Urban Blue Space Quality Promotion and Health of Residents: Evidence from Qingdao, China

Abstract

Urban blue spaces (UBSs) play a pivotal role in supporting ecological integrity and public health, yet the causal mechanisms connecting the magnitude of water quality improvements to specific health outcomes remain insufficiently explored. The objective of the study is to investigate the effects of two large-scale surface water quality initiatives on the health and physical activity patterns of residents in Qingdao, China: a river pollution clean-up program and a shoreline and estuary monitoring program. Employing a quasi-experimental difference-in-differences (DID) framework with repeated cross-sectional survey data (2017 and 2018; n = 735), we evaluate effects on self-rated health (SRH), happiness, physical activity levels, and body mass index (BMI). Results reveal that both programs significantly enhanced exercise frequency. River pollution clean-up could promote SRH by 0.319 points on the 5-point scale (β = 0.319, p < 0.05), while shoreline quality monitoring program boosted happiness by 0.233 points on the 5-point scale (β = 0.233, p < 0.05). In contrast, both interventions had no significant effect on BMI (p > 0.1). Subgroup analysis revealed that the health and well-being benefits of water quality improvements varied by gender, age, education, and income level. These findings emphasize the importance of prioritizing high-impact UBS interventions in degraded urban areas and integrating them with accessible infrastructure to optimize health gains in a more effective and equitable way.

1. Introduction

Urban blue spaces (UBSs)—including rivers, coastlines, lakes, ponds, and canals—are indispensable components of urban natural infrastructure. They regulate microclimates, attenuate flooding and stormwater runoff, and provide daily settings for recreation and social interaction. Beyond their ecological services, these waterscapes serve as emblematic symbols of urban identity and important determinants of life quality [1]. Recognizing these multifaceted roles, cities worldwide have prioritized waterfront enhancement—improving surface water quality, integrating river corridors with adjacent green spaces to form continuous green–blue systems, and creating venues for water-based recreation [2]. Such measures not only elevate real estate values but also encourage outdoor activity [3], fostering mental, physical, and social well-being [4].

An expanding body of evidence underscores the health benefits of blue-space exposure. Visually appealing and ecologically healthy water surface reduces stress [5], improve mood and emotional regulation [6,7,8], and offer restorative sensory experiences that enhance subjective well-being and life satisfaction [9,10,11,12,13,14]. Such benefits are frequently related to proximity, where residents living nearer to coasts and other types of waterbodies report better mental health [15,16]. Blue spaces also promote physical activity—walking, running, cycling, and water sports [17,18]—contributing to improved cardiopulmonary function [8], weight control, and reduced chronic-disease risk [19,20]. Even in cities with scarce water features, urban blue spaces can still attract residents to participate in physical activities [21,22].

Water quality is central to delivering these benefits. In dense urban settings, residents are especially sensitive to pollution [23], and poor water quality is consistently linked to diminished life satisfaction, health status, and overall well-being [24]. Multiple mechanisms could explain this association. High-quality water enhances visual clarity and esthetic appeal, reinforcing psychological restoration [25,26]. Clean, odor-free environments invite frequent visits [27], whereas polluted waters with unpleasant odors or visible contamination deter use and diminish mental-health gains [28]. Sustained exposure to clean waterfronts fosters place attachment and environmental identity [29], while ecological degradation can trigger environmental anxiety [30]. As social hubs, high-quality blue spaces encourage community interaction, reduce social isolation, and strengthen cohesion [31,32,33]. Physical health benefits are also contingent on water quality: pollution raises concerns over safety and infection risk, discouraging water-adjacent activities [34], and, in severe cases, posing direct health hazards [35]. Visual clarity and landscape integrity enhance environmental attractiveness, extending activity duration and frequency, while pollution characteristics weaken willingness to participate [36]. Conversely, well-maintained waterfronts, often accompanied by upgraded infrastructure, support more frequent and sustained physical activity [3,37].

