Next Article in Journal
Spatiotemporal Evolution of Water Quality and Pollution Source Identification in Baiyangdian Lake: Focus on the Extreme Precipitation Event
Previous Article in Journal
Study on the Allowable Gradient of Soil at the Base of a Cutoff Wall Considering Stress State
Previous Article in Special Issue
Assessing Three Decades of Groundwater Modelling Applications in Greece: An Overview of Progress, Tools and Future Challenges
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Water
Volume 18
Issue 12
10.3390/w18121421
water-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Landscape-Derived Indicators of Water-Related Ecological Risks: Multi-Scale Drivers and Zoned Governance in Yangtze River Basin Urban Agglomerations

by
Jing Tao
1,
Tianli Ma
2 and
Huajun Meng
1,*
1
College of Information, Shanghai Ocean University, Shanghai 201306, China
2
College of Engineering, Shanghai Ocean University, Shanghai 201306, China
*
Author to whom correspondence should be addressed.
Water 2026, 18(12), 1421; https://doi.org/10.3390/w18121421 (registering DOI)
Submission received: 24 April 2026 / Revised: 4 June 2026 / Accepted: 7 June 2026 / Published: 10 June 2026
(This article belongs to the Special Issue Optimization–Simulation Modeling of Sustainable Water Resource, 2nd Edition)
Browse Figures
Versions Notes

Abstract

Climate change and rapid urbanization increasingly threaten water security in large river basins, yet existing assessments often fail to capture the multi-scale interactions between hydroclimatic extremes and human activities. To address this gap, we developed an integrated framework combining risk assessment, multi-method driver diagnosis (Geodetector, Multi-Scale Geographically Weighted Regression (MGWR), and Structural Equation Modeling (SEM)), and Zoned Management. Using a landscape-derived Ecological Risk Index (ERI) as a proxy indicator of runoff and non-point source potential, based on established empirical linkages between landscape metrics and hydrological processes, we applied the framework to three major urban agglomerations in the Yangtze River Basin from 2000 to 2020. Our results reveal three distinct risk mechanisms: in the Chengdu–Chongqing area (CYUA), a 165.8% increase in impervious surfaces drives altered runoff; in the Middle Reaches (MRC), the q-value of the Standardized Precipitation Index (SPI) rose from 0.017 in 2000 to 0.146 in 2020, corresponding to a 759% relative increase. Although the absolute q-value of SPI remains moderate at around 0.15, its rapid rise suggests increasing hydrological sensitivity of the MRC’s river–lake system to precipitation extremes; in the Yangtze River Delta (YRD), socioeconomic activities exert overriding pressure. Based on these diagnostics, we propose tailored strategies for water environment management, adaptive planning, and disaster mitigation. This framework offers a scientific basis for differentiated water governance in large river basins facing coupled anthropogenic and hydroclimatic pressures.

1. Introduction

The sustainable development of major river basins is fundamentally contingent upon water security. For urban agglomerations within these basins, accelerating climate change and rapid urbanization are converging to intensify a suite of interconnected water challenges. These include escalating water scarcity, pervasive water quality deterioration, degradation of aquatic ecosystems, and heightened risks of both floods and droughts [1]. These pressures collectively threaten hydrological resilience and pose a critical constraint on long-term socio-ecological sustainability. In the Yangtze River Basin, the interplay of climate variability and intensive human activities—such as large-scale reservoir operations and land use change—has significantly altered natural hydrological regimes since the late 20th century [2,3]. This has led to pronounced spatial gradients in water-related environmental risks across the basin [4,5]. The increasing frequency and intensity of hydroclimatic extremes, superimposed on anthropogenic pressures, not only directly impair water quality [6] and degrade critical aquatic habitats [7] but also systematically undermine the basin’s capacity to buffer hydrological shocks, thereby complicating integrated water resource management and disaster risk reduction [8].
To assess environmental risks, landscape pattern indices have been widely used to construct composite indices like the Ecological Risk Index (ERI) [9]. Recent studies have confirmed that landscape composition and spatial configuration strongly regulate runoff generation, nutrient transport, and non-point source pollution across watersheds. Qi et al. (2026) integrated landscape source–transport–sink mechanisms with GeoAI modeling to improve the surrogate prediction of watershed nutrient loads, verifying that landscape spatial structure acts as a fundamental constraint on nutrient migration and accumulation [10]. Matomela et al. (2023) further evidenced that land use transition and landscape pattern alterations significantly reshape spatial runoff distribution and non-point source pollution dynamics, with landscape metrics closely tied to basin pollutant loading [11]. Yin et al. (2025) quantified how landscape pattern changes control runoff variations at annual and seasonal scales in ecologically fragile basins, highlighting the deterministic role of landscape structure in regulating hydrological processes [12]. Building on these consistent findings, the landscape-based ERI, synthesized from fragmentation, isolation, and dominance metrics, can reasonably serve as a surrogate indicator for water-related ecological risk, reflecting comprehensive landscape-induced pressures on hydrological routing, nutrient retention, and non-point pollution potential.
Despite these advances, significant gaps remain in the current understanding of water-related ecological risks. First, many assessments lack fine-scale, dynamic quantification of hydroclimatic extremes—such as floods and droughts characterized by indices like the Standardized Precipitation Index (SPI)—as discrete, potent drivers of risk [13]. While long-term trends are often considered, the spatially heterogeneous ecological impacts triggered by short-term extreme wet and dry anomalies are frequently overlooked [14]. Second, there is a need for a more explicit integration of the multi-process interactions between landscape change and core water cycle components. Specifically, the pathways through which altered land use affects runoff generation, soil erosion, and non-point source pollution transport are not fully articulated within many risk assessment frameworks [15,16]. Third, and crucially for management, existing studies often fail to dissect the multi-scale spatial heterogeneity of driving factors [17]. While tools like the Geodetector can identify dominant drivers and their interactions, they typically provide global statistics that mask local variations. Subsequent analyses using Geographically Weighted Regression (GWR) often impose a single, uniform spatial scale on all drivers, which is a geographic oversimplification. This methodological shortfall hinders the identification of whether and how the influence of a factor like SPI or economic density varies from one location to another. Furthermore, understanding the complex causal pathways—including indirect and mediating effects—through which natural and anthropogenic factors jointly influence water risk requires analytical techniques like Structural Equation Modeling (SEM), which are not yet routinely integrated into such spatial diagnostic frameworks.
Therefore, to bridge these gaps, this study develops and applies an integrated analytical framework: Water-Related Risk Assessment–Multi-Scale Driver Diagnosis–Zoned Management. This framework innovatively combines Geodetector, Multi-Scale Geographically Weighted Regression (MGWR)—which allows each driver to operate at its own optimal spatial scale [18]—and SEM to deconstruct the mechanisms shaping water-related ecological risks. We focus on three major urban agglomerations in the Yangtze River Basin from 2000 to 2020 [19]. The study is guided by four specific research questions: (1) What are the spatiotemporal patterns of water-related ecological risk? (2) What are the key driving factors and their interaction effects, and, critically, how does their influence vary spatially across scales? (3) How do natural and anthropogenic factors interconnect through direct and indirect pathways to form a causal network governing risk? (4) Based on this mechanistic diagnosis, what are the region-specific, targeted strategies for water environment management and disaster risk mitigation?
The principal contributions of this work are fourfold. First, it provides a targeted methodological integration for water-related ecological risk analysis, establishing a progressive and coherent workflow coupling Geodetector, MGWR, and SEM. Unlike simple method superposition, this integrated framework can deliver substantive and exclusive insights: it quantifies the scale hierarchy of multi-type driving factors, identifies prominent inter-regional divergence in vegetation’s hydrological regulatory capacity, and supports scale-matched, zoned risk governance, effectively overcoming the one-sided limitations of single-model analysis. Second, it explicitly quantifies the role of hydroclimatic extremes (SPI) and reveals their strong contextual dependence [20]. Third, it identifies distinct dominant risk mechanisms for the Chengdu–Chongqing (CYUA), Middle Reaches (MRC), and Yangtze River Delta (YRD) urban agglomerations, linking drivers directly to water processes (e.g., GDP to pollutant loads, nighttime lights to impervious surfaces and urban runoff, SPI to flood–drought hazards) [21]. Finally, it translates these diagnostics into actionable, zoned governance strategies for water security [22], offering a replicable framework for large river basins globally.

