nodes in the water system network.
	RNCOM [31]	RC = N0 × (L/Lm), RC refers to the complexity of the river network, and N0 represents the number of river orders. In this study, according to the “Shanghai River Management Regulations,” rivers are divided into four categories: municipal-managed, district-managed, town-managed, and village-level. L represents the total length of all rivers, and Lm represents the total length of main rivers, here referring to municipal-managed rivers.
Ecological elements	NDVI [26]	NDVI = (NIR − R)/(NIR + R), NDVI refers to the Normalized Difference Vegetation Index, NIR represents the reflectance of the near-infrared band, and R represents the reflectance of the visible red band.
	CONTAG [26]	$CONTAG=\left[1+{\displaystyle \frac{{\displaystyle {\sum }_{i=1}^x}{\displaystyle {\sum }_{k=1}^x}\left[\left(P_i\right)\left[\frac{G_{ik}}{{\displaystyle {\sum }_{k=1}^x}{G}_{ik}}\right]\right]\times \left[\mathit{ln}\left(P_i\right)\left[\frac{G_{ik}}{{\displaystyle {\sum }_{k=1}^x}{G}_{ik}}\right]\right]}{2\mathit{ln}\left(x\right)}}\right]\times 100,$ CONTAG refers to the contagion index, Pi represents the ratio of the area of landscape type i to the total landscape area, and x is the total number of different landscape types. Gik represents the probability that two randomly selected adjacent grids are of landscape types i and k.
Waterfront public service dimension	GSPDI [32]	$G={\displaystyle \frac{p}{S}},$ G refers to the green space park distribution index, p represents the number of POI points classified as green space parks in each area, and S is the area of each region.
	CLUI [33]	$L=100\times {\displaystyle {\sum }_{i=1}^k}\left(S_i\times A_i\right),$ L refers to the comprehensive index of land use, Si indicates the grading of land use degree, and Ai represents the ratio of each type to the total area.

and

Table 3. Calculation methods for various river-scale indicators.

Evaluation Aspect	Evaluation Indicators	Calculation Method
Hydrological elements	RFV [34]	The river flow velocity is denoted by v and is measured using the LS45A cup-type current meter.
	WFCR [26]	WC = TC/T, WC refers to water connectivity, TC indicates the time of water free flow, and T represents a year.
	RFTG [23]	According to the 10 types of water flow listed in the RHS manual, four types suitable for Shanghai urban rivers were selected; under windless conditions, the survey river sections were scored, with the scoring criteria referring to the manual, as detailed in Table S1.
Geomorpholo-gical elements	RCC [26]	S = Lr/Lv, S refers to the river corridor curvature, Lr is the actual length of the river centerline, and Lv is the straight-line distance between the start and end points of the river.
	RPSR	$R_p={\displaystyle \frac{{\sum }_{i=1}^m{A}_{pi}}{A_b}},$ Rp refers to the permeable surface ratio of the shoreline, Api is the area of the i-th type of permeable land use, and Ab is the area of the buffer zone
	PBPMG [23]	Based on the four grades of advantageous revetments listed in the RHS manual, corresponding scoring standards are set, as detailed in Table S1.
Ecological elements	CCME WQI [26]	$WQI=100-\left({\displaystyle \frac{\sqrt{F_1^2+F_2^2+F_3^2}}{1.732}}\right), F_1={\displaystyle \frac{n}{N}}\times 100,$ F1 represents the percentage of parameters that exceeded the standard limit at least once among all monitored water quality parameters, N refers to the total number of parameters in water quality monitoring, and n is the number of parameters that exceeded the standard limit. $F_2=\left({\displaystyle \frac{m}{M}}\right)\times 100,$ F2 represents the proportion of any monitored parameter exceeding its specified standard limit within the time range of the collected samples, where M is the total number of water quality monitoring samples, and m is the number of samples exceeding the standard limit. $F_3={\displaystyle \frac{P}{0.01P+0.01}}, P={\displaystyle \frac{{\displaystyle {\sum }_{i=1}^n}Q}{M}}, Q=\left\{\begin{array}{c}{\displaystyle \frac{V_{ob}}{V_s}},V_{ob}<V_s\\ V_s,V_{ob}=0\\ {\displaystyle \frac{V_s}{V_{ob}}},V_{ob}>V_s\end{array}\right.,$ Vob represents the specified standard value of a water quality parameter, Vs represents the actual measured value of a water quality parameter, P is the normalized index used to calculate the ratio of the cumulative sum of exceedance multiples to the total number of samples, and F3 defines the range of P between [0, 100]. The water quality grades and score ranges are shown in Table S2.
	B-IBI [34]	Set reference points and disturbance points to establish candidate indicators. For biological indicators whose index value decreases with increased disturbance, the formula is $B_i=B/B_{95\%}$, where Bi is the calculated score of the i-th indicator, B is the measured value of the indicator, and B95% is the 95th percentile value of the measured indicator; for indicators whose index value increases with increased disturbance, the 5th percentile value is chosen as the optimal expected value. The formula is $B_i=(B_{max}-B)/(B_{max}-B_{5\%})$, where Bmax is the maximum measured value of the i-th indicator, and B5% is the 5th percentile value of the measured indicator. By calculating the indices of each biological indicator and accumulating them, the B-IBI value at each sampling site is obtained, and the health evaluation levels are set based on the reference points, as shown in Table S2.
	F-IBI [34]	Set reference points and disturbance points to establish candidate indicators. For biological indicators whose index decreases with increased disturbance, the formula is $F_i=F/F_{95\%}$, where Fi is the calculated score of the i-th indicator, F is the measured value of the indicator, and F95% is the 95th percentile of the measured values of the indicator; for indicators whose index increases with increased disturbance, the 5th percentile is used as the optimal expected value. Its formula is $F_i=(F_{max}-F)/(F_{max}-F_{5\%})$, where Fmax is the maximum measured value of the i-th indicator and F5% is the 5th percentile of the measured values of the indicator. By calculating the indices of each biological indicator and summing them, the F-IBI value of each sampling site is obtained, and the health evaluation level is set according to the reference points as shown in Table S2.
Waterfront public service dimension	RCD [26]	Using the Buffer tool in ArcToolbox of the ArcGIS software and the Extract by Mask tool in Spatial Analyst Tools, you can obtain the mean population density data within the buffer zone.
	PWA [34]	By combining remote sensing image analysis, Gaode Map data, and field survey methods, the left and right banks of the selected rivers are delineated, and the waterfront distances available for public leisure and recreational activities are determined. In this process, any buildings that may hinder the public from freely walking, engaging in sports, or enjoying waterfront services, such as residential communities, industrial facilities, and educational institutions, will be marked as inaccessible areas.
	ASRF [23]	Based on the total number of types of waterfront facilities on both sides of each river, determine the richness of waterfront facilities for each river. The types of waterfront facilities, specific scoring criteria, and on-site photos are shown in Table S1.

