Assessing the Riparian Squeeze in the Greater Bay Area based on multisource data

Water is fundamental to human civilization and urban development. Throughout various historical epochs, humanity has cultivated intricate relationships with water, a universal phenomenon that has catalyzed the emergence of distinctive societal structures, religious beliefs, and political frameworks, shaping civilizations across geographical and temporal boundaries. Within the intricate urban systems fashioned by humans, rivers, lakes, wetlands, and oceans serve as mediums through which matter, energy, and information flow. The dynamic interplay between urban environments and water has garnered significant scholarly attention across disciplines such as anthropology, history, urban planning, and environmental studies. Embedded within the urban fabric, human interaction with water generates a variety of persistent physical forms and patterns. Water and its uses mold the urban layout in general but also on the micro-level of building practices. Not least, the presence of water decisively bears on ‘softer’ factors like a city's visual character and the quality of its ambience (Hauer et al. 2016). For instance, a significant proportion of Pompeii's population resided within a mere 80 m of water (Tainter 2019); archival research and the GIS (Geographic Information System)-based reconstruction of the past 500 years of riverscape in Wien allow for a new view of the co-evolution of the city and the river and reveal how human behaviors have substantially influenced fluvial dynamics and enabled urban development in parts of the former floodplain (Hohensinner et al. 2013).

As global urbanization accelerates in tandem with population growth, the anthropogenic impacts on waterfront areas have become increasingly pronounced. Currently, more than 50% of the world's population resides within 3 km of a surface freshwater body (Kummu et al. 2011). Anthropogenic interventions have resulted in the construction of a minimum of 30,549 flow-obstructing structures along major river systems globally (Yang et al. 2022). These modifications, implemented primarily for irrigation purposes (Wine & Laronne 2020) and transportation enhancement (Blanton & Marcus 2009), have substantially altered the natural riverscape. Consequently, human activities have induced changes in the flow regimes of approximately two-thirds of global river systems (Wang et al. 2024). These anthropogenic modifications manifest multi-dimensional impacts. Traditional pollutants and emerging contaminants (e.g. microplastics) have contaminated the majority of global river systems (Wilkinson et al. 2022; Tan et al. 2023), which, in conjunction with climate change, poses significant threats to both biodiversity (Kozlowski & Bondallaz 2013) and sustainable water supplies (Van Vliet et al. 2023). Furthermore, an estimated 2 billion people globally inhabit floodplains (Devitt et al. 2023), with human settlement expansion in flood-prone zones exhibiting an accelerating trend (Rentschler et al. 2023).

Several key concepts in urban planning and landscape architecture have emerged to assess the impacts of urban activities on waterbodies. Terms such as ‘Open Water Surface’, ‘Visible/Invisible Waters’ (Watson 2019; Arborino et al. 2024), and ‘River Fragmentation’ (Jumani et al. 2020) are frequently employed to describe these interactions. For coastal regions, researchers often utilize the concept of ‘Coastal Squeeze’ (Lithgow et al. 2019). Recently, researchers in the Netherlands conducted an extensive study to measure the proximity of the world's coastlines to nearby infrastructure or buildings using the coastline dataset from the mean high-water springs line, offering a quantitative assessment of how human activities are encroaching on coastal areas (Lansu et al. 2024). While this approach provides valuable insights, it has notable limitations. First, the study does not explore potential applications or implications of the measured distances, which restricts understanding of the broader relevance and utility of these findings. Second, although the Dutch researchers' focus on maritime environments is well-founded, it may not fully capture the complex dynamics of human activities around diverse waterbodies, particularly in nations where rivers and lakes play crucial roles in urban development. For nations such as China (Wittfogel 2017), India (Amrith 2018), and the United States (Doyle 2018), a framework that also considers rivers and lakes would offer a more comprehensive assessment of human impacts on various waterbodies.

This study addresses this gap by developing a new analytical framework – the Riparian Squeeze Index (RSI) – designed to evaluate human–water interactions across multiple types of waterbodies. Using the Greater Bay Area (GBA) as a stage case study, this research makes three primary contributions: (1) it establishes a multi-dimensional measurement system that integrates spatial distances between waterbodies and infrastructure with demographic, economic, and environmental metrics; (2) it introduces the RSI as a standardized tool for assessing human pressure on riparian zones, adaptable to diverse geographical contexts; and (3) it offers data-driven insights into the effectiveness of existing water protection legislation through systematic comparisons with RSI values. This study enhances our understanding of human–water dynamics in urban settings and provides practical tools for sustainable waterfront management.