Despite mounting evidence on the benefits of blue spaces, several gaps constrain our understanding of how water-quality improvements translate into health outcomes. First, the magnitude of water-quality improvements required to produce measurable health benefits remains poorly understood. Many studies identify associations between “good” water quality and health outcomes but rarely quantify the gradient of benefits as water quality shifts from poor to moderate or from poor to excellent. Without this, planners and decision makers lack evidence on whether incremental improvements—such as meeting basic safety thresholds—are sufficient, or whether substantial ecological restoration is necessary to achieve significant mental and physical health gains [19,24,38]. Second, few studies explicitly compare different waterbody types, such as rivers versus coastal shorelines. These environments differ markedly in ecological processes, patterns of use, and social meaning. Rivers in dense urban cores often provide narrow, linear spaces used for commuting and short recreational visits, while coastal shorelines offer broader, more open landscapes that can support extended leisure activities and stronger place identity. The relative health impacts of quality improvements in these contexts remain unclear, limiting the ability to tailor interventions to local environmental and cultural conditions [5,29,39]. Third, the predominance of cross-sectional designs in existing literature limits causal inference. While such studies demonstrate correlations, they have limited capability to rule out reverse causality or confounding factors [13,40].

Against the backdrop of growing global efforts to rehabilitate coastal ecosystems and enhance blue spaces, there is an increasing demand for integrated remediation strategies that address both ecological and recreational functions. In China, rapid urbanization and intensive coastal development have imposed significant pressures on both the inland river network and nearshore environments, underscoring the urgent need for targeted policies and actions to mitigate pollution and restore marine habitats. Responding to these challenges, the Chinese government launched the “Blue Bay Remediation Action” in 2016 through the Ministry of Finance and the State Oceanic Administration [41]. This national initiative provided financial and policy support to coastal cities for implementing comprehensive ecological remediation projects to address severe issues, including land-based pollution and shoreline degradation. Key measures included coastal rehabilitation, pollution control, wetland vegetation restoration, and enhanced marine ecosystem monitoring. As one of the key beneficiary cities, Qingdao is among the earliest to implement a series of specialized projects that provide a concrete context and numerous cases for evaluating urban blue space management and recovery.

This study addresses these abovementioned gaps by taking advantage of two large-scale water-quality initiatives in Qingdao, Shandong Province, China: (1) a river-cleaning program targeting heavily polluted reaches in built-up areas, representing intensive physical restoration, and (2) a shoreline and estuary monitoring program establishing continuous surveillance and rapid response to emerging coastal pollution. These interventions, both implemented at the end of the year 2017, differ in magnitude and operational mode, creating a natural quasi-experimental setting. Using repeated cross-sectional survey data from 2017 and 2018, we apply a difference-in-differences framework to estimate how exposure to these programs affected self-rated health (SRH), happiness, physical activity, and body mass index (BMI). The primary objective of this study is threefold: (1) to examine the extent to which the magnitude of water-quality improvements influences self-rated health, happiness, physical activity levels, and body mass index (BMI) of residents; (2) to assess whether the health impacts differ between built-up riverine environments and coastal shorelines; and (3) to explore the heterogeneous effects of water-quality promotion across different population groups.

2. Materials and Methods

2.1. Study Area

Qingdao, located on the southern coast of the Shandong Peninsula along the Yellow Sea, has a land area of approximately 11,300 km² [42] and a sea area of 12,200 km² [43]. As of 2024, the permanent population is 10.44 million, with 78.87% residing in urban areas [44]. Qingdao features a temperate monsoon climate and is administratively divided into seven districts and three county-level areas. The city has abundant water resources, boasting a dense water network with a total of 224 rivers distributed across three major river basins—the Dagu River, North Jiaolai River, and coastal river basins, among which 50 rivers flow into the sea. Additionally, the city has an extensive coastline with a total length of 905.2 km [43], covering coastal zones along the Yellow Sea. The abundant urban blue space resources, mainly rivers and oceans, are crucial to daily life and tourism. However, rapid economic and population growth triggered resource over-exploitation and environmental degradation, causing severe pollution in water bodies—including wastewater discharge and heavy metal contamination in both rivers within urban built-up and coastal lines [45]. This significantly degraded coastal and inland water quality, directly impairing public recreational activities and threatening the tourism-dependent economy [46].