2. Materials and Methods

2.1. Study Area

This study focuses on three major urban agglomerations within the Yangtze River Basin: CYUA, MRC, and YRD (Figure 1). These regions are the core socioeconomic engines of the basin, representing hotspots where intense human activities intersect with sensitive aquatic ecosystems [23,24]. From a water security perspective, the three agglomerations form a compelling gradient study system: CYUA acts as the source disturbance zone (upper mountainous reaches, terrain-constrained urbanization, soil erosion and non-point source risks) [16]; MRC serves as the process modulation zone (Middle Reaches, largest freshwater lake cluster, flood regulation but stressed by hydrological variability) [1,25]; YRD represents the sink pressure zone (river mouth, low-lying delta, extreme socioeconomic intensity, coastal–estuarine pollution and flood risks) [26,27]. This upstream-to-downstream gradient encapsulates distinct yet interconnected water risk typologies [28], providing a holistic basin-scale framework for zoned water governance [17,29].

2.2. Data

All data are categorized into three types: land use/cover, natural factors, and socioeconomic factors. The China Land Cover Dataset (CLCD) [30] provides annual land cover (2000–2020). Natural factors include a 30 m DEM (Chinese Academy of Sciences), annual precipitation and temperature (RESDC), annual maximum NDVI (MODIS MOD13A3), and the 6-month Standardized Precipitation Index (SPI-6) [31]. Monthly SPI-6 values were averaged to obtain annual indicators, and data for 2000, 2005, 2010, 2015 and 2020 were extracted to match the 5-year interval of land use data. Socioeconomic data include gridded GDP, population density (LandScan) [32], nighttime light intensity [33], and per capita GDP (derived). All raw data were uniformly preprocessed (format conversion, clipping, resampling, algebraic operations) using ArcGIS 10.2 and Python 3.8.5 to ensure consistent spatiotemporal range and resolution (Albers equal-area conic projection). Table 1 summarizes the data sources, resolution, temporal coverage, and primary use.

2.3. Model and Methodology

This study constructed a systematic analytical framework to uncover the multi-scale driving mechanisms behind water-related ecological risk. This framework consists of three sequential components: landscape ecological risk assessment, multi-method driver diagnosis, and zoning governance. For driver diagnosis, we adopted a three-stage workflow integrating Geodetector, MGWR and SEM.
We used Geodetector to quantify factor explanatory power and identify interaction patterns. All factors were retained given their evident spatiotemporal heterogeneity and widespread synergistic effects between variables (see Figure A1). Factors with steady high explanatory power were selected as core predictors, and factor interaction features guided covariance settings in SEM. We then applied MGWR to acquire local coefficients and optimal bandwidths, which reflected multi-scale spatial disparities of driving forces and supported the definition of SEM causal paths. Finally, we established and optimized the SEM model by removing insignificant paths (p > 0.01) and adding reasonable covariances. We emphasize that the SEM analysis in this study is essentially exploratory. The final path model is obtained through iterative, data-driven optimization, rather than strictly predefined theoretical assumptions. Accordingly, the causal interpretations should be regarded as hypotheses to be verified by independent datasets in future research.
We further examined model robustness via a sub-period test on core SEM pathways, with results presented in Table A3. The temporal stability of key paths verifies model reliability. Formal split-sample cross-validation was not conducted due to medium sample size, and this limitation is discussed in Section 4.4.

2.3.1. Landscape Ecological Risk Index Assessment

We used ERI as a proxy for key water-related processes, namely runoff generation and non-point source pollution potential. To quantitatively characterize the macroscopic spatial pattern of ecological risk across three large urban agglomerations, a regular fishnet grid with a cell size of 30 km × 30 km was created as the basic assessment unit. This scale was selected by comprehensively considering research objectives, data features and computational efficiency. It is suitable for regional comparative analysis, reduces the computational load of long time-series data, and matches the aggregation of 1 km land use datasets. This grid setting is adopted in large-scale landscape risk and hydrological function assessments for river basins and urban agglomerations [34,35]. Nevertheless, a fixed grid inevitably leads to the Modifiable Areal Unit Problem (MAUP), which may mask fine-scale ecological differences in complex terrain and highly urbanized areas. We further conducted multi-scale sensitivity tests using 15 km, 30 km and 45 km grids to verify result reliability, and relevant details are provided in Table A4. The ERI for each unit was then calculated as follows:
E R I = ∑ i = 1 n ( A k i A k ) × R i
where E R I is the landscape Ecological Risk Index for the kth assessment unit; n is the number of landscape types; A k i is the area of landscape type i within the kth assessment unit; A k is the total area of the kth assessment unit; and R i is the landscape loss index for landscape type.
The landscape loss index R i is calculated by multiplying the landscape disturbance index E i and the landscape vulnerability index F i :
R i = E i × F i
The landscape disturbance index E i integrates three indicators—fragmentation, isolation, and dominance—with calculation formulas and weight assignments following established research [35]:
E i = a C i + b S i + c D i
where weights a , b , and c are set to 0.5, 0.3, and 0.2, respectively, with a + b + c = 1. The detailed expressions for C i , S i , and D i as well as the vulnerability values assigned to nine land use types are provided in Table A1. Finally, all computed ERI values were classified into five risk levels (from lowest to highest) using the natural breaks method. It should be noted that the ERI is a landscape-based proxy for water-related ecological risk; direct hydrological validation is a direction for future research (see Section 4.4).

2.3.2. Geodetector

We employed the factor detection and interaction detection functions of Geodetector to quantify the explanatory power of each driving factor on the spatial differentiation of ERI. The q statistic, which measures this explanatory power, is calculated as
q = 1 − ∑ h = 1 L N h σ h 2 N σ 2
where L is the number of classes or strata of driving factor X; N h and N are the numbers of sample units in stratum h and the entire region, respectively; and σ h 2 and σ 2 are the variances of the ERI in stratum h and the entire region, respectively.
Nine driving factors were selected—elevation (X1), GDP (X2), annual precipitation (X3), NDVI (X4), average annual temperature (X5), population density (X6), nighttime light intensity (X7), GDP per capita (X8), and SPI (X9)—with ERI as the dependent variable (Y). Geodetector analysis was conducted after resampling and preprocessing data from five temporal periods.

2.3.3. Multi-Scale Geographically Weighted Regression Analysis

Before applying local regression, we first assessed the spatial clustering pattern of ERI using the Global Moran’s I index. The computed Moran’s I values ranged from 0.736 to 0.784 across all years (p < 0.01), indicating significant positive spatial autocorrelation. This result justified the use of a local modeling approach. We then turned to MGWR, which overcomes a key limitation of traditional GWR by allowing each independent variable to have its own optimized bandwidth. The MGWR model is specified as
y i = ∑ j = 1 k β b w j ( u i , v i ) x i j + ϵ i
where y i is the ERI value at location i ; x i j is the value of the j -th independent variable at location i ; ( u i , v i ) represents the coordinates of location i ; β b w j ( u i , v i ) is the spatially varying regression coefficient for the j -th variable with bandwidth b w j , indicating that each variable has its own independent spatial smoothing parameter; and εi is the error term. The MGWR tool in ArcGIS Pro was employed, with bandwidth selection optimized independently for each variable based on the Akaike Information Criterion corrected (AICc) to reflect the inherent spatial scales of different driving processes.