• (1) Waterbodies and roads

The waterbodies dataset for this study was obtained from OpenStreetMap (OSM), which categorizes water features using distinct geometrical representations. Large waterbodies – including lakes, reservoirs, wetlands, and river surfaces – are mapped as polygons, while smaller rivers (typically channels less than 10 m wide or seasonal streams) are represented as polylines delineating the watercourse centerline. Polygon-represented waterbodies are defined using the high-water mark, aligning with regulations in the GBA. Although regional variations exist, core delineation principles remain fundamentally consistent. For instance, as articulated in the Shenzhen Blue Line Planning, protection boundaries for rivers and lakes without embankments are demarcated from the high-water mark, whereas, for embanked rivers, delineation commences at the embankment line (in such instances, the water bank coincides with the high-water mark).

Despite potential measurement limitations associated with smaller rivers, we deliberately retained these features in our analysis. The study area in the Pearl River Delta comprises a complex hydrological network extensively integrated with urban development. Excluding these features would significantly compromise the comprehensive representation of local hydrology. Recent research has, moreover, emphasized the critical yet frequently overlooked role of seasonal streams in modulating water quantity and quality (Brinkerhoff et al. 2024). Recognizing the inherent challenges in precisely defining water body boundaries, we validated the OSM dataset by cross-referencing high-resolution satellite imagery. This verification process confirmed the dataset's accuracy in representing the study area's hydrological network and its appropriateness for our research objectives.

The road data used in our study were also sourced from OSM. And mainly, this study focuses exclusively on paved roads. This methodological decision is grounded in the established understanding that unpaved pathways, such as dirt tracks, do not substantially influence the infiltration coefficient of land cover. Evidence demonstrates that linear infrastructure's hydrological impact is predominantly associated with paved roads, which exhibit an impact 5–6 times greater than that of unpaved roads (Raiter et al. 2018). Moreover, obtaining accurate datasets for unpaved roads presents significant methodological challenges, recent research has highlighted significant underrepresentation of road networks in Asia, largely due to informal or unauthorized constructions, bulldozed paths in logging areas, and other ‘ghost’ roads that are missing from existing datasets for various reasons (Engert et al. 2024). Furthermore, the study area benefits from relatively comprehensive data coverage, which makes this dataset appropriate for our analysis. In total, there are 150,279 km of roads and railways in the study area, with roads in opposite directions counted separately.

• (2) Buildings

Points of interests (POIs) serve as suitable proxy variables for capturing active human activities and residential development, making them widely utilized in urban research (Song et al. 2018). When interpreted using various language models, POIs can reveal semantically meaningful topics related to functional zones within cities. This vast dataset can be leveraged to scrutinize city functions at an inter-city scale, facilitating analyses such as traffic patterns, urban infrastructure assessments, and evaluations of livability (Qian et al. 2021). The POIs data utilized in this study were sourced from Amap, one of the commonly used map service providers in China. Utilizing their API, we acquired 3,838,114 POIs encompassing various sectors such as catering, commerce, transportation, education, and healthcare within the study area. Some comparable characteristics of the GIS dataset are provided in Table 1.

• (4) Demographic and other statistical data

Gender, age, income, and other social factors were additional factors considered in this study. Extensive research indicates that marginalized communities characterized by poverty, gender imbalances, and minority status tend to exhibit heightened sensitivity and structural vulnerabilities in the face of water-related disasters like droughts and floods, as well as pollution incidents (Ermagun et al. 2024; Fujiki et al. 2024; Rajput et al. 2024). The population data used in this study originate from the 100-m resolution grid data, which, based on the seventh census data, utilized random forest models for stacking ensemble learning to delineate inhabited areas (Chen et al. 2024). The remaining statistics were sourced from the Statistics Bureau of Guangdong Province, the Government of Macao Special Administrative Region Statistics and Census Service, and the Census and Statistics Department of Hong Kong.