Since 2016, Qingdao municipality has proposed a series of water quality control measures aiming to eliminate polluted water bodies by the end of 2017 [47], which include a wide range of vigorous management approaches [45]. Among them, two programs aim to recover the inland and coastal lines’ water quality. One is a river pollution clean-up program and the other a shoreline and estuary monitoring program. The former aims to clean all inland river pollution through dredging, bank reinforcement, ecological restoration, and point-source pollution control, and restore them to Class IV water standards. According to the national standard, Class IV water quality, ranked fourth among the five-level surface water quality classification, is normally required for water sources of industrial production and entertainment areas, such as rivers with no direct human contact [48]. As a result, by December 2017, all heavily polluted 14 rivers (see Figure 1 for example) had been cleaned up [49]. The latter program is a monitoring program, also called “The River Chief System and Bay Chief System”, which was launched at the end of 2017. This institutional framework mandated regular monitoring, reporting, and remediation actions for coastal areas. The major purpose is to strengthen the regulatory responsibilities of government departments and to promote tangible improvements along the coastline. Similarly, the outcome of the continuous efforts led to an excellent quality of 71.8% of water area in Jiaozhou Bay, and the coastal water quality was significantly improved [50].

2.2. Survey Design

The data for this study were derived from the 2017 and 2018 waves of the Shandong General Social Survey (SGSS), administered by Shandong University. The survey consistently employed a multi-stage stratified sampling design covering all 17 prefecture-level cities in Shandong Province, including Qingdao. In the first stage, counties or districts were selected. In the second stage, villages and urban communities were sampled within those areas. From each selected site, 25 households were drawn using probability proportional to size (PPS) sampling. Finally, one adult aged 18 or above was randomly selected from each household using the Kish grid method. We extracted the residents living within Qingdao’s administrative area, yielding 387 valid questionnaires in 2017 and 348 in 2018. The detailed sample sizes for each district over two years are shown in Figure S1 in the Supplementary Material. The 2017 and 2018 samples are independent cross-sectional cohorts instead of a longitudinal follow-up of the same individuals. All respondents met the core criteria of being older than 18 years and having resided in Qingdao for at least six months at the time of the survey. Geographic coordinates were obtained by matching detailed address records of the residential community with Amap (an equivalent of Google Maps in China) geocoding service and manually verified for accuracy [51]. The annual survey starts in June and collects the majority of respondents from June to September for several reasons. First, it avoids major public holidays, such as the national holidays in April, May, and October that could impede participant availability. Second, June often has mild weather, which promotes greater community engagement in outdoor activities such as visiting waterfronts or shorelines. Lastly, the season aligns with several previous national and local survey practices in the country, whose outcomes support the idea that this period often has the best sampling efficiency and response rates. This approach reinforces the robustness and representativeness of our data collection framework.

2.3. Outcome Variables

To comprehensively assess the health and well-being of local residents, we utilized five indicators as dependent variables from the survey: self-rated health (SRH), self-reported happiness, weekly moderate physical activity times, frequency of physical exercise, and Body Mass Index (BMI). Self-rated health was measured by the question, “How would you rate your current health?”, with responses coded as: 1 (Very unhealthy), 2 (Somewhat unhealthy), 3 (Neutral), 4 (Somewhat healthy), and 5 (Very healthy). Self-reported happiness was assessed through the question, “Overall, how happy do you feel with your life?”, with response options coded as: 1 (Very unhappy), 2 (Somewhat unhappy), 3 (Neither happy nor unhappy), 4 (Somewhat happy), and 5 (Very happy). The weekly physical activity times were captured by asking, “In the past 12 months, how many times per week did you usually engage in physical activity that lasted at least 30 min that caused you to sweat?”, for which respondents filled in a numerical value. Frequency of physical exercise was assessed by the question, “Over the past year, how often did you engage in physical activity during your free time?”, with options: 1 (Every day), 2 (Several times a week), 3 (Several times a month), 4 (Several times a year or less), and 5 (Never). BMI was calculated as the respondent’s weight in kilograms divided by the square of their height in meters [52].

2.4. Key Variables: Exposure to UBS Quality Improvement

We defined urban blue spaces as inland surface waters, oceans, and estuary environments in this study. To identify the location of surface waterbody clean-up, we utilized Qingdao municipal records that textually describe locations of fourteen designated polluted water bodies that were fully cleaned up at the end of 2017 [49]. The locations and segments (see Figure 1) of these rivers were then manually traced based on a digital map on Amap. We then measured linear distances from the community centroid to the nearest cleaned-up rivers and coastal line. We consider residents who live within 2 km of community centroids to be the ones who can benefit from the clean-up actions, based on previous studies. This distance threshold captures the feasible range for daily leisure activities and routine access, aligning with evidence that urban residents exhibit strong preferences for and derive restorative benefits from nearby blue spaces, especially given that accessible surface waters are often scarce resources in urban settings, which discourages long-distance travel for such amenities [53]. For the exposure to the coastal line, we consider residents living within 3 km distance from the coastal line exposed to the coastal environmental quality monitoring program. This distance threshold is consistent with the accessibility analysis used in urban environmental studies [54]. Another reason we are not using 2 km consistently to measure the exposure is that, in the sample, we do not have residents living within 2 km of a shoreline. Those who live beyond these two distances were considered to have limited benefits from such UBS quality promotions, thus serving as control groups. We also used people living within 2 km of all rivers as a treatment group and those beyond the distance as a control group. But this model returned insignificant results, thus not reported in the main text. All the geolocations tracing and spatial analyses were performed in ArcGIS Pro (version 3.1).