.

2.4. Dual-Scale Scoring for River Network-River

For the evaluation of urban river networks in the Shanghai Plain, this study employs a dual-scale coupled scoring method based on existing standards and threshold-setting approaches for urban river network assessment [35,36]. Scoring benchmarks are developed by integrating the key threshold method, natural breakpoint method and expert consultation method (

Table 4. Scoring standards for dual-scale evaluation.

Evaluation Criteria	Evaluation Indicators	Scoring Methods
River network scale	RNCON	Evaluated jointly by node connectivity rate β and water system integration γ: 0 < β < 1, score 1; β > 1, score 2; β > 1 and γ < 1/3, score 3; β > 1 and γ > 1/3, score 4.
	RWSR	Wp < 0.1%, score 1; Wp = [0.1%, 1%), score 2; Wp = [1%, 5%), score 3; Wp = [5%, 10%), score 4; Wp ≥ 10, score 5.
	RNLR	α = 0, score 1; α = (0,0.2], score 2; α = (0.2,0.5], score 3; α = (0.5,0.8], score 4; α > 0.8, score 5.
	RNCOM	RC < 4, score 1; RC = [4, 8), score 2; RC = [8, 16), score 3; RC > 16, score 4.
	NDVI	NDVI < 0.1, score 1; NDVI = [0.1, 0.4), score 2; NDVI = [0.4, 0.7), score 3; NDVI > 0.7, score 4.
	CONTAG	CONTAG = (0, 20), score 1; CONTAG = [20, 40), score 2; CONTAG = [40, 60), score 3; CONTAG = [60, 80), score 4; CONTAG = [80, 100), score 5.
	GSPDI	G = [0, 5%), score 1; G = [5%, 15%), score 2; G = [15%, 30%), score 3; G = [30%, 50%), score 4; G ≥ 50%, score 5.
	CLUI	L = [100, 200), score 1; L = [200, 300), score 2; L = [300, 400), score 3.
River scale	RFV	v < 0.1, score 1; v = [0.1, 0.3), score 2; v = [0.3, 0.6), score 3; v ≥ 0.6, score 4.
	WFCR	WC < 40%, score 1; WC = [40%, 60%), score 2; WC = [60%, 80%), score 3; WC ≥ 80%, score 4.
	RLTG	Still, score 1 point; uniform wave, score 2 points; ripple, score 3 points; broken standing wave, score 4 points. If it is a transitional state between two, take the average of the two.
	RCC	S = [1.0, 1.05], score 1; S = [1.06, 1.29], score 2; S = [1.3, 3.0], score 3; S > 3.0, score 4.
	RPSR	Rp < 30%, score 1; Rp = [30%, 40%), score 2; Rp = [40%, 60%), score 3; Rp > 60%, score 4.
	PBPMG	Hard protective shore, score 1; open protective shore, score 2; degraded protective shore, score 3; no protective shore, score 4.
	CCME WQI	Bad, score 1; Poor, score 2; Average, score 3; Good, score 4; Excellent, score 5.
	B-IBI	Bad, score 1; Poor, score 2; Average, score 3; Sub-healthy, score 4; Healthy, score 5.
	F-IBI	Bad, score 1; Poor, score 2; Average, score 3; Sub-healthy, score 4; Healthy, score 5.
	RCD	Low aggregation level = [12.94, 19.18), score 1; medium-low aggregation level = [19.18, 22.56), score 2; medium-high aggregation level = [22.56, 29.64), score 3; high aggregation level = [29.64, 39.88], score 4
	PWA	Low accessibility < 45%, score 1; Medium accessibility = [45%, 90%], score 2; High accessibility > 90%, score 3.
	ASRF	1–2 types of facilities, score 1; 3–4 types of facilities, score 2; 5–6 types of facilities, score 3; 7–8 types of facilities, score 4