The concept of Riparian Squeeze Distances itself is neutral. Scholars often present differing perspectives on the proximity of infrastructure and buildings to waterbodies. Environmental researchers argue that shorter distances may harm aquatic ecosystems and heighten flood risks. In contrast, property owners and urban planners may emphasize the potential benefits of closer proximity in developing water-friendly areas that foster economic growth. Recognizing the complexity of this issue, we developed the Riparian Squeeze Index (RSI), which includes three thematic dimensions with various indicators to facilitate a comprehensive evaluation. The RSI serves as an analytical framework, with selected indicators and their respective weights adjustable based on the characteristics of the study area and available data, as outlined in Table 2. Further applications and refinements of the RSI await future research.

The construction of the Spatial Distance theme draws on the Measures for the Administration of Urban Blue Line (Decree No. 145-2006). The ‘blue line’ serves as the principal regulatory boundary for the planning and preservation of waterbodies in China, designating protected zones around rivers, lakes, reservoirs, canals, and wetlands to ensure water safety and environmental conservation. In principle, the blue line prohibits construction and infrastructure development within its bounds. While the specific delineation of the blue line is subject to regional discretion, its width typically ranges from 4 to 30 m. Adherence to the blue line is a mandatory criterion during the approval process for construction projects and forms the basis for law enforcement actions by administrative authorities. To assess the level of water body protection within the study area, this study evaluates the proportion of sample points with Riparian Squeeze Distances below 5, 10, and 20 m.

For the Human Activities theme, we draw upon recent research on the ‘15-min city’ concept. While the 15-min walking distance reflects urban morphology and activity intensity, it was deemed excessive for this study. Instead, we adopted a ‘complete community’ approach, defining a Riparian Zone as the area within a 5-min walking radius (400 m) along waterbodies (Liu et al. 2024). We statistically analyzed the population and POI density within this Riparian Zone, as such areas tend to exhibit higher human activity density, reflecting a natural inclination for people to reside and interact near waterbodies.

Within the Socioeconomic Status theme, the objective is to evaluate inequality and vulnerability across different regions. Disruptions to critical infrastructure during extreme events disproportionately affect communities based on their socioeconomic and demographic characteristics. Moreover, the level of economic development in a region plays a pivotal role in determining its resilience and capacity for recovery. Economically vulnerable regions often lack the necessary resources for effective disaster preparedness and recovery, thereby exacerbating the susceptibility of their populations to adverse outcomes. This theme facilitates the identification of potential system upgrades aimed at benefiting vulnerable communities, with an emphasis on prioritizing measures that enhance resilience in the most affected areas. The selection of indicators was guided by the resilience framework established for China's National Urban Agglomeration (NUA) policy (Li et al. 2024).

Among the cities analyzed, Macau exhibits the lowest median distance at 22.28 m, highlighting the considerable pressure exerted on waterbodies due to the region's high level of urbanization. In contrast, the cities of Zhaoqing, Jiangmen, and Huizhou, characterized by lower levels of urbanization, recorded median distances of 75.33, 82.83, and 128.35 m, respectively. These findings are consistent with the observed urbanization patterns in the region. Further spatial analysis of the Riparian Squeeze Distances reveals several notable patterns. First, a clear concentric distribution pattern emerges, with distances gradually increasing from urban centers to peripheral areas. This trend is particularly pronounced in Guangzhou and Shenzhen, where the central business districts demonstrate significantly shorter distances (typically under 20 m) compared to their suburban areas (often exceeding 50 m). This spatial gradient reflects the historical development patterns of these cities, where early urban settlements were concentrated along waterways before expanding outward.

The distribution of Riparian Squeeze Distances demonstrates a strong correlation with urban morphology. In areas characterized by traditional block layouts, such as the historic districts of Guangzhou and Macau, these distances are consistently short, typically ranging from 15 to 25 m. This pattern reflects historical reliance on water-dependent development, where proximity to waterbodies was integral to urban functionality. Conversely, newer industrial zones and planned communities tend to exhibit larger distances, generally between 30 and 50 m. These increased separations are intentional, designed to mitigate environmental impacts and align with contemporary urban planning principles, such as preserving ecological integrity and adhering to regulatory frameworks such as the aforementioned ‘blue line’.