2.5. Control Variables

To account for potential confounding effects and ensure robust estimation, a standard range of demographic and socioeconomic covariates was included, such as age, gender, marital status, education level, household registration type, employment experience and status, type of residential community, total household income in the year 2016, house area, household car, and self-rated economic status. We also included factors that might be related to personal health, including illness status, health issues, and frequency of feeling depressed. Illness status was measured by “In the past 12 months, how many times have you been hospitalized due to illness or injury?” and coded into a binary variable: 1 (Being hospitalized more than once), 0 (Never). Health issues were measured by asking “In the past four weeks, how often have your health problems interfered with your work or other daily activities?”, with responses coded as: 1 (Far below average), 2 (Below average), 3 (Average), 4 (Above average), and 5 (Far above average). Depressive symptoms were measured by asking, “In the past four weeks, how often have you felt depressed or down?”, with response categories: 1 (Always), 2 (Often), 3 (Sometimes), 4 (Rarely), and 5 (Never). These health-related variables, along with other socioeconomic controls such as “community type,” were effects-coded into sets of dummy variables to avoid perfect multicollinearity. Categories with very small sample sizes (n < 10) were merged with conceptually similar groups to enhance model stability and interpretability. These decisions were made to maximize model fit, reduce bias from omitted variables, and align with conventional approaches in health geography and social epidemiology. The abovementioned survey questions in both English and Chinese were in Table S1 in Supplementary Material.

2.6. Estimation Approach

Starting from the core logic of the difference-in-difference (DID) model, and combining with the exposure distance rules for the clean-up of surface water (including river pollution clean-up program and shoreline and estuary monitoring program) in this study, a spatial DID model is constructed as follows:

$$Y_{i,t=}{\alpha }_{j,t}+{\beta }_{1,UBS,d}{D}_{UBS,d,i}+{\beta }_2\left(D_{UBS,d,i}\times {Post}_{UBS,t}\right)+\gamma X_i+{\epsilon }_{i,t}$$

In the model specification, the coefficient ${\beta }_1$ is the coefficient of the treatment group dummy variable $D_{UBS,d,i}$, commonly referred to as the “main effect”. It captures the pre-existing, inherent differences between the treatment and control groups before the policy implementation. The key parameter of interest in this study is the coefficient of the interaction term, ${\beta }_2$ ($D_{UBS,d,i}\times {Post}_{UBS,t}$), which represents the differential change in the treatment group relative to the control group after the policy implementation, that is, the net effect of the policy.

The formula uses the interaction term of “intervention time point × spatial treatment group”. Here, i represents the individual resident, i.e., the resident. j indicates the community where residents live. t is the time node, where t = 0 indicates the period before the implementation of two programs (2017) and t = 1 indicates the period after the implementation (2018). D_UBS,d,i is a dummy variable for the treatment group. It should be noted that for river treatment, d = 1 when resident i lives within 2 km from the treated river segment, otherwise d = 0; for shoreline treatment, d = 1 when resident i lives within 3 km from the shoreline, otherwise d = 0. ${\beta }_1$ and ${\beta }_2$ are the coefficients of exposure and interaction term. Y_i,t is the five outcome variables. X_i represents individual control variables, including demographic and socioeconomic characteristics, and γ is the coefficient of the control variables. α_j,t represents community-year fixed effects, controlling for time-invariant spatial characteristics of communities and time-varying trends within communities. And ε_i,t is the random error term. A triple interaction term was added to examine the heterogeneous effects of two types of water quality promotion on the subgroup. All statistical analyses were completed in STATA (version 17.0).