), enabling coordinated assessment of macro-structural characteristics and micro-functional performance. At the river network scale, the evaluation focuses on verifying whether the regional background health baseline is met. At the river scale, principal component analysis (PCA) is applied to accurately identify functional deficiencies of specific river sections.

This study adopts the analytic hierarchy process (AHP) to establish a dual-scale weighting system and comprehensive evaluation framework for the river network scale and river scale. This method quantifies the contribution of each scale to watershed landscape health assessment. A three-level hierarchical structure is established, as shown in

Table 5. Indicator system and weights for the dual-scale assessment framework.

Goal Layer	Criteria Layer	Weight	Sub-Criteria Layer	Weight	Final Weight
Comprehensive health assessment of urban river systems	River network scale	0.70	Hydrological elements	0.35	0.24
			Geomorphological elements	0.19	0.13
			Ecological elements	0.35	0.24
			Waterfront public service dimension	0.11	0.08
	River scale	0.30	Hydrological elements	0.28	0.09
			Geomorphological elements	0.16	0.05
			Ecological elements	0.39	0.12
			Waterfront public service dimension	0.17	0.05

. Eight experts in related fields conduct pairwise comparisons to determine the relative importance of indicators at each level. The geometric mean method is used to integrate expert judgments, and a reciprocal judgment matrix is constructed accordingly. A consistency test (consistency ratio CR < 0.1) is performed to verify the logical rationality of the matrix, and the weight values of all indicators are calculated.

This study uses a 100-point quantitative scoring system. For the river network scale, standardized scores of sub-indicators are weighted and summed to obtain the comprehensive score. For the river scale, each river in the study area is evaluated independently using corresponding indicator weights, and the average score of all rivers is regarded as the comprehensive score of the river scale. The final comprehensive assessment score of regional watershed landscape health is obtained through the weighted summation of the two scale scores based on the weights determined by the AHP method. Based on the 100-point scoring system, the water system health status is divided into five grades: Good (85–100 points), Relatively good (69–84 points), Average (53–68 points), Relatively poor (37–52 points), and Poor (0–36 points). This classification is uniformly applied to the river network scale, river scale, and the comprehensive evaluation score.

For the river network scale, the scoring criteria of each evaluation indicator differ. To eliminate dimensional differences, raw scores of all indicators are first linearly standardized to dimensionless values in the range of 0–1, as follows:

$$S_i={\displaystyle \frac{X_i-X_{i,min}}{X_{i,max}-X_{i,min}}}$$

(1)

where S_i is the standardized score of indicator i, derived from its raw score X_i, with X_{i, min} and X_{i, max} denoting the minimum and maximum scores defined in the corresponding scoring criteria. Next, the average standardized score for each sub-criterion layer is calculated: the arithmetic mean of the standardized scores of all indicators under the same sub-criterion layer is taken as the composite score of that element. Subsequently, the composite score of the river network scale is computed via weighted summation, using the sub-criterion layer weights determined by the AHP:

$$R_{network}=100\times {\sum }_{j=1}^4{S}_j\times w_j$$

(2)

where R_network is the composite score of the river network scale on a 100-point scale, S_j is the average standardized score of the j-th sub-criterion layer, and w_j is its corresponding weight.