The data further reveal three primary clusters based on Theme 1 spatial distance values, as detailed in Table 3. These findings are classified as follows:

• (1) High-pressure cities (value > 35): Macau SAR (42.773) and Zhongshan (41.312) exhibit the highest pressure on waterbodies, with over 20% of their sample points located less than 10 m from the infrastructure. This indicates significant spatial constraints on waterbodies in these regions.

• (2) Moderate-pressure cities (value between 25 and 35): This cluster includes Shenzhen (33.687), Foshan (33.329), Hong Kong SAR (29.841), and Guangzhou (26.277). These cities demonstrate balanced yet significant urban pressure on waterbodies, with approximately 13–16% of sample points situated within 10 m of infrastructure.

• (3) Low-pressure cities (value < 25): Huizhou (6.508), Zhaoqing (10.994), and Dongguan (18.688) exhibit relatively lower pressure on their waterbodies. Notably, in Huizhou, only 3.2% of sample points are located within 10 m of infrastructure, suggesting a better preservation of riparian zones.

The spatial distribution of these pressures reveals a clear core-periphery pattern within the GBA. Cities in the Pearl River Delta core area generally exhibit higher pressure values, while those on the periphery demonstrate lower values. This pattern aligns with the historical development trajectory of the region, where urbanization initially concentrated around the Pearl River estuary before expanding outward.

Furthermore, the data indicate an intriguing relationship between city size and riparian pressure. While one might expect larger cities to consistently exhibit higher pressure values, the data suggest a more nuanced relationship. For instance, Zhongshan, a relatively smaller city, exhibits higher pressure (41.312) than megacities such as Guangzhou (26.277) and Shenzhen (33.687), indicating that factors beyond city size, including urban planning policies and historical development patterns, play crucial roles in determining the pressure on waterbodies.

These findings underscore the necessity for differentiated water management strategies across the GBA, particularly in high-pressure areas.

We calculated the proportion of POIs and the population within the 400 m Riparian Zone range for each city, relative to the total number. To ensure comparability despite differences in water network density across cities, we normalized these figures by the proportion of the Riparian Zone area to the overall city area. Across the GBA, Riparian Zones occupy 37% of the total area, yet they encompass 57% of the total POIs and 59% of the population, with respective relative densities of 1.55 and 1.62. This underscores Riparian Zone areas as hubs of frequent economic and human activities, highlighting the pivotal role of water in shaping urban form and vitality. It is observed that only Dongguan, Shenzhen, and Zhongshan demonstrate a relative POI density that is less than the value of 1. This peculiarity can be attributed to the manufacturing-centric character of these urban areas, which results in a significant presence of slums and a high degree of land development intensity. This situation has led to an almost exhaustive utilization of the waterfront spaces, thereby prompting a migration of human activities toward inland areas.

The trend in relative population density closely parallels that of relative POI density, with a strong correlation (R² = 0.98). This relationship is unsurprising, as numerous studies have demonstrated that POIs serve as a reliable proxy for capturing population and economic activity, particularly in contexts where data density is high, such as in China (Zhang & Zhao 2024). This finding highlights the complex interplay between human settlements and waterbodies across various stages of urbanization, emphasizing a consistent preference for proximity to water whenever feasible. Moreover, the strong correlation underscores the utility of POIs in big data applications, particularly for identifying patterns of human settlement. Analyzing POI densities offers valuable insights into the spatial dynamics of urban development and the interrelations between urbanization and the spatial proximity of human activities to waterbodies.

We have endeavored to ensure that the contributions of each theme to the overall index are as comparable as possible, maintaining a reasonable degree of balance. Our analysis of the GBA yields an average RSI of 77.75; however, this figure obscures significant spatial heterogeneity across the region. In cities experiencing above-average stress, we observe three distinct patterns in RSI distribution. First, there are high-risk urban cores, such as Guangzhou (77.939) and Macau SAR (80.342), characterized by intense urban development pressure combined with lower socioeconomic vulnerabilities. Second, we find high-risk transition zones, including Zhongshan (89.021) and Foshan (83.567), which display balanced yet significant pressures across all three themes. Third, there are high-risk peripheral areas, such as Zhaoqing (97.306) and Jiangmen (78.82), that exhibit lower spatial pressure but often have higher socioeconomic vulnerabilities.