2.3.4. Structural Equation Modeling Analysis

To disentangle the complex interaction networks among driving factors and to quantify both direct and indirect pathways influencing ERI, we adopted exploratory Structural Equation Modeling analysis. Instead of relying on generic ecological theory, we constructed initial conceptual models for CYUA, MRC, and YRD based on the results obtained from Geodetector and MGWR. These initial models were then iteratively refined: non-significant pathways (p > 0.01) were removed step by step, and necessary covariances among error terms were added according to modification indices and theoretical plausibility. Model parameters were estimated using the maximum likelihood method. The following fit indices were used to evaluate model performance: a chi-square-to-degrees-of-freedom ratio (Chi/DF) below 5 was considered acceptable and below 3 indicated a good fit; GFI, AGFI, and CFI values above 0.90 were acceptable and above 0.95 indicated a good fit; an RMSEA below 0.08 was reasonable and below 0.05 indicated a good fit.

3. Results

3.1. Spatiotemporal Evolution of Water-Related Ecological Risk

Before examining ecological risk dynamics, we briefly summarize land use changes most relevant to water processes. From 2000 to 2020, impervious surfaces expanded by 108.3% across the three agglomerations, with the highest growth rate in CYUA (+165.8%), followed by YRD (+107.5%) and MRC (+93.0%). Water bodies shrank notably only in MRC (−7.4%), whereas cropland decreased most severely in YRD (−9.3%).
Higher risk areas exhibited a distinct “point-axis” agglomeration pattern along the Yangtze mainstem and major transportation corridors [36] (Figure 2). Between 2000 and 2020, the extent of the highest-risk area shrank substantially: by 77.5% in YRD, by 44.0% in CYUA, and by 25.0% in MRC. Meanwhile, the lowest-risk area expanded in all three agglomerations (by 13.8% in YRD, 9.4% in CYUA, and 5.3% in MRC). Despite these improvements, moderate-risk areas still accounted for a non-negligible proportion (e.g., 23.0% in CYUA in 2020), indicating that continued risk management efforts remain necessary.

3.2. Geodetector Analysis of Landscape Ecological Risk Drivers

We used Geodetector to quantify the explanatory power, denoted as the q-value, of each natural and socioeconomic driver for the spatial differentiation of water-related ecological risk. The analysis reveals a dual structure consisting of a stable baseline layer and a fluctuating modulation layer.

3.2.1. Factor Detection

The explanatory power of the SPI exhibited strong spatiotemporal heterogeneity across the three agglomerations (Figure 3). In CYUA, the q-value of SPI increased from 0.031 in 2000 to 0.142 in 2020, a rise of 358%. This trend indicates growing sensitivity to precipitation variability. A notable increase in SPI explanatory power was observed in MRC: its q-value rose from 0.017 in 2000 to 0.146 in 2020, a relative increase of 759%. This substantial percentage change is mainly attributed to the extremely low initial baseline. Hydrologically, although the 2020 q-value only explains around 15% of ERI spatial differentiation and remains moderate in absolute terms, this growing trend indicates that the lake–wetland and floodplain system of the MRC has become far more sensitive to hydroclimatic anomalies. Under climate change and wetland functional degradation, SPI has gradually developed into an important triggering factor for flood–drought ecological risks. In YRD, the q-value of SPI fluctuated over time, declining from 0.112 in 2000 to 0.004 in 2015 before rebounding to 0.083 in 2020. This pattern reflects the buffering capacity of the region’s developed economy and flood control systems [17,18], though the volatility also reveals the complexity of modulating climatic effects under strong socioeconomic backgrounds [37,38].

3.2.2. Overall Structure

Across the entire basin, socioeconomic factors such as population density (X6) and GDP (X2) form a stable baseline layer. In water terms, they represent persistent pressure on water pollution load and water demand. Climatic factors, particularly annual precipitation (X3) and SPI (X9), constitute a highly fluctuating modulation layer. Their inter-annual variability can substantially amplify or mitigate the intensity of water-related risk during specific periods. Importantly, the effect of SPI is not uniform across the basin; it is selectively amplified or dampened by local natural backgrounds and the resistance capacity of socioeconomic systems (Figure 4).

3.3. Spatial Heterogeneity of Water-Related Natural Drivers (MGWR)

Global Moran’s I for ERI ranged from 0.736 to 0.784 (p < 0.01), confirming significant positive spatial autocorrelation and justifying the use of MGWR. The MGWR results revealed that natural factors operate at distinct spatial scales across the three agglomerations (Figure 5).
In CYUA, elevation acted as a rigid topographic constraint with a strongly negative effect (coefficient up to −1.77). SPI showed a positive correlation with risk in rapidly expanding urban fringes, indicating an emerging superposition of hydroclimatic pressure onto urbanization. NDVI generally had negative but limited effects on risk mitigation.
In MRC, hydrothermal conditions exhibited a functional reversal. In intensive agricultural areas such as the Jianghan Plain and Dongting Lake Plain, annual precipitation turned into a pressure factor (positive coefficient up to 0.66), suggesting that excessive water can induce waterlogging and non-point source pollution. This reversal coincided with strong positive effects of SPI (coefficient range 0.31–0.67), making this region highly sensitive to extreme hydrological events. NDVI showed negative effects but with limited spatial continuity.
In YRD, the influence of natural factors was strongly overprinted by human activities. Elevation effects approached zero. The signals of annual precipitation and temperature were significantly attenuated, masked by urban hydrology and the urban heat island effect. SPI effects were weak and spatially patchy, indicating that infrastructure buffers regular precipitation fluctuations, though latent vulnerability to prolonged extreme drought persists. NDVI negative effects were confined to scattered green patches.
Spatial scales from MGWR bandwidths: MGWR optimal bandwidths reveal a clear hierarchy (Table A2). Climate drivers (precipitation, temperature, SPI) have the largest bandwidths (1450–1680 km), indicating basin-scale processes requiring cross-provincial coordination. Population density shows moderate bandwidths (~720 km), matching metropolitan extents. Socioeconomic drivers (GDP per capita, nighttime light) exhibit the smallest bandwidths (260–310 km), reflecting localized urban core impacts. This bandwidth hierarchy directly informs the scale-sensitive governance strategies in Section 4.2.

3.4. Regional Differentiation of Water-Related Causal Pathways: SEM Analysis

Structural Equation Modeling was used to identify key natural and anthropogenic pathways affecting the ERI. All models showed good fit (Table 2). Core significant pathways for CYUA, MRC and YRD are displayed in Figure 6, Figure 7, and Figure 8, respectively.

3.4.1. Common Basin-Wide Patterns

Elevation correlates negatively with temperature and ERI, indicating higher risk in low-lying plains. Annual precipitation determines the SPI, while GDP drives population density and nighttime light intensity.