For the river scale, the same standardization and weighting procedure was applied. First, raw indicator scores were linearly standardized, and average standardized scores were computed for each sub-criterion layer. The composite score for each individual river was then calculated as:

$$R_i=100\times {\sum }_{j=1}^4{S}_{ij}\times w_j$$

(3)

where R_i is the composite score of the i-th river on a 100-point scale, S_ij is the average standardized score of the j-th sub-criterion layer for river i, and w_j is the weight of the j-th sub-criterion layer under the river scale. Finally, the arithmetic mean of the composite scores of all rivers in the study area is computed to obtain the composite score of the regional river scale:

$$R_{river}={\displaystyle \frac{1}{n}}\times {\sum }_{i=1}^n{R}_i$$

(4)

where R_river is the composite score of the river scale, and n is the total number of rivers in the study area. The final comprehensive assessment score of regional river landscape health is then calculated as the weighted sum of the dual-scale composite scores, using the AHP-derived weights for the two scales:

$$R_{final}=w_{network}\times R_{network}+w_{river}\times R_{river}$$

(5)

where R_final is the overall health score of the river system on a 100-point scale, and R_network and R_river are the weights of the river network scale and river scale, respectively, determined by the AHP (Table 5).

To evaluate the robustness of the assessment results, reduce the subjective uncertainty of AHP-derived weights, and quantify the contribution of each scale’s weight to the final health score, this study adopted the one-factor-at-a-time (OAT) method, a widely used local sensitivity analysis approach in environmental assessment models [37]. We set a series of reasonable weight ratios between the river network scale and single river scale within the range of 0.1–0.9, while ensuring the sum of the two weights remained equal to 1. For each scenario, the final comprehensive health score and corresponding health grade were recalculated using the established evaluation formula to verify both the stability of the assessment outcomes and the relative influence of each scale’s weight on the final result.

3. Results

The indicator weights for the dual-scale assessment, determined by the analytic hierarchy process, are presented in Table 5. The corresponding questionnaire used in this method is provided in Table S3 of the Supplementary Material.

3.1. Assessment Results at the River Network Scale

Based on the weighting framework, the comprehensive score of the Qingxi area’s river network is 58.34, rated as average. The region has stable hydrological and geomorphological support. Human development and ecological protection maintain a relatively high degree of coordination. Overall, the river network structure remains intact, while the ecological background and waterfront public service capacity are relatively insufficient.

For hydrological and geomorphological characteristics, the node connectivity (1.24) and water system integration (0.42) indicate that the river network has developed into a mature, typical network structure. The 10.91% river water surface ratio reflects a high retention rate of natural water bodies in the region. The river network loop rate (12.27%) and complexity (61.53) confirm that its tributary system is well-developed and hierarchically rich, providing excellent geomorphological heterogeneity. These hydrological and geomorphological indicators collectively demonstrate the structural integrity of the study area’s river network.

For the ecological and waterfront public service, the Normalized Difference Vegetation Index (NDVI, average 0.04) and spread index (average 53.75) exhibit a spatial pattern that is low in the northwest and high in the southeast (Figure 3). This pattern stems from habitat fragmentation caused by dense lakes and marshes in the northwest, which further leads to spatial fluctuations in landscape connectivity. The Qingxi area has nine green parks, with a green park distribution index of 0.03, reflecting low green park distribution density and inadequate spatial supply capacity for the waterfront public service dimension. The land use degree index is 252.92, with over 77.80% of land classified as natural or agricultural. These indicators confirm the region’s low-intensity urbanization characteristics, which are based on lakes and polders. Restricted by ecological spatial differentiation and inadequate supporting layout of waterfront public service facilities, the overall ecological and waterfront public service capacity of the river network fails to reach a good level.

3.2. Assessment Results at the River Scale

The average comprehensive score of the 21 monitored rivers is 61.80 points, which is an average grade. Among the 21 rivers, one is rated relatively good, one is rated relatively poor, and the remaining 19 are classified as average, with no rivers reaching the good level (

Table 6. Results of indicators at the river scale.