Our cross-thematic analysis reveals critical insights into the complex relationship between development and environmental risk. A notable development-vulnerability paradox emerges in less-developed cities (Adger 2006), such as Jiangmen and Zhaoqing, which demonstrate higher Theme 3 scores (33.281 and 33.801, respectively). Despite experiencing lower spatial pressure, these areas face greater risks due to their limited adaptive capacities, thereby challenging traditional assumptions about the interplay between development and environmental risk. Conversely, economically advanced cities like Hong Kong, Guangzhou, and Shenzhen exhibit higher pressures in Themes 1 and 2 but lower vulnerability in Theme 3, suggesting that economic development may enhance resilience, even in the face of increased environmental pressures. Dongguan's unique performance (RSI: 60.704) across all themes underscores the significant role of administrative frameworks in shaping urban–water relationships. Unlike most other cities in China, Dongguan has not established the third-level administrative division (county). Instead, its governance structure really consolidates all planning and construction activities under the direct management of second-level administrative agencies (city). This streamlined approach has mitigated inter-jurisdictional conflicts and inefficiencies commonly observed in other cities, thereby enabling more cohesive and effective policy interventions to manage riparian pressures. Such administrative context, particularly for Dongguan, enhances our understanding of how specific governance structures influence urban–environment interactions.

The RSI presented in this study offers an innovative approach to assessing the pressures exerted by urbanization and human activities on waterbodies within the Guangdong–Hong Kong–Macau GBA. This research contributes to the field by establishing a multi-dimensional measurement system that integrates spatial distances between water banks and infrastructure with demographic, economic, and environmental metrics. This framework provides a standardized tool for evaluating human pressure on riparian zones across diverse geographical contexts.

Our key findings reveal significant pressures on waterbodies resulting from urban development, including:

• (1) The median distance from a waterbody to its nearest building or road within the GBA is 55.13 m, indicating substantial pressures on the aquatic environment and other water users like agriculture, industry, and recreation.

• (2) Riparian zones, which constitute 37% of the total area, contain 57% of POIs and 59% of the population, demonstrating a human preference for proximity to water. This concentration underscores the critical role of waterbodies in shaping urban vitality and economic activity.

• (3) The RSI highlights the intricate interplay between socioeconomic development and environmental risk vulnerability, particularly in sensitive and marginalized areas.

These findings are significant as they emphasize the necessity for balanced urban development and environmental conservation strategies, which are essential for sustainable water resource management and urban planning. While the RSI provides a comprehensive framework, it is not without limitations, primarily related to the data and methods employed. The current analysis is based on snapshot data and does not account for future projections or temporal dynamics. Furthermore, the analysis is conducted at the city level, which may obscure nuances at finer spatial scales, such as districts or neighborhoods.

Future research should aim to address these limitations by incorporating longitudinal data to track changes over time and developing predictive modeling capabilities. Refining the analysis to the level of districts or neighborhoods could yield more nuanced insights. The integration of additional variables, such as climate change impacts, ecological health indicators, and the cultural significance of waterbodies, could further enhance the RSI framework. Furthermore, future work should explore the applicability of the RSI in other urban agglomerations and river basins to identify shared patterns and distinct challenges.

This tool provides actionable insights for policymakers and environmental managers, enabling the formulation of evidence-based sustainable urban growth strategies. The RSI framework can play a pivotal role in supporting the implementation of the ‘blue line’ policies, providing essential data for the identification of critical hotspots along waterbodies. By integrating RSI findings, policymakers can delineate effective blue lines that balance water resource protection, disaster prevention, and economic development, particularly in vulnerable and marginal areas. In the forthcoming stages of our research, we intend to extend the application of this framework to encompass the entirety of China.

In conclusion, the RSI represents a valuable instrument for understanding and managing the complex interplay between urbanization and waterbodies. We hope that this research will promote more sustainable urban planning practices and policies, ensuring the preservation of these critical ecosystems for future generations. The adaptability of the RSI makes it a promising tool for researchers and practitioners worldwide, offering a framework for identifying vulnerabilities and guiding targeted interventions to protect both water resources and the communities that rely on them.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.