3.4.2. Three Distinct Regimes

In CYUA, human activities reduce NDVI and indirectly elevate water risk, with a path coefficient of minus 0.30 from NDVI to the risk index. Temperature directly increases risk by 0.33, whereas SPI shows no direct effect. Elevation exerts a strong negative constraint of minus 0.63. This regime can be described as terrain-constrained, where topography dominates the spatial pattern of water risk (Figure 6).
The MRC exhibits a transitional climate–human coupled regime. The NDVI pathway remains significant at minus 0.25. SPI directly increases risk by 0.12, but, interestingly, annual precipitation reduces risk by minus 0.66, likely because moderate precipitation sustains wetland buffering. Elevation again suppresses risk, with a coefficient of minus 0.43. This region exhibits a climate-triggered risk regime where hydroclimatic extremes act as the primary risk amplifier (Figure 7).
The YRD represents a fundamentally engineered regime. Here, the mediating effect of NDVI becomes statistically insignificant in the causal network. SPI shows a significant positive effect of 0.19, revealing a latent vulnerability to hydroclimatic extremes that engineered systems cannot fully eliminate. Elevation has the strongest negative effect among all three regions, reaching minus 0.98, meaning that even slight elevation differences become critical in this low-lying delta. The non-significant NDVI-mediated pathway mentioned above does not indicate a complete collapse of natural hydrological regulation. This phenomenon may be partially attributed to high vegetation homogeneity in highly urbanized delta areas, potential multicollinearity among socioeconomic variables, and the dominant role of artificial hydraulic infrastructure. It implies that natural vegetation mediation is gradually overshadowed by engineered systems and human disturbances, rather than marking an abrupt regime shift (Figure 8).

4. Discussion

4.1. Water Risk Mechanisms in Three Contrasting Hydro-Social Contexts

The three urban agglomerations exhibit fundamentally different water risk regimes, each shaped by a distinct interplay between natural settings, development stages, and human–climate interactions.
For CYUA, rapid urbanization unfolds under complex terrain conditions. The dramatic spread of impervious surfaces fundamentally alters how rainfall turns into runoff, leading to higher flood peaks and more severe non-point source pollution from urban fringes and agricultural slopes. Water risk here is therefore constrained by topography and amplified by land use change. In MRC, by contrast, hydroclimatic extremes have become the primary risk driver. Short-term precipitation excess easily triggers waterlogging, nutrient flushing, and inundation because natural floodplain storage and wetland filtration have been partially degraded. This region operates under a climate-triggered regime where natural buffering is only partial. Finally, YRD represents a different reality. High-intensity socioeconomic activities largely override natural climate variability. Engineered infrastructure buffers routine hydrological fluctuations, but this also creates a regime of socioeconomic endogenous coupling, where water environmental capacity and estuarine ecological stress are dominated by GDP growth and impervious surface expansion. Latent vulnerabilities to extreme events remain, even if average conditions appear well managed.
Across the three urban agglomerations, similarities include significant spatial clustering of water ecological risk along the Yangtze River corridor and persistent socioeconomic pressure from population and economic activities as basin-wide baseline drivers.
Regionally distinct regimes are clear: CYUA is constrained by terrain and dominated by impervious surface expansion altering runoff pathways; MRC exhibits rising sensitivity to hydroclimatic extremes, with SPI becoming a key triggering factor governing flood–drought risk evolution; YRD is overwhelmingly shaped by intensive urbanization and engineered infrastructure, where vegetation’s mediating role is substantially weakened. This consistent background yet regionally differentiated mechanism underpins the necessity of zoned water governance.
In summary, the three agglomerations form a coherent gradient from terrain-constrained urbanization impacts (CYUA) to climate-triggered floodplain vulnerability (MRC) to socioeconomic-dominated deltaic stress (YRD). Each demands a distinct water management strategy.

4.2. Scale-Sensitive Policy Implications

MGWR coefficient patterns (Figure 5) reveal three distinct operational scales for water risk drivers.
Basin-wide (SPI): SPI coefficients are uniformly distributed across the whole basin, indicating that hydroclimatic extremes require coordinated, cross-provincial drought–flood management rather than local actions.
Metropolitan (population density): Population density shows moderate spatial heterogeneity, aligning with metropolitan extents. Interventions should target county- or city-level water demand control and pollution reduction.
Local (GDP): GDP varies sharply, concentrating in urban cores. Effective measures include sponge city construction, green infrastructure, and industrial water recycling at hotspot scales.
Consistent with the MGWR bandwidth hierarchy, our scale-sensitive policy recommendations are organized as follows: (1) basin-wide actions for climate drivers (SPI, precipitation, temperature), such as cross-provincial drought–flood early warning systems; (2) metropolitan-scale actions for population density, including water demand management and wastewater treatment infrastructure; and (3) local-scale actions for socioeconomic drivers (GDP, nighttime light), such as sponge city construction, green infrastructure, and industrial water recycling at hotspot locations.
SEM integration further refines these strategies: protect green space in CYUA to break the density–NDVI–risk chain; establish SPI-based early warning in MRC; and adopt blue–green–gray infrastructure in YRD to decouple growth from hydrological degradation.
These spatial patterns provide a multi-scale, evidence-based foundation for zoned water governance.

4.3. Zone-Based Water Management Strategies

Based on the diagnosed risk mechanisms, we propose three zone-specific water management strategies with actionable interventions, as illustrated in Figure 9. For the upper Yangtze ecological barrier zone (CYUA), we prioritize green infrastructure to manage terrain-constrained runoff and erosion, including riparian buffers, sloping land restoration, and conservation tillage. For the river–lake connected floodplain zone (MRC), we focus on climate-adaptive measures such as seasonal floodplain reconnection, coordinated reservoir–lake operations, and SPI-based early warning systems. For the high-intensity development compound risk zone (YRD), we mandate engineered solutions including tiered transboundary compensation for water quality, algal bloom management, saltwater intrusion early warning, and sponge city requirements (e.g., at least 40% permeable surface in new districts). Together, these strategies form a tiered, spatially explicit framework for water management in the Yangtze River Basin.
To further enhance the scientific support and policy operability of the above qualitative zoned governance strategies, this study further supplements a scientifically credible quantitative prediction of ecological risk mitigation effects. Referring to the empirical governance effect and landscape ecological risk reduction results of similar ecological restoration, green infrastructure construction, reservoir–lake joint regulation and sponge city practices in the Yangtze River Basin [26,39,40,41,42,43,44], combined with long-term field monitoring data of typical watersheds and urban agglomerations in CYUA, MRC and YRD [17,40,45,46], as well as the marginal driving contribution constraints of dominant natural and socioeconomic factors derived from Geodetector and MGWR in this study, the prospective ERI reduction intervals of tailored strategies are determined as follows: the targeted green infrastructure and ecological restoration measures are expected to reduce the average ERI by 8–15% in CYUA [40,45]; climate-adaptive early warning and river–lake coordinated regulation can achieve an ERI decrease of 10–18% in MRC [39,46]; and integrated blue–green–gray infrastructure together with sponge city implementation may lower the comprehensive water ecological risk by 6–12% in YRD [41,42,43,47]. Considering the inherent uncertainty of coupled human–water systems, this study adopts interval estimation instead of single-value prediction to avoid arbitrary overestimation [48,49,50].

4.4. Limitations and Future Directions

This study has four main limitations. (1) The landscape-based ERI acts as an indirect indicator of water-related ecological risk. Owing to insufficient long-term spatially continuous hydrological data across the Yangtze River Basin, ERI was not validated with field-measured hydrological variables. (2) Restricted by sample size, we used a sub-period internal consistency test for SEM instead of standard split-sample cross-validation, which may introduce potential sampling bias. (3) This study is subject to the Modifiable Areal Unit Problem (MAUP) resulting from the fixed 30 km × 30 km evaluation grid. Although supplementary multi-scale sensitivity analysis (15 km, 30 km, and 45 km) verified the robustness of the core spatial patterns and driving mechanisms, this single grid size may smooth fine ecological heterogeneity in highly urbanized delta areas and subtle terrain gradients in mountainous urban agglomerations, and may produce heteroscedastic averaged values in complex topographic regions. (4) This study adopted 5-year interval sampling and annual mean SPI-6 values to characterize climatic dry–wet conditions, which inevitably smooths intra-annual and short-term inter-annual precipitation variability and limits the capacity to capture abrupt hydroclimatic extreme events.
For future research, we will integrate field monitoring data and hydrological models (e.g., SWAT) to improve the accuracy and reliability of landscape-based risk assessment. Larger sample datasets will be adopted to implement rigorous split-sample cross-validation and strengthen the credibility of SEM causal inference. We will also employ continuous annual SPI records and combine annual maximum and minimum SPI-6 values to distinguish the influences of long-term dry–wet background and hydroclimatic extremes. Furthermore, multi-scale grid comparison or combined grid–administrative unit analysis is recommended to mitigate MAUP effects and achieve more robust ecological risk evaluation. In addition, although this study has supplemented interval estimation of ERI reduction benefits for zoned governance strategies to compensate for the lack of quantitative evaluation, future work can further introduce scenario simulation models. Such tools can support dynamic, refined, and multi-intensity quantitative prediction of ecological risk mitigation effects under diverse policy implementation scenarios, thereby providing more precise and adaptive decision-making references for watershed fine-scale ecological management.