River Name	RFV	WFCR	RLTG	RCC	RPSR	PBPMG	CCME-WQI	B-IBI	F-IBI	RCD	PWA	ASRF
Beiheng Gang	0.07	100.00%	3.50	1.04	96.50%	1.00	84.33	2.70	3.87	19.83	60.15%	2.00
Dazheng Tang	0.61	75.07%	4.00	1.02	94.65%	1.00	92.76	1.42	4.49	14.15	98.46%	1.00
Dianpu River	0.21	99.45%	3.75	1.04	76.87%	1.75	71.01	2.35	3.54	31.23	72.41%	3.00
Dianshan Gang	0.11	99.45%	3.00	1.12	93.53%	2.00	69.90	2.20	3.83	30.69	47.44%	1.00
Dongtang Gang	0.14	87.12%	3.00	1.00	78.79%	2.00	84.17	2.20	3.87	21.01	74.74%	1.00
Fan Tang	0.34	75.07%	4.00	1.13	91.63%	1.00	83.20	2.40	4.49	16.65	91.71%	1.00
Huatian Jing	0.19	99.45%	4.00	1.00	98.34%	1.00	83.17	1.87	3.56	12.94	32.41%	1.00
Jishui Gang	0.21	85.48%	3.50	1.06	86.15%	1.00	69.90	2.23	3.51	19.27	82.41%	1.00
Lanlu Gang	0.26	98.36%	4.00	1.02	96.70%	1.00	74.64	2.54	3.79	15.84	100.00%	3.00
Liansheng Shuhe	0.18	97.53%	3.00	1.15	90.75%	1.00	83.75	2.32	3.87	22.17	40.86%	1.00
Maoyang Gang	0.40	57.53%	3.50	1.01	88.68%	1.50	83.76	2.09	4.18	26.74	25.22%	1.00
Nanheng Gang	0.16	97.53%	3.00	1.02	93.16%	1.00	83.79	2.56	3.88	19.18	32.39%	1.00
Shenxiang Zhongxin He	0.16	99.45%	3.00	1.01	77.92%	1.00	69.65	2.45	3.67	29.64	65.94%	1.00
Shitang Gang	0.08	100.00%	3.00	1.01	68.90%	3.00	83.73	2.60	3.87	29.16	52.62%	1.00
Taipu River	0.35	86.85%	4.00	1.01	97.21%	1.00	85.37	2.38	3.47	22.76	95.75%	2.00
Wangyang Gang	0.05	85.48%	3.00	1.01	42.07%	1.00	69.81	2.14	3.51	22.35	76.78%	1.00
Xintang Jiang	0.03	99.45%	3.00	1.12	93.82%	3.00	70.73	2.38	3.84	26.72	24.16%	1.00
Xuqi Jiang	0.03	100.00%	3.00	1.34	82.17%	3.50	69.45	2.59	3.51	13.98	57.94%	1.00
Yuhui Tang	0.39	75.07%	3.50	1.29	84.97%	1.00	84.65	1.69	4.49	22.56	38.91%	1.00
Zhukun River	0.07	99.45%	3.50	1.02	84.97%	2.50	70.49	2.22	3.84	33.36	41.73%	1.00
Zhumao River	0.14	99.45%	3.00	1.02	80.14%	1.00	83.18	2.65	3.90	39.88	15.05%	1.00

and Figure 4).

For hydrological characteristics, the rivers generally exhibit gentle flows. A total of 76.19% of them have flow speeds ranging from 0.03 to 0.26 m/s, with main streams flowing faster than tributaries. A total of 61.90% of the rivers maintain a connectivity rate of 97.53% to 100% and remain mostly unobstructed year-round. The only exception is the Maoyang Gang, where strict water-gate management leads to lower connectivity. Flow types are dominated by ripples, broken standing waves, and transitional states, with no stagnant water observed. These patterns reflect the effectiveness of active water regulation in the embanked area as well as the integrity of natural micro-topography. Hydrological conditions are a relatively advantageous dimension of river-scale evaluation, supporting the basic ecological functions of the river system.

Geomorphologically, natural meanders are retained in the Xuqi Jiang and Yuhui Tang. River channel curvature varies across the study area. 19.05% of rivers are moderately curved, 61.90% are slightly curved, and 9.52% are nearly straight. Shoreline permeability also differs among sites. For instance, 47.62% of rivers have permeable shorelines exceeding 90%, while the Wangyang Gang shows notably limited shoreline permeability. Embankment types are diverse. Ecological embankments such as those using timber piles and soil in the Xuqi Jiang or reed and timber piles in the Xintang Jiang help balance ecological needs with flood control. Nevertheless, hard embankments still dominate and account for 61.90% of rivers. Widespread river straightening and hard revetment construction have suppressed the geomorphological diversity of most rivers, becoming one of the main limiting factors for river health improvement.

Ecological monitoring results indicate that 57.14% of the rivers exhibit good water quality. These rivers support a total of 43 large benthic invertebrate species and 38 fish species. Clear spatial patterns emerge in biological integrity. B-IBI values show a declining trend from northeast to southwest. In contrast, F-IBI values first increase and then decrease from west to east. Generally, water quality and aquatic biological integrity are maintained at a medium-to-good level, forming a favorable ecological foundation for river landscape health.

From a waterfront public service perspective, the average riverbank population density is 23.34 people/hm². High-density clusters are located in the northeast urbanized and tourist areas, such as the intersections of the town of Zhujiakou and Shenxiang Zhongxin He. Low-density areas lie at the regional periphery, such as Dazheng Tang and Xuqi Jiang, among others. Shore accessibility is highly variable. Approximately 47.61% of rivers have accessibility above 60%, yet 52.38% lack sufficient accessibility due to building or farmland encroachment. Waterfront facilities are generally underdeveloped. Only a few areas, like the Lanlu Gang, offer relatively diverse facilities. Waterfront boardwalks and steps, among other features, indicate consideration for public accessibility and environmental compatibility. Insufficient riverside accessibility, limited provision of leisure service facilities, and unbalanced spatial layout are prominent limiting factors at the river scale, which may constrain the overall health status of the river system.