3. Results

3.1. Descriptive Statistics

Table S2 presents descriptive statistics of all observed variables, encompassing key variables, dependent variables, and covariates. In 2017, 40.31% of respondents reported being “Very healthy” and 32.04% “Somewhat healthy” in terms of self-rated health (SRH); by 2018, these figures were 24.71% and 37.07%, respectively. For self-reported happiness, 63.64% described themselves as “Somewhat happy” and 27.01% as “Very happy” in 2017, while in 2018, the corresponding percentages were 52.17% and 35.07%.

The average weekly moderate physical activity was 1.852 (range: 0–20; SD: 2.036) in 2017 and 2.883 (range: 0–29; SD: 3.228) in 2018. Regarding exercise frequency, 41.60% of respondents reported “Never” engaging in physical exercise in 2017, a proportion that increased to 49.86% in 2018. The average BMI was 23.678 (range: 9.375–33.121; SD: 3.576) in 2017 and 23.826 (range: 10.381–32.951; SD: 3.453) in 2018.

Covariates include household income, household economic status, gender, marital status, age, highest degree earned, hukou status, and work experience and status, among others. Missing values were treated as separate categories to retain more observations. Detailed statistical information for these variables is presented in Table S2.

3.2. UBS Quality Promotion on Subjective Well-Being

Figure 2 plots the coefficients of the main effect and the interaction term only. Full regression outcomes could be found in Table S3.

The river clean-up program was associated with significant improvements in self-rated health (SRH) after implementation. The interaction term here represents the differential change in the treatment group relative to the control group after the policy implementation, which is our key interest. While the main effect of living near polluted rivers was negative and not statistically significant (β = −0.257, p > 0.1), the interaction term for post-intervention exposure was positive and statistically significant (β = 0.319, p < 0.05), indicating improved SRH after water quality enhancements (Table S3, Model 1a). However, the program showed no significant effect on happiness. The main effect was negative and significant (β = −0.382, p < 0.01), while the interaction term was positive but not statistically significant (β = 0.179, p > 0.1), suggesting no measurable post-intervention gain in happiness (Table S3, Model 2a).

In contrast, the shoreline monitoring program did not significantly affect SRH. Both the main effect (β = −0.170, p > 0.1) and the interaction term (β = 0.105, p > 0.1) were not significant (Table S3, Model 1b). However, happiness significantly improved after monitoring was introduced. While the main effect was negative (β = −0.408, p < 0.01), the interaction term was positive and statistically significant (β = 0.233, p < 0.05), suggesting enhanced happiness following program implementation (Table S3, Model 2b).

3.3. UBS Quality Promotion on Physical Activity and BMI

The river clean-up program significantly increased physical activity levels. The interaction term here reflects the net policy impact, which is measured by the difference in the treatment and control groups after the policy implementation. For weekly moderate activity frequency, the interaction term was positive and highly significant (β = 1.255, p < 0.01), though the main effect was not significant (β = 0.026, p > 0.1) (Table S3, Model 3a). Similarly, for exercise frequency (reverse-coded; lower values indicate higher frequency), the post-intervention interaction term was negative and highly significant (β = −0.779, p < 0.01), implying an increase in regular exercise near cleaned-up rivers (Table S3, Model 4a). For BMI, the main effect of river treatment was negative and significant (β = −1.448, p < 0.05), but no significant post-intervention change was detected (interaction β = −0.559, p > 0.1) (Table S3, Model 5a).

For the shoreline monitoring program, weekly moderate activity was not significantly affected (main effect: β = 0.223, p > 0.1; interaction: β = 0.736, p > 0.1) (Table S3, Model 3b). However, exercise frequency improved post-monitoring, with a negative and significant interaction term (β = −0.639, p < 0.05) (Table S3, Model 4b). BMI was significantly lower for those exposed to shoreline monitoring (main effect: β = −1.732, p < 0.05), although the interaction effect remained non-significant (β = 0.124, p > 0.1) (Table S3, Model 5b).

3.4. Subgroup Analyses

Figure 3 plots the coefficients of interaction terms by different groups only. Full regression outcomes of subgroup analyses are available in Tables S4–S7 in the Supplementary Material.