5. Conclusions

From the integrated analysis of water-related ecological risk in three Yangtze River Basin urban agglomerations over the period 2000–2020, we draw four main conclusions.
First, three fundamentally different water risk regimes have been identified. The Chengdu–Chongqing area exhibits a land use-dominated regime where rapid impervious expansion alters runoff and erosion. The Middle Reaches operate under a climate-driven regime where hydroclimatic extremes, particularly SPI, act as the primary trigger of flood–drought hazards. The Yangtze River Delta represents a socioeconomically dominated regime where intensive human activities override natural hydrological regulation.
Second, multi-scale driver effects and causal pathways have been clarified through MGWR and SEM. Natural factors such as SPI operate at near-basin scales, whereas socioeconomic drivers like GDP and nighttime light demand localized interventions. More critically, vegetation (NDVI) acts as a significant mediator of water ecological risk in CYUA and MRC, while its mediating function becomes statistically insignificant in YRD. This phenomenon implies diminished natural hydrological regulation under intensive urbanization.
Third, the MGWR-derived bandwidths provide direct guidance for scale-sensitive policy design. Basin-wide coordination is required for SPI driven flood–drought risks, metropolitan-level actions should target water demand and pollution control, and local interventions need to focus on green infrastructure and sponge city construction in high-intensity urban cores.
Fourth, based on the diagnosed mechanisms, we propose three zone-specific management strategies. For the upper Yangtze ecological barrier zone (CYUA), priority actions include riparian buffers, sloping land restoration, and conservation tillage. For the river–lake connected floodplain zone (MRC), climate adaptive measures such as seasonal floodplain reconnection, coordinated reservoir lake operations, and SPI-based early warning are recommended. For the high-intensity development compound risk zone (YRD), engineered solutions like tiered transboundary compensation, algal bloom management, saltwater intrusion early warning, and sponge city mandates should be implemented.
Collectively, this progressive multi-model framework moves beyond simple methodological superposition to generate substantive mechanistic insights, including the quantified scale hierarchy of diverse driving factors and the divergent regional disparities in vegetation’s hydrological regulatory performance. It provides a scientific basis for scale-matched and zoned differentiated water management in the Yangtze River Basin and offers transferable insights for other large river basins facing coupled pressures of urbanization and hydroclimatic extremes.

Author Contributions

J.T.: Conceptualization, Methodology, Software, Writing—original draft; T.M.: Methodology, Data curation; H.M.: Writing—review and editing. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data Availability Statement

Data will be made available on request.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A

Table A1. Detailed formulas for landscape metrics and vulnerability values used in ERI calculation.
Table A1. Detailed formulas for landscape metrics and vulnerability values used in ERI calculation.
ComponentFormula/ValueDescription
Fragmentation ( C i ) C i = n i A i n i is the number of patches in landscape i and A i is the total area of the landscape i
Isolation ( S i ) S i = A 2 A i n i A A is the total landscape area
Dominance ( D i ) D i = 1 4 ( n i M + m i M ) + A i 2 A M is the total number of sampling points and m i is the number of sampling points containing landscape
Vulnerability values ( F i )
Snow/ice6Highest vulnerability
Water bodies5
Wetlands5
Cropland4
Shrubland3
Grassland3
Bare land3
Forest2
Impervious surface1Lowest vulnerability
Table A2. Representative MGWR optimal bandwidths for major drivers.
Table A2. Representative MGWR optimal bandwidths for major drivers.
VariableRepresentative Bandwidth (km)Spatial Scale Interpretation
Annual precipitation1520Basin-wide
Annual average
temperature		1480Basin-wide
SPI1450–1680Basin-wide
Population density720Metropolitan/city cluster scale
GDP280Urban core/local hotspot
GDP per capita260Urban core/local hotspot
Nighttime light
intensity		310Urban core/local hotspot
Table A3. Consistency of key SEM pathways across two sub-periods.
Table A3. Consistency of key SEM pathways across two sub-periods.
Study AreaKey Causal Pathway2000–2010
Coefficient (p-Value)		2010–2020
Coefficient (p-Value)		Consistency Characteristic
CYUAElevation → ERI−0.617 (p < 0.001)−0.628 (p < 0.001)Stable negative effect
NDVI → ERI−0.304 (p < 0.001)−0.298 (p < 0.001)Stable negative effect
Annual precipitation → ERI−0.425 (p < 0.001)−0.438 (p < 0.001)Stable negative effect
MRCAnnual precipitation → ERI−0.648 (p < 0.001)−0.661 (p < 0.001)Stable negative effect
SPI → ERI0.102 (p < 0.001)0.127 (p < 0.001)Stable positive effect
NDVI → ERI−0.249 (p < 0.001)−0.253 (p < 0.001)Stable negative effect
Elevation → ERI−0.431 (p < 0.001)−0.439 (p < 0.001)Stable negative effect
YRDElevation → ERI−0.969 (p < 0.001)−0.982 (p < 0.001)Stable negative effect
Annual precipitation → ERI−0.672 (p < 0.001)−0.683 (p < 0.001)Stable negative effect
GDP → Population density0.447 (p < 0.001)0.461 (p < 0.001)Stable positive effect
Table A4. Sensitivity analysis of research results under different grid sizes.
Table A4. Sensitivity analysis of research results under different grid sizes.
Evaluation Index15 km × 15 km30 km × 30 km45 km × 45 kmConsistency
Moran’s I of ERI0.7510.7360.724Differences ≤ 0.012
Top three drivers (Geodetector q-value, 2020)NDVI (0.212), population density (0.194), annual precipitation (0.144)NDVI (0.204), population density (0.190), annual precipitation (0.141)NDVI (0.198), population density (0.185), annual precipitation (0.137)Ranking unchanged
SPI q-value in MRC (2020)0.1530.1460.140Variation ≤ 5.5%
SEM path: SPI → ERI (MRC)0.1280.1240.118Sign and significance unchanged
SEM path: NDVI → ERI (CYUA)−0.315−0.303−0.294Sign and significance unchanged
SEM path: Elevation → ERI (YRD)−0.986−0.979−0.968Sign and significance unchanged
Water 18 01421 g0a1
Figure A1. Single-factor q-value and factor interaction heatmaps. (a) CYUA; (b) MRC; (c) YRD; (d) the entire Yangtze River Basin.
Figure A1. Single-factor q-value and factor interaction heatmaps. (a) CYUA; (b) MRC; (c) YRD; (d) the entire Yangtze River Basin.
Water 18 01421 g0a1