Integrating PCA conducted on 12 river-scale indicators, this study precisely identifies the functional characteristics of river landscapes. The analysis yielded a KMO value of 0.537 and a Bartlett’s test of sphericity of 110.373 (p < 0.001). These results confirm the data’s suitability for factor analysis. Following the eigenvalue-greater-than-1 criterion, four principal components were extracted. They cumulatively explain 75.223% of the total variance. After varimax rotation (

Table 7. Loading coefficients of principal components.

Evaluation Indicators	Component 1	Component 2	Component 3	Component 4
River Flow Velocity	−0.778	0.377	0.316	−0.221
Water Flow Connectivity	0.866	−0.141	0.070	0.055
River Channel Flow Type Score	−0.283	0.706	0.416	−0.255
River Corridor Curvature	−0.074	−0.008	0.087	0.873
Shore Permeable Surface Rate	−0.019	0.142	0.918	0.087
Dominant Bank Protection Material	0.341	−0.286	−0.159	0.610
CCME WQI	−0.581	−0.069	0.531	−0.268
B-IBI	0.786	−0.167	0.060	−0.062
F-IBI	−0.760	−0.164	0.350	0.073
Riverside Population Density	0.252	−0.646	−0.214	−0.376
Accessibility of Public Waterfront	−0.084	0.867	−0.185	−0.124
Shoreline Recreational Facility Richness Score	0.443	0.606	0.216	−0.355

), each component was linked to distinct ecological and functional attributes. Component 1 corresponds to the water ecology and hydrological connectivity factor. It is strongly correlated with flow connectivity, B-IBI, F-IBI, and river flow velocity. Component 2 represents the waterfront public service function factor. It is closely linked to public waterfront accessibility, river flow types, and the richness of waterfront facilities. Component 3 denotes the riparian environment and water quality factor. It is highly associated with riparian permeable surface rate and CCME WQI. Component 4 reflects the river morphology and bank protection factor. It is strongly associated with river corridor sinuosity and bank protection materials.

Combined with the indicator scores, these PCA results reveal the following points: (1) Most rivers achieve scores between 3 and 4 in flow connectivity, B-IBI, and F-IBI, which confirms generally good hydrological connectivity and sound aquatic biological health across the region. (2) Scores for riverside accessibility and leisure facility richness mostly fall between 1 and 3, highlighting widespread deficiencies in waterfront public service functions. (3) Riparian permeable surface rate and CCME WQI scores are generally in the range of 3 to 4, suggesting that the degree of shoreline hardening and overall water quality remain at a moderately favorable level. (4) River corridor sinuosity and dominant bank material scores lie between 1 and 2, indicating the need to enhance morphological diversification and reduce hardening modifications.

3.3. Cross-Scale Coupling Results of River Network–River System

Based on the weighted calculation results, the dual-scale comprehensive assessment yields an overall score of 59.39 points in the Qingxi area, belonging to the average grade. Both macro river network and micro river scale are at the medium level, with no obvious hierarchical differentiation in health grade. At the river network scale, the system provides fundamental structural support for individual rivers through its hydrological and geomorphological characteristics. It features a mature network structure, high node connectivity, a favorable water surface ratio, and a well-developed tributary hierarchy, which together maintain stable hydrological connectivity and support basic ecological functions at the river scale.

However, the ecological dimension of the river network exhibits pronounced spatial heterogeneity, with NDVI and the spread index displaying a distinct northwest-low and southeast-high pattern. This heterogeneity is strongly coupled with spatial variations in the river-scale biological integrity indices (B-IBI and F-IBI). The pattern suggests that the ecological background of the river network exerts hierarchical constraints on the ecological quality of individual rivers. At the river scale, most watercourses are classified as average grade, accompanied by a small number of relatively good and relatively poor samples. This reflects the overall moderate health background of the river network yet also reveals localized functional deficits. In particular, the relatively low evaluation performance in waterfront public service, river morphology and bank protection at the river scale is largely consistent with the insufficient green space arrangement and inadequate service provision at the river network scale. Accordingly, a certain degree of cross-scale synchronous constraint can be observed in terms of service function and morphological structure.

Overall, the Qingxi area exhibits a typical coupling pattern characterized by overall support from the macro scale and localized responses at the micro scale. The macro-level river network framework largely determines the basic functional attributes of individual rivers. Meanwhile, the micro-level shortcomings at the river scale reflect the structural and functional weaknesses of the river network itself.

The results obtained via the OAT sensitivity analysis are presented in

Table 8. Sensitivity analysis results of dual-scale evaluation weights.