Using triple-difference terms, we identified subgroup heterogeneity in post-treatment effects. By gender, among those living close to the cleaned-up river, compared with females, males showed higher self-rated health (β = 0.404, p < 0.01), lower happiness (β = −0.276, p < 0.01), lower moderate physical activity times (β = −1.164, p < 0.05), reduced exercise frequency (β = 0.679, p < 0.01), and higher BMI (β = 2.859, p < 0.01) (Table S4). However, among those living close to the shoreline, male residents only reported significantly lower happiness than female residents after the monitoring program was implemented (β = −0.247, p < 0.05) (Table S4). By age, among those living close to the cleaned-up river, adults aged 60 years and above showed higher self-rated health (β = 0.539, p < 0.01) and lower happiness (β = −0.191, p < 0.01) than younger adults (Table S5). Along the monitored shoreline, older adults reported lower happiness (β = −0.267, p < 0.05) but an increased exercise rate (β = −0.374, p < 0.05) after program implementation (Table S5). By education, among those living close to the cleaned-up river, respondents with higher education reported lower self-rated health (β = −0.728, p < 0.01), higher happiness (β = 0.592, p < 0.01), and a lower moderate physical activity time (β = −3.496, p < 0.01) (Table S6). Similarly, higher-education respondents living close to the monitored shoreline showed lower self-rated health (β = −0.459, p < 0.05), higher happiness (β = 0.480, p < 0.01), and a lower moderate physical activity time (β = −2.593, p < 0.01) than those with lower education (Table S6). By economic status, among those living close to the cleaned-up river, residents with above-average status reported lower happiness (β = −0.258, p < 0.05), a lower moderate physical activity time (β = 0.970, p < 0.01), and a lower BMI (β = −2.510, p < 0.01) (Table S7). All other interactions were not significant at the p < 0.05 level.

3.5. Robustness Tests

To ensure the robustness of our findings, we conducted a series of complementary tests (see Supplementary Material). First, a pre-treatment covariate balance check was conducted by using samples in the year 2017. The results revealed that while age and gender were balanced across groups, other covariates showed imbalance and were statistically significant at p < 0.1 level (Table S8). Second, we tested whether there was compositional change between the 2017 and 2018 samples regarding pre-determined socioeconomic and demographical characteristics. The results showed that only the community types of residents were significantly changed (p < 0.001) (Table S9). To mitigate potential bias, we included these controls and additional control variables, including house area, household car ownership, illness status, health issues, and depression in the regression models and interpreted the results with appropriate caution. Further, we applied inverse probability weighting (IPW) to assign weights to the intervention and control groups using the same demographic and socioeconomic factors [55]. The weighted results aligned closely with the main DID estimates (Table S10). Third, a falsification (placebo) outcome test was performed by applying the DID model to outcomes theoretically unrelated to the intervention [56]. We utilized a number of visiting hospitals in the past four weeks as a placebo, and all models showed no significant effects (Table S11). Fourth, to assess the robustness of our difference-in-differences estimate given the relatively small sample size and potential violations of standard parametric assumptions, we conducted a randomization inference (RI) procedure with 1000 permutations of the treatment assignment [57]. The results (Table S12) showed that several observed DID estimates—particularly for physical activity frequency (p = 0.023 and 0.028) in both programs and physical activity times under the river clean-up program (p = 0.090)—are unlikely to have occurred by chance, suggesting statistically significant effects. The SRH for the river cleanup program and happiness for ocean monitoring program returned a marginal p-value; therefore, we recommend exercising caution when interpreting these outcomes. In contrast, BMI in both programs and SRH in the shoreline program showed high p-values, indicating limited evidence of treatment effects. Power analysis also suggests we can reliably detect only moderate changes (≥1-point difference), while the observed effect size for BMI (~0.56 points) yields low power (~40%).

4. Discussion

4.1. UBS Quality Promotion and Subjective Well-Being

Our study found that while significant improvements in water quality can enhance both self-rated health (SRH) and happiness, this effect was specific to the intervention. The river pollution clean-up program improved SRH, whereas the shoreline and estuary monitoring program was associated with improved happiness. Prior studies have demonstrated that living close to blue space is positively associated with SRH [19] and suggest that visible and tangible improvements in water conditions translate more readily into favorable health self-assessments [16,58]. In our case, the river clean-up program produced highly visible changes—clearer water, restored vegetation, and more coherent landscapes—which likely reinforced perceptions of environmental quality through repeated, embodied experiences and greater use [17]. These direct sensory cues support positive health appraisal, while newly upgraded riverbanks functioned as inviting, socially credible spaces that facilitated informal interaction and contributed to well-being [13]. Moreover, where water quality visibly improved, residents felt more comfortable engaging in a wider range of near-water activities such as lingering at the edge, informal play, or recreational fishing, thereby multiplying the occasions from which positive SRH is drawn [26]. In contrast, the shoreline and estuary monitoring program delivered preventive but largely invisible benefits such as early pollution detection, warning systems, and management response that residents may not perceive as directly improving physical conditions [15,39]. This divergence also reflects how subjective outcomes are shaped by perception and environmental preferences. SRH is more closely tied to direct, visible, and embodied experiences of environmental quality, while happiness can respond to symbolic cues of care, order, and even culture, even when physical improvements are less perceptible [33].