References

  1. Gong, L.; Zhang, X.; Liu, J.; Gui, H. Exploring the Influence of Urban Agglomeration on Extreme Precipitation: Evidence from the Middle Reaches of the Yangtze River, China. J. Hydrol. Reg. Stud. 2024, 55, 101932. [Google Scholar] [CrossRef]
  2. Tan, Z.; Li, H.; Song, Q.; Wang, Z.; Cao, Y. Synergistic Optimization and Interaction Evaluation of Water-Energy-Food-Ecology Nexus Under Uncertainty from the Perspective of Urban Agglomeration. Sustain. Cities Soc. 2025, 124, 106291. [Google Scholar] [CrossRef]
  3. He, N.; Guo, W.; Lan, J.; Yu, Z.; Wang, H. The Impact of Human Activities and Climate Change on the Eco-Hydrological Processes in the Yangtze River Basin. J. Hydrol. Reg. Stud. 2024, 53, 101753. [Google Scholar] [CrossRef]
  4. Wang, Z.; Ma, C.; Tan, Z.; Wu, T. Low-Carbon Development Pathways for the Water-Energy-Food-Carbon Nexus in the Yangtze River Economic Belt: Insights from Coupling Coordination and Obstacle Degree Analysis. J. Clean. Prod. 2025, 523, 146399. [Google Scholar] [CrossRef]
  5. Bai, J.; Guo, K.; Liu, M.; Jiang, T. Spatial Variability, Evolution, and Agglomeration of Eco-Environmental Risks in the Yangtze River Economic Belt, China. Ecol. Indic. 2023, 152, 110375. [Google Scholar] [CrossRef]
  6. Zhu, Z.; Wu, H.; Wang, Z.; Zhang, X.; Kong, J.; Liu, C.; Tan, Z.; Liu, Q. Physics-AI Synergized Optimization-Learning-Simulation Framework for Robust Cascade Reservoir Scheduling Under Future Hydrological Uncertainty. Water Resour. Res. 2026, 62, e2025WR042149. [Google Scholar] [CrossRef]
  7. Li, Z.; Peng, L.; Wu, F. The Impacts of Impervious Surface on Water Quality in the Urban Agglomerations of Middle and Lower Reaches of the Yangtze River Economic Belt From Remotely Sensed Data. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2021, 14, 8398–8406. [Google Scholar] [CrossRef]
  8. Ding, C.; Wang, Z.; Tan, Z.; Zhang, X. Multi-source Heterogeneous Data-Driven Interpretable Model Based on Transformer and Kolmogorov-Arnold Networks for Streamflow Prediction. Expert Syst. Appl. 2026, 328, 132862. [Google Scholar] [CrossRef]
  9. Jia, L.; Yu, K.; Li, Z.; Li, P.; Cong, P.; Li, B. Spatiotemporal Pattern of Landscape Ecological Risk in the Yangtze River Basin and Its Influence on NPP. Front. For. Glob. Change 2024, 6, 1335116. [Google Scholar] [CrossRef]
  10. Qi, Z.; Cai, Y.; Ye, Y.; Yang, H.; Cai, S.; Tan, Q.; Zhang, S.; Yang, Z. Integrating Landscape “Source-Transport-Sink” Mechanisms into GeoAI to Enhance Surrogate Modeling of Watershed Nutrient Loads. Water Res. 2026, 291, 125263. [Google Scholar] [CrossRef] [PubMed]
  11. Matomela, N.; Li, T.; Zhang, P.; Ikhumhen, H.O.; Lopes, N.D.R. Role of Landscape and Land-Use Transformation on Nonpoint Source Pollution and Runoff Distribution in the Dongsheng Basin, China. Sustainability 2023, 15, 8325. [Google Scholar] [CrossRef]
  12. Yin, S.; Wang, Y.; Lei, C.; Zhang, J. Runoff Responses to Landscape Pattern Changes and Their Quantitative Attributions Across Different Time Scales in Ecologically Fragile Basins. Catena 2025, 249, 108716. [Google Scholar] [CrossRef]
  13. Wei, J.; Wang, Z.; Han, L.; Shang, J.; Zhao, B. Analysis of Spatio-Temporal Evolution Characteristics of Drought and Its Driving Factors in Yangtze River Basin Based on SPEI. Atmosphere 2022, 13, 1986. [Google Scholar] [CrossRef]
  14. Canedo Rosso, C.; Nyberg, L.; Pechlivanidis, I. Drought Hazard Assessment across Sweden’s Diverse Hydro-Climatic Regimes. Nat. Hazards Earth Syst. Sci. 2025, 25, 4577–4592. [Google Scholar] [CrossRef]
  15. Song, Q.; Wang, Z.; Wu, T. Risk Analysis and Assessment of Water Resource Carrying Capacity Based on Weighted Gray Model with Improved Entropy Weighting Method in the Central Plains Region of China. Ecol. Indic. 2024, 160, 111907. [Google Scholar] [CrossRef]
  16. Pan, Z.; Gao, G.; Fu, B. Spatiotemporal Changes and Driving Forces of Ecosystem Vulnerability in the Yangtze River Basin, China: Quantification Using Habitat-Structure-Function Framework. Sci. Total Environ. 2022, 835, 155494. [Google Scholar] [CrossRef]
  17. Zhao, Y.-L.; Sun, H.-J.; Wang, X.-D.; Ding, J.; Lu, M.-Y.; Pang, J.-W.; Zhou, D.-P.; Liang, M.; Ren, N.-Q.; Yang, S.-S. Spatiotemporal Drivers of Urban Water Pollution: Assessment of 102 Cities across the Yangtze River Basin. Environ. Sci. Ecotechnol. 2024, 20, 100412. [Google Scholar] [CrossRef] [PubMed]
  18. Li, Q.; Ge, J.; Zhang, X.; Wu, X.; Fan, H.; Yang, L. Assessment of the Interaction Between Digital Infrastructure and Ecological Resilience: Insights from Yangtze River Delta Urban Agglomeration in China. J. Clean. Prod. 2025, 486, 144364. [Google Scholar] [CrossRef]
  19. Sun, J.; Che, M.; Yang, F.; Zhang, C.; Yin, S.; Wei, C. Temporal and Spatial Variability of Landscape Ecological Risk in the Yangtze River Midstream Urban Agglomeration in the Context of Climate Change. Hum. Ecol. Risk Assess. Int. J. 2025, 31, 165–194. [Google Scholar] [CrossRef]
  20. Sun, J.; Bi, S.; Bashir, B.; Ge, Z.; Wu, K.; Alsalman, A.; Ayugi, B.O.; Alsafadi, K. Historical Trends and Characteristics of Meteorological Drought Based on Standardized Precipitation Index and Standardized Precipitation Evapotranspiration Index over the Past 70 Years in China (1951–2020). Sustainability 2023, 15, 10875. [Google Scholar] [CrossRef]
  21. Zhu, X.; Wei, S.; Wu, Y. Eco-Environmental Assessment and Trend Analysis of the Yangtze River Middle Reaches Megalopolis Based on a Modified Remote Sensing Ecological Index. Sustainability 2024, 16, 8118. [Google Scholar] [CrossRef]
  22. Peng, L.; Wu, H.; Li, Z. Spatial–Temporal Evolutions of Ecological Environment Quality and Ecological Resilience Pattern in the Middle and Lower Reaches of the Yangtze River Economic Belt. Remote Sens. 2023, 15, 430. [Google Scholar] [CrossRef]
  23. Li, Y.; Wang, Z.; Liu, S. Enhance Carbon Emission Prediction Using Bidirectional Long Short-Term Memory Model Based on Text-Based and Data-Driven Multimodal Information Fusion. J. Clean. Prod. 2024, 471, 143301. [Google Scholar] [CrossRef]
  24. Li, Z.; Yu, K.; Zhong, J.; Yang, J.; Zhang, D.; Zhu, J. Spatial Correlation Network Characteristics and Influencing Factors of Water Environmental Efficiency in Three Major Urban Agglomerations in the Yangtze River Basin, China. Sustain. Cities Soc. 2024, 104, 105311. [Google Scholar] [CrossRef]
  25. Shen, W.; Zheng, Z.; Pan, L.; Qin, Y.; Li, Y. A Integrated Method for Assessing the Urban Ecosystem Health of Rapid Urbanized Area in China Based on SFPHD Framework. Ecol. Indic. 2021, 121, 107071. [Google Scholar] [CrossRef]