Scenario No.	River Network Weight	Single River Weight	Final Comprehensive Score	Health Grade
S1	0.10	0.90	61.45	Average
S2	0.20	0.80	61.11	Average
S3	0.30	0.70	60.76	Average
S4	0.40	0.60	60.42	Average
S5	0.50	0.50	60.07	Average
S6	0.60	0.40	59.72	Average
S7 (Baseline, AHP)	0.70	0.30	59.38	Average
S8	0.80	0.20	59.03	Average
S9	0.90	0.10	58.69	Average

Note: The weight combination of S7 is the original AHP weight used in this study. All weight settings belong to the scientifically acceptable reasonable range.

. Across all reasonable weight allocation scenarios, the overall river network health score of the Qingxi area shows only minor fluctuations, ranging from 58.69 to 61.45. Notably, the health grade remains consistently classified as “Average” in all scenarios, with no grade shift observed. This confirms that the core assessment conclusion is robust and insensitive to reasonable variations in the dual-scale weights. The small range of score changes further indicates that neither scale’s weight dominates the final outcome, reinforcing that the AHP-based weighting scheme is balanced and scientifically defensible. Accordingly, the resulting assessment outcomes are reliable.

4. Discussion

4.1. Analysis and Interpretation of Results

Among the four dimensions at the river network scale, hydrology and geomorphology perform the strongest. This is supported by high node connectivity, water system integration, river network loop rate, and complexity. These indicators reflect the well-preserved natural water structure and the mature urban river network pattern in the Qingxi area. Rapid urbanization has driven substantial transformations of river networks in many highly urbanized regions worldwide [38,39,40]. The ecological dimension shows moderate overall performance but pronounced spatial differentiation. This pattern is driven mainly by habitat fragmentation in the northwestern lake-marsh zone and uneven vegetation distribution. Hydrological characteristics of urban catchments largely determine how the system responds to urban expansion and river modification [41,42]. Waterfront public service exhibits the weakest performance, characterized by low green park distribution density and limited supply of public recreational space. This weakness stems from the low-intensity urbanization pattern centered on lakes and polders, as well as insufficient integration of human-made recreational facilities into the natural water system. In fact, stronger local community attachment and sense of belonging to their living ecosystems can greatly enhance ecological improvement and restoration efforts [43].

At the river scale, the hydrological and ecological dimensions perform best. Most rivers maintain high flow connectivity, gentle flow regimes, good water quality, and rich benthic and fish communities. This can be attributed to effective sluice regulation and the relatively intact natural micro-topography in the study area. The geomorphological dimension is moderately favorable. Most rivers retain a slight to moderate curvature yet are dominated by hardened embankments. Main rivers are typically preserved to support flood control and water storage, whereas smaller tributaries are far more vulnerable to burial or infilling during urban expansion [44]. This observation aligns with existing findings that urbanization tends to exert a weaker influence on main rivers, while primarily driving the degradation and simplification of small tributaries [45]. Meanwhile, the public’s view of urban rivers tends to focus on waterfront aesthetics, recreational opportunities, and water quality rather than on purely ecological or structural features [46]. As Junker and Buchecker observed, citizens’ everyday perceptions often diverge from the ecological and environmental standards applied by experts [47]. People also develop complex, meaningful attachments to informal green spaces such as street verges, waterfront edges, vacant lots, railway corridors, and brownfields, attributing considerable value to them [48]. Among the dimensions assessed, waterfront public service performed the weakest. This was reflected in limited shoreline accessibility, uneven provision of waterfront facilities, and insufficient public services. Such shortcomings are widespread in urban river systems, where traditional assessment frameworks commonly underrepresent social service values [49]. They largely stem from riverbank encroachment by farmland and construction, poor planning of recreational infrastructure, and a general lack of human-centered design in river management.

Taken together, these coupled findings confirm that single-scale assessments alone cannot fully capture the structural and functional dynamics of urban river networks. Evaluating only the river network scale ignores localized functional defects of individual rivers. In contrast, focusing solely on river segments overlooks the macro-structural constraints of the entire water system. The dual-scale coupling assessment employed in this study reveals the hierarchical linkages between macro-scale river network patterns and micro-scale river characteristics. It clarifies cross-scale mismatches between ecological background conditions and waterfront public service supply. At the same time, it fills a critical gap in traditional river health evaluations, which typically lack multi-scale and multi-dimensional integration. Given its relatively low technical threshold and low-cost field investigation features, this method can effectively support multi-stakeholder participation and promote the popularization of scientific evaluation [50,51]. This confirms the necessity and importance of coupled river network–river assessment. It also provides a methodological reference for systematic and refined diagnosis of urban river landscapes.