4.2. UBS Quality Promotion and Physical Health

In our study, both water quality promotion programs—the high-magnitude river pollution clean-up program and the lower-magnitude shoreline and estuary monitoring program—were associated with discernible shifts in physical-activity patterns, reflected in higher exercise frequency. These associations are consistent with prior findings that high-quality waterfronts attract routine activity [59]. The programs worked through changes that are readily legible to users: the removal of malodors and visible litter reduced avoidance and made stays along the edge more acceptable; routine maintenance and advisories under monitoring increased confidence that conditions were safe; and, in several corridors, the interventions coincided with parallel place-management actions such as adding linear jogging and cycling lanes, smoothing path continuity, and improving seating and access points [3,60,61]. Together, these improvements made previously underused river segments and shorelines more usable on an everyday basis, lengthening dwell time, encouraging repeat visits, and shifting short utilitarian trips into purposeful bouts of walking and light exercise [62].

By contrast, we found limited and unstable effects on BMI, which accords with mixed results in the literature on blue-space proximity and adiposity. Cross-sectional studies have identified associations between obesity and BMI [19], although the results are often mixed [16]. Longitudinal evidence indicates that a greater distance from blue space is associated with a higher risk of being overweight [63]. Several features of our setting help explain our modest BMI response. The proximity bands we observe are larger than the very short distances at which BMI differences are most often detected, such as within a clearly walkable catchment [63]. The observation window spans two survey years, which is likely too short to capture changes in routine activity and translate them into measurable weight change in repeated cross-sections [63]. Activity along our waterfronts is dominated by light to moderate intensity and short duration, which can improve perceived health without materially shifting weight in the absence of dietary change. Future research should extend follow-up, sharpen exposure measurement by using finer distance thresholds and route accessibility, incorporate objective activity tracking, and account for diet so that the conditions under which water quality promotion programs influence adiposity can be identified with greater precision [64].

4.3. Heterogeneous Effects of Water Quality Promotion on Different Groups

The subgroup patterns point to differences in how residents perceive and use improved waterscapes, which help explain the uneven gains we observed. The river pollution clean-up program was highly visible, which likely raised environmental appraisals and translated into better self-rated health among men and older adults, yet it coincided with lower reported happiness in these groups. A plausible reading is that global health judgments respond to clear environmental upgrades and status cues [65], whereas day-to-day affect is more sensitive to crowding, shifting use norms, enforcement accompanying restoration, and differences in family roles and preferences [19,31,66]. For gender, men’s pattern of fewer moderate-activity bouts, less frequent exercise, and higher BMI relative to women suggests that clean-up may have enabled women to incorporate more light movement into caregiving-related visits with children and older relatives along safer, cleaner edges [66], while women’s stronger positive affect toward blue spaces could help explain their higher happiness after quality promotion [67]. For older adults, higher self-rated health near cleaned rivers is consistent with perceived gains, whereas lower happiness may reflect noise, congestion, or reduced seating comfort that matters more to seniors [19,31,68]. Education gradients show another evaluative asymmetry. Highly educated residents reported lower self-rated health but higher happiness and fewer moderate-activity bouts in both settings, consistent with stricter self-assessment of health and greater sensitivity to esthetic and symbolic uplift [69,70]. Economic differences near cleaned rivers align with the capacity to convert improved waterfront quality into structured exercise, with above-average residents reporting more moderate activity and lower BMI, but their lower happiness may reflect higher expectations for maintenance quality [39]. Despite heavier workload constraints, lower-income residents may derive greater mental-health gains from the water clean-up because even low-intensity exposure—improved visibility, brief visits, and ancillary environmental upgrades—can yield meaningful affective benefits [15,71]. Along monitored shorelines, effects were narrower, which fits a preventive intervention whose benefits are less visible; older adults did exercise more frequently after monitoring commenced, but happiness fell, suggesting that reassurance about concerns such as safety and climate factors can nudge routine activity without delivering the same affective payoff as tangible environmental change [72].