  26. Wang, Z.; Xu, N.; Bao, X.; Wu, J.; Cui, X. Spatio-Temporal Deep Learning Model for Accurate Streamflow Prediction with Multi-Source Data Fusion. Environ. Model. Softw. 2024, 178, 106091. [Google Scholar] [CrossRef]
  27. Cheng, L.; Wang, Z.; Pei, R.; Wu, J. Sustainable Urban Management and Flood Resilience in China’s Yangtze River Economic Belt: Drivers, Patterns, and Policy Synergies. Sustain. Cities Soc. 2025, 131, 106737. [Google Scholar] [CrossRef]
  28. Huang, S.; Gan, Y.; Chen, N.; Wang, C.; Zhang, X.; Li, C.; Horton, D.E. Urbanization Enhances Channel and Surface Runoff: A Quantitative Analysis Using Both Physical and Empirical Models over the Yangtze River Basin. J. Hydrol. 2024, 635, 131194. [Google Scholar] [CrossRef]
  29. Liang, H.; Zhang, D.; Wang, W.; Yu, S.; Nimai, S. Evaluating Future Water Security in the Upper Yangtze River Basin Under a Changing Environment. Sci. Total Environ. 2023, 889, 164101. [Google Scholar] [CrossRef] [PubMed]
  30. Yang, J.; Huang, X. The 30 m Annual Land Cover Dataset and Its Dynamics in China from 1990 to 2019. Earth Syst. Sci. Data 2021, 13, 3907–3925. [Google Scholar] [CrossRef]
  31. Zhang, Q.; Miao, C.; Su, J.; Gou, J.; Hu, J.; Zhao, X.; Xu, Y. A New High-Resolution Multi-Drought-Index Dataset for Mainland China. Earth Syst. Sci. Data 2025, 17, 837–853. [Google Scholar] [CrossRef]
  32. Sims, K.; Reith, A.; Bright, E.; Kaufman, J.; Pyle, J.; Epting, J.; Gonzales, J.; Adams, D.; Powell, E.; Urban, M.; et al. LandScan Global 2022 [Data Set]; Oak Ridge National Laboratory: Oak Ridge, TN, USA, 2023. [CrossRef]
  33. Chen, Z.; Yu, B.; Yang, C.; Zhou, Y.; Yao, S.; Qian, X.; Wang, C.; Wu, B.; Wu, J. An Extended Time Series (2000–2018) of Global NPP-VIIRS-like Nighttime Light Data from a Cross-Sensor Calibration. Earth Syst. Sci. Data 2021, 13, 889–906. [Google Scholar] [CrossRef]
  34. Shen, J.; Du, S.; Ma, Q.; Huang, Q.; Wen, J.; Yin, Z.; Gao, J. A New Multiple Return-Period Framework of Flood Regulation Service—Applied in Yangtze River Basin. Ecol. Indic. 2021, 125, 107441. [Google Scholar] [CrossRef]
  35. Qing, X.; Li, Y.; Li, W.; Lu, Z.; Yue, R. To Refine Differential Land Use Strategies by Developing Landscape Risk Assessment for Urban Agglomerations in the Yellow River Basin of China. Environ. Impact Assess. Rev. 2026, 117, 108162. [Google Scholar] [CrossRef]
  36. Li, Z.-T.; Li, M.; Xia, B.-C. Spatio-Temporal Dynamics of Ecological Security Pattern of the Pearl River Delta Urban Agglomeration Based on LUCC Simulation. Ecol. Indic. 2020, 114, 106319. [Google Scholar] [CrossRef]
  37. Urban, M.C.; Alberti, M.; De Meester, L.; Zhou, Y.; Verrelli, B.C.; Szulkin, M.; Schmidt, C.; Savage, A.M.; Roberts, P.; Rivkin, L.R.; et al. Interactions Between Climate Change and Urbanization Will Shape the Future of Biodiversity. Nat. Clim. Change 2024, 14, 436–447. [Google Scholar] [CrossRef]
  38. Chen, E.; Yu, X. (Bill) Real-Time Control of Urban Drainage System for Flood and Combined Sewer Overflow Mitigation with a Novel Recurrent Deep Reinforcement Learning Framework. Water Res. 2026, 299, 125797. [Google Scholar] [CrossRef] [PubMed]
  39. Kuang, W.; Huang, G.; Zhou, X.; Wang, S.; Li, Y.; Yu, B. Intensifying Extreme Rainfall Will Threaten the Socioeconomic Development of the Yangtze River Basin. Int. J. Climatol. 2025, 45, e70062. [Google Scholar] [CrossRef]
  40. Xia, X.; Zhou, F.; Chen, J. Spatiotemporal Heterogeneity and Zoning Strategies for Urban Ecological Resilience in Yichang, China. Land Degrad. Dev. 2025, 36, 5957–5975. [Google Scholar] [CrossRef]
  41. Li, Y.; Ji, C.; Wang, P.; Huang, L. Proactive Intervention of Green Infrastructure on Flood Regulation and Mitigation Service Based on Landscape Pattern. J. Clean. Prod. 2023, 419, 138152. [Google Scholar] [CrossRef]
  42. Hettiarachchi, S.; Wasko, C.; Sharma, A. Rethinking Urban Storm Water Management through Resilience—The Case for Using Green Infrastructure in Our Warming World. Cities 2022, 128, 103789. [Google Scholar] [CrossRef]
  43. Leng, L.; Jia, H.; Chen, A.S.; Zhu, D.Z.; Xu, T.; Yu, S. Multi-Objective Optimization for Green-Grey Infrastructures in Response to External Uncertainties. Sci. Total Environ. 2021, 775, 145831. [Google Scholar] [CrossRef]
  44. Yang, W.; Zhang, J. Assessing the Performance of Gray and Green Strategies for Sustainable Urban Drainage System Development: A Multi-Criteria Decision-Making Analysis. J. Clean. Prod. 2021, 293, 126191. [Google Scholar] [CrossRef]
  45. Yang, M.; Xue, L.; Liu, Y.; Liu, S.; Han, Q.; Yang, L.; Chi, Y. Asymmetric Response of Vegetation GPP to Impervious Surface Expansion: Case Studies in the Yellow and Yangtze River Basins. Environ. Res. 2024, 243, 117813. [Google Scholar] [CrossRef]
  46. Zhu, B.; Lin, J.; Liu, Y.; Zhang, D.; Peng, Q.; Ren, Y.; Chen, J.; Xu, Y. Multi-Risk Interaction Analysis of Cascade Hydropower Stations Based on System Dynamics Simulation. Water Resour. Manag. 2024, 38, 45–62. [Google Scholar] [CrossRef]
  47. Jiang, Q.; Xiong, S.; Yang, F.; Huang, J. Simulation Study on Rain-Flood Regulation in Urban “Gray-Green-Blue” Spaces Based on System Dynamics: A Case Study of the Guitang River Basin in Changsha. Water 2023, 16, 109. [Google Scholar] [CrossRef]
  48. Dai, K.; Shen, S.; Cheng, C.; Song, Y. Integrated Evaluation and Attribution of Urban Flood Risk Mitigation Capacity: A Case of Zhengzhou, China. J. Hydrol. Reg. Stud. 2023, 50, 101567. [Google Scholar] [CrossRef]
  49. Wang, Z.; Zhu, Z.; Luan, H.; Wu, T. Multi-Objective Optimal Scheduling of Cascade Reservoirs in Complex Basin Systems: Case Study of the Jinsha River-Yalong River Confluence Basin in China. J. Hydrol. Reg. Stud. 2025, 58, 102240. [Google Scholar] [CrossRef]
  50. Qi, W.; Ma, C.; Xu, H.; Chen, Z.; Zhao, K.; Han, H. Low Impact Development Measures Spatial Arrangement for Urban Flood Mitigation: An Exploratory Optimal Framework Based on Source Tracking. Water Resour. Manag. 2021, 35, 3755–3770. [Google Scholar] [CrossRef]
Water 18 01421 g001
Figure 1. Geographic location of the three major urban agglomerations in the Yangtze River Basin. (a) National location of the study area; (b) Scope of CYUA, MRC and YRD; (c) DEM terrain and spatial distribution of typical lakes.
Figure 1. Geographic location of the three major urban agglomerations in the Yangtze River Basin. (a) National location of the study area; (b) Scope of CYUA, MRC and YRD; (c) DEM terrain and spatial distribution of typical lakes.