4.2. Governance Strategies and Policy Recommendations

Based on the dual-scale assessment results and coupling characteristics, targeted governance strategies are proposed from three levels to promote synergistic improvement of the river network-river system. Flood frequency has been shown to increase as the fractal dimension of river networks decreases [52]. Many global cities have experienced severe declines in stream channels during urbanization, with similar patterns observed in China [45]. At the river network scale, strictly protect the existing hydrological and geomorphological structure of the Qingxi river network. Prohibit channel straightening of naturally meandering rivers such as Xuqi Jiang and Yuhui Tang to maintain stable node connectivity and water surface ratio. To address the significant spatial heterogeneity of ecological elements, prioritize improving landscape connectivity in the northwestern lake-marsh dense area. Construct ecological greenways and wetland patches to connect fragmented water bodies and reduce habitat fragmentation. Relying on the lake-marsh polder ecosystem, implement strict planning control and reduce construction intensity in ecologically sensitive areas. In addition, improve vegetation coverage along key lake shorelines such as Dianshan Lake to alleviate ecological pressure in the northeastern urbanized and tourist areas.

Classify rivers according to assessment grades for differentiated management. For the Taipu River, classified into the relatively good health grade, protective and developmental measures should be implemented. Maintain their high hydrological and biological integrity while expanding popular science and ecological recreation functions. For the 19 rivers classified as having an average health grade, targeted optimization and remediation should be implemented. Clear occupied riverbank spaces and add slow-traffic trails and waterfront platforms to improve shoreline accessibility and facility richness. Promote ecological embankment transformation by referencing the wooden pile and soil structure of Xuqi Jiang. Changes in hydrological connectivity further affect hydrological stability and ecological responses [53]. This will moderately restore natural meandering morphology, reduce the dominance of hard revetments, and enhance habitat diversity.

Establish a long-term dual-scale dynamic monitoring system to track cross-scale mismatches in waterfront public service supply and ecological background support. Strengthen inter-departmental collaborative governance among water conservancy, ecological environment, natural resources, and cultural tourism authorities, and clarify management responsibilities for sluice ecological regulation, shoreline space optimization, and waterfront public service layout. Leveraging the cultural resources of Qingxi Ancient Town, develop a public participation mechanism to encourage community involvement in river maintenance and ecological monitoring. Public support and civil society engagement are fundamental to the success of river restoration efforts [54]. This also requires a more detailed assessment of social river connectivity for rational waterfront planning [55]. This approach will foster a sustainable governance model led by the government, coordinated by multiple departments, and supported by public participation.

4.3. Limitations and Future Prospects

The proposed dual-scale assessment framework in this study exhibits good repeatability and can be applied to multi-year datasets for long-term sequential evaluation, enabling the tracking of temporal variations in river system health and identification of its driving factors. Nevertheless, constrained by data availability and research costs, the present study still has certain limitations.

This study conducts a static cross-sectional assessment without carrying out long-term dynamic analysis on the evolutionary trends of river health. Meanwhile, the current framework lacks in-depth socio-cultural indicators, as systematic acquisition of multi-source subjective survey data remains difficult. The waterfront public service dimension can be further enriched by incorporating targeted socio-cultural metrics, including public perception surveys, cultural mapping, and historical water-town interface evaluation, to support a more comprehensive assessment of river landscape cultural value.

Furthermore, the existing indicator system and scoring criteria are calibrated primarily for urban river basins and have not been validated in rural or low-disturbance natural watersheds. Indicator settings, scoring thresholds and AHP weights can be adjusted to adapt to non-urban contexts, further improving the general applicability of the methodology across broader watershed scenarios.

5. Conclusions

Based on the localized optimization of the URBAN framework, this study developed a multi-dimensional assessment system for urban water landscape health. The system includes four dimensions, including hydrology, geomorphology, ecology, and waterfront public service. By combining the macro-scale structure of river networks with the micro-scale functions of rivers, it establishes a dual-scale coupled evaluation method. This approach overcomes the limitations of traditional single-river or single-scale assessments, which often overemphasize local characteristics.

The empirical application in Shanghai’s Qingxi area confirms the system’s strong practicality and scientific validity. It effectively identifies the overall landscape health status as well as key weak points of urban water systems. With only moderate professional requirements for operators, the system can be readily promoted and applied after simple training for grassroots personnel. This enhances public participation in water system assessment and provides essential data support for building a long-term, large-scale river landscape database.

Empirical results indicate that the Qingxi river network possesses superior hydrological and geomorphological foundations. Nevertheless, obvious spatial heterogeneity of ecological quality and inadequate supply of waterfront public services are observed at the river network scale. At the river scale, hydrological and ecological performance remains at a moderately favorable level, whereas waterfront public service and river morphology constitute the primary limiting factors. An overall support and local response coupling pattern exists between the river network and individual rivers. This pattern highlights the hierarchical constraints imposed by macro-scale water system structures on micro-scale river functions. It also reveals the feedback effects of local deficiencies on the overall system. This study strengthens the scientific rigor, sustainability, and long-term effectiveness of watershed governance. It also offers a valuable reference that supports the synergistic advancement of ecological functions and human well-being in densely urbanized plain areas with well-developed river networks.