4.4. Planning and Policy Recommendations

Our findings support differentiated, evidence-based planning strategies for river clean-up and shoreline monitoring interventions. First, river pollution clean-up programs should be prioritized in highly degraded, densely populated, and heavily trafficked areas where visible improvements in water quality are more likely to yield rapid gains in self-rated health and promote moderate exercise. Second, shoreline and estuary monitoring programs, while less visible, should be leveraged to promote subjective well-being and feelings of joy, especially in areas where water quality risks are intermittent but public use is high. To enhance their perceptual impact, such programs should be communicated visibly through real-time signage, water safety ratings, or digital dashboards to symbolically convey care, order, and responsiveness. Finally, both programs should be strategically targeted to disadvantaged communities and older populations, who may benefit disproportionately from even light exposure and symbolic signals of environmental care. Integration with easy access networks and inclusive space design is essential to ensure equitable reach. Together, these strategies link visible transformation with embodied health gains, and institutional assurance with emotional well-being, responding directly to the observed divergence in SRH and happiness outcomes.

4.5. Strengths, Limitations, and Future Studies

This study advances blue space and health research by examining temporal change in water quality rather than treating exposure as static. We use high-resolution community locations together with detailed coordinates for river clean-up segments and monitored shoreline and estuary reaches, which allows for finer spatiotemporal matching between residents and interventions. We also differentiate two water quality promotion programs with distinct magnitudes and across two waterbody types, which enables comparison of water quality promotion intensity and context. The natural policy and project rollout provide quasi-experimental leverage that strengthens interpretation beyond simple cross-sectional association, and the citywide scale enhances practical relevance for planning.

Several limitations remain. The two-wave repeated cross-sectional design is weaker than longitudinal follow-up for capturing individual change and for tracing long-run effects. We did not directly record whether respondents used the specific river or shoreline segments, which limits explanatory power. Furthermore, the use of straight-line distance from residences to water bodies may not accurately reflect actual accessibility, as it does not account for physical barriers, transportation networks, or actual travel behavior. Moreover, the two-year window is short for detecting a change in body mass index, which responds slowly to behavior. Compounding this limitation, the absence of detailed dietary data and baseline weight information restricted our ability to rigorously test potential explanations for the lack of significant change in BMI, such as compensatory dietary behaviors or pre-existing weight conditions. Given the modest sample size (n = 735) and substantial outcome variability (e.g., BMI SD = 3.5), our study is likely underpowered to detect small effect sizes. This limitation is especially important for outcomes like BMI, which are known to respond slowly to environmental changes. Future research should build multi-wave panel cohorts, link respondents to actual waterfront visits through travel diaries or privacy-preserving GPS, add objective activity tracking and higher-frequency water quality monitoring, and code micro-scale design features and pollution categories to identify the most effective levers. Longer follow-up and replication across multiple cities would improve external validity and permit assessment of trajectories in body mass index and disease risk, while event-study designs and mediation analysis can clarify how program magnitude and communication translate into durable health gains.

5. Conclusions

By using two-year repeated cross-sectional data and a DID estimation in Qingdao City, Shandong, China, this study offers evidence that improving urban blue space (UBS) quality of different magnitudes generates significant yet variable health benefits. The river pollution clean-up program enhanced self-rated health (β = 0.319, p < 0.05) and exercise frequency (β = −0.779, p < 0.01) of the residents, driven by visible pollution reduction, restored ecological esthetics, and greater recreational use. In contrast, the shoreline and estuary monitoring program boosted happiness (β = 0.233, p < 0.05) and exercise frequency (β = −0.639, p < 0.05), likely due to its preventive focus and less perceptible impacts. Neither intervention significantly affected BMI (p > 0.1), indicating short-term activity changes alone may not reduce weight without corresponding dietary or long-term behavioral adjustments. The subgroup differences highlight that the health and well-being benefits of water quality improvements are unevenly distributed, shaped by gender, age, education, and economic status. Nonetheless, this study advances urban planning by demonstrating that intensive urban blue space upgrades can effectively enhance subjective well-being and physical activity of the residents but must be implemented with attention to socioeconomic disparities to ensure health equity.