Water 18 01421 g001
Water 18 01421 g002
Figure 2. Spatiotemporal distribution of ecological risk index (ERI) in the three major urban agglomerations from 2000 to 2020. (a) ERI spatial pattern in 2000; (b) ERI spatial pattern in 2005; (c) ERI spatial pattern in 2010; (d) ERI spatial pattern in 2015; (e) ERI spatial pattern in 2020.
Figure 2. Spatiotemporal distribution of ecological risk index (ERI) in the three major urban agglomerations from 2000 to 2020. (a) ERI spatial pattern in 2000; (b) ERI spatial pattern in 2005; (c) ERI spatial pattern in 2010; (d) ERI spatial pattern in 2015; (e) ERI spatial pattern in 2020.
Water 18 01421 g002
Water 18 01421 g003
Figure 3. Explanatory power of SPI and population density on ERI.
Figure 3. Explanatory power of SPI and population density on ERI.
Water 18 01421 g003
Water 18 01421 g004
Figure 4. Explanatory power of driving factors on ERI. (a) Chengdu-Chongqing Urban Agglomeration (CYUA); (b) Middle Reaches of the Yangtze River City Cluster (MRC); (c) Yangtze River Delta (YRD).
Figure 4. Explanatory power of driving factors on ERI. (a) Chengdu-Chongqing Urban Agglomeration (CYUA); (b) Middle Reaches of the Yangtze River City Cluster (MRC); (c) Yangtze River Delta (YRD).
Water 18 01421 g004
Water 18 01421 g005
Figure 5. Regression coefficients between driving factors and ERI.
Figure 5. Regression coefficients between driving factors and ERI.
Water 18 01421 g005
Water 18 01421 g006
Figure 6. Structural equation model path diagrams for CYUA. Standardized coefficients are shown; *** p < 0.01. Non-significant paths are omitted. Green arrows denote negative standardized path coefficients, red arrows denote positive standardized path coefficients, and thicker arrows correspond to larger absolute values of coefficients.
Figure 6. Structural equation model path diagrams for CYUA. Standardized coefficients are shown; *** p < 0.01. Non-significant paths are omitted. Green arrows denote negative standardized path coefficients, red arrows denote positive standardized path coefficients, and thicker arrows correspond to larger absolute values of coefficients.
Water 18 01421 g006
Water 18 01421 g007
Figure 7. Structural equation model path diagrams for MRC. Standardized coefficients are shown; *** p < 0.01. Non-significant paths are omitted. Green arrows denote negative standardized path coefficients, red arrows denote positive standardized path coefficients, and thicker arrows correspond to larger absolute values of coefficients.
Figure 7. Structural equation model path diagrams for MRC. Standardized coefficients are shown; *** p < 0.01. Non-significant paths are omitted. Green arrows denote negative standardized path coefficients, red arrows denote positive standardized path coefficients, and thicker arrows correspond to larger absolute values of coefficients.
Water 18 01421 g007
Water 18 01421 g008
Figure 8. Structural equation model path diagrams for YRD. Standardized coefficients are shown; *** p < 0.01. Non-significant paths are omitted. Green arrows denote negative standardized path coefficients, red arrows denote positive standardized path coefficients, and thicker arrows correspond to larger absolute values of coefficients.
Figure 8. Structural equation model path diagrams for YRD. Standardized coefficients are shown; *** p < 0.01. Non-significant paths are omitted. Green arrows denote negative standardized path coefficients, red arrows denote positive standardized path coefficients, and thicker arrows correspond to larger absolute values of coefficients.
Water 18 01421 g008
Water 18 01421 g009
Figure 9. Zone-based water risk management strategies for the three major urban agglomerations in the Yangtze River Basin.
Figure 9. Zone-based water risk management strategies for the three major urban agglomerations in the Yangtze River Basin.
Water 18 01421 g009
Table 1. Summary of multi-source data used in this study.
Table 1. Summary of multi-source data used in this study.
CategoryData VariableSource/ProductResolutionTemporal RangePrimary Use
Land use/coverLand cover typeCLCD (Wuhan University)30 m2000–2020 (annual)ERI calculation (landscape metrics)
NaturalElevationDEM, Chinese Academy of Sciences30 m–Topographic control factor
NaturalAnnual precipitationRESDC, Chinese Academy of Sciences1 km2000–2020Climatic driver
NaturalAverage annual temperatureRESDC, Chinese Academy of Sciences1 km2000–2020Climatic driver
NaturalNDVIMODIS MOD13A31 km2000–2020Vegetation regulation
NaturalSPI-6Derived from precipitation (Zhang et al., 2025) [31]1 km2000–2020 (annual)Hydroclimatic extreme indicator
SocioeconomicGDPRESDC, Chinese Academy of Sciences1 km2000–2020Anthropogenic pressure
SocioeconomicPopulation densityLandScan [32]1 km2000–2020Anthropogenic pressure
SocioeconomicNighttime light intensityNPP-VIIRS-like [33] 1 km2000–2020Impervious surface proxy, urban activity
SocioeconomicPer capita GDPCalculated (GDP/population density)1 km2000–2020Economic intensity
Table 2. Fit indices for the three major urban agglomerations.
Table 2. Fit indices for the three major urban agglomerations.
Urban AgglomerationSample SizeChi/DFGFIAGFICFIRMSEA
CYUA13123.4870.9900.9720.9890.044
MRC19151.8740.9960.9890.9970.021
YRD11153.1850.9920.9700.9920.044
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Tao, J.; Ma, T.; Meng, H. Landscape-Derived Indicators of Water-Related Ecological Risks: Multi-Scale Drivers and Zoned Governance in Yangtze River Basin Urban Agglomerations. Water 2026, 18, 1421. https://doi.org/10.3390/w18121421

AMA Style

Tao J, Ma T, Meng H. Landscape-Derived Indicators of Water-Related Ecological Risks: Multi-Scale Drivers and Zoned Governance in Yangtze River Basin Urban Agglomerations. Water. 2026; 18(12):1421. https://doi.org/10.3390/w18121421

Chicago/Turabian Style

Tao, Jing, Tianli Ma, and Huajun Meng. 2026. "Landscape-Derived Indicators of Water-Related Ecological Risks: Multi-Scale Drivers and Zoned Governance in Yangtze River Basin Urban Agglomerations" Water 18, no. 12: 1421. https://doi.org/10.3390/w18121421

APA Style

Tao, J., Ma, T., & Meng, H. (2026). Landscape-Derived Indicators of Water-Related Ecological Risks: Multi-Scale Drivers and Zoned Governance in Yangtze River Basin Urban Agglomerations. Water, 18(12), 1421. https://doi.org/10.3390/w18121421

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Article metric data becomes available approximately 24 hours after publication online.
Back to TopTop