Spatio-Temporal Variability of Key Habitat Drivers in China’s Coastal Waters

Abstract

China’s coastal fisheries face challenges to their sustainability due to climate and human-induced pressures on key habitat drivers. This study provides an 18-year (2003–2020) assessment of six key ecological and data-available environmental factors (sea-surface temperature (SST), salinity, transparency, currents (eastward velocity, EV; northward velocity, NV), and net primary productivity (NPP), selected for their ecological relevance and data availability, across the Bohai, Yellow, and East China Seas at a spatial resolution of 0.083°. Non-parametric trend tests and seasonal climatologies were applied using MODIS-Aqua and CMEMS data with a refined quasi-analytical algorithm (QAA-v6). The results show distinct gradients: SST ranging from 9 to 13 °C (Bohai Sea) to >20 °C (East China Sea); transparency ranging from <5 m (turbid coasts) to 29.20 m (offshore). Seasonal peaks occurred for SST (summer: 18.92 °C), transparency (summer: 12.54 m), and primary productivity (spring: 1289 mg/m²). Long-term trends reveal regional SST warming in the northern Yellow Sea (9.78% of the area), but cooling in the central East China Sea. Widespread increases in transparency were observed (65.14% of the area), though productivity declined significantly (27.3%). The drivers showed spatial coupling (e.g., SST–salinity r = 0.95), but the long-term trends were decoupled. This study provides a comprehensive and long-term assessment of multiple key habitat drivers across China’s coastal seas. The results provide an unprecedented empirical baseline and dynamic management tools for China’s changing coastal ecosystems.

1. Introduction

China’s coastal waters, spanning the Bohai Sea, Yellow Sea, and East China Sea, constitute one of the world’s most productive marine ecoregions and support multi-million-ton fisheries that are critical for regional food security, livelihoods, and economies [1]. However, the sustainability of these fisheries is increasingly dependent on the quality and stability of fish habitats, which are influenced by various environmental factors such as sea-surface and bottom temperature, salinity, water transparency, current velocity, and pelagic primary and secondary productivity [2]. These drivers control not only fish physiology, distribution, and trophic interactions [3,4], but also modulate the carrying capacity of marine ecosystems under rapid global change [5].

Over the past two decades, China’s coastal ecosystems have been subjected to the combined pressures of climate change and intensive anthropogenic activities [1]. Rising sea-surface temperatures, altered precipitation and river discharge patterns, expanding marine aquaculture, and large-scale hydraulic engineering have collectively shifted the baseline conditions of key habitat drivers [6]. These changes may result in spatial mismatches between critical life-history stages (e.g., spawning, nursery, and feeding grounds) and suitable environmental conditions, which could ultimately undermine the resilience of commercially important species such as the small yellow croaker (Larimichthys polyactis), hairtail (Trichiurus lepturus), and the endangered Chinese sturgeon (Acipenser sinensis) [1,7].

Despite their central role in ecosystem-based fisheries management (EBFM), there is still a lack of comprehensive, long-term assessments of these habitat drivers at resolutions relevant to fish ecology (≤10 km) for China’s coastal seas. Most existing studies either rely on sparse in situ measurements that are limited to estuarine or shelf-break transects or use coarse global reanalysis products (>1°) that poorly resolve nearshore heterogeneity [8,9,10]. Consequently, our ability to detect subtle yet ecologically significant trends, attribute them to specific climatic or anthropogenic influences, and translate these findings into adaptive management strategies, remains limited.

Recent advances in ocean-color remote sensing and semi-analytical inversion algorithms now offer unprecedented opportunities to fill these data gaps. Notably, improvements in the retrieval of water transparency—an integrative proxy for light availability, turbidity, and eutrophication—have been achieved by refining traditional quasi-analytical algorithms to account for optically complex coastal waters [11]. Similarly, the open-access Copernicus Marine Service and MODIS-Aqua archives provide long-term daily records of physical and biogeochemical variables at a resolution of ~9 km, enabling robust spatio-temporal analyses.

This study builds on technological advances to present a systematic, 18-year (2003–2020) assessment of six key habitat drivers across the China Seas at a spatial resolution of (0.083°). Our primary contribution is the integration of this unprecedented spatial resolution and long-term temporal coverage with methodological refinement for coastal waters, which has not been achieved in previous regional assessments. Specifically, we quantify the spatial heterogeneity and long-term trends of each driver using non-parametric trend tests and slope estimators; evaluate seasonal cycles and their inter-annual stability; and discuss the implications of the observed patterns for EBFM and climate-adaptation strategies under the overarching framework of the UN Sustainable Development Goals. Our findings aim to support spatially explicit decision-making processes, such as dynamic habitat closures, the adaptive zoning of marine protected areas, and the implementation of early-warning systems for habitat degradation in China’s rapidly changing coastal seas.

2. Materials and Methods

2.1. Study Area

The study area (115–126° E, 26–41° N) encompasses China’s eastern continental shelf seas: the Bohai Sea, Yellow Sea, and East China Seas. This region is a key area for the interaction between the East Asian continent and the ocean, composed of three semi-enclosed marginal seas (Figure 1). It spans multiple climatic zones and is influenced by various ocean currents and river inputs. The area features high productivity, rich biodiversity, and diverse ecosystems, including estuarine wetlands, mangroves, seagrass beds, and coral reefs. These provide habitats for many species of fish [1,10]. It is also the main habitat for the critically endangered Chinese sturgeon, as well as the spawning/nursery grounds of many commercially important species [7]. Water depths range from less than 5 m to over 7000 m, providing ideal conditions for large-scale habitat analysis.

2.2. Data Sources and Pre-Processing

These six environmental drivers—SST, salinity, transparency, NPP, EV, and NV —were selected based on two primary principles: (1) their established physiological and ecological significance as key habitat determinants for major fishery species and endangered fauna in China’s coastal waters, such as the Chinese sturgeon (Acipenser sinensis), small yellow croaker (Larimichthys polyactis), and hairtail (Trichiurus lepturus) [12,13,14,15,16,17,18]; and (2) the availability of consistent, long-term (2003–2020), and high-resolution (≤9 km) data products capable of supporting basin-scale spatio-temporal analysis. While other variables like pH and dissolved oxygen (DO) are undoubtedly important habitat drivers, they were not included in this study due to the current lack of robust, high-resolution remote sensing or reanalysis products providing continuous long-term coverage at the scale of China’s coastal seas. The chosen suite of drivers comprehensively captures the fundamental physical (temperature, salinity, currents, and light availability) and biological (primary and secondary production) dimensions of the pelagic habitat, providing a holistic basis for ecosystem-based management.

2.2.1. Satellite Data

Daily MODIS-Aqua Level-2 ocean-color products (2003–2020) were obtained from NASA Ocean Biology Processing Group (OBPG; http://oceancolor.gsfc.nasa.gov/, accessed on 27 December 2024). Native resolution is 1 km at nadir. We extracted remote-sensing reflectance R_rs(λ) at seven visible bands (412–667 nm) for Secchi-disk depth (SDD) inversion.

Physical and biogeochemical variables were acquired from the Copernicus Marine Environment Monitoring Service (CMEMS): Global Ocean Physics Reanalysis (GLORYS12V1) for sea-surface and bottom temperatures (SST/BT), practical salinity, and eastward and northward surface current velocities (UO and VO) at a resolution of 0.083° × 0.083° and a frequency of once per day. Global Ocean Biogeochemistry Reanalysis for NPP was performed at the same spatio-temporal resolution. Bathymetry data were obtained from the 1-arc-minute ETOPO1 global relief model provided by NOAA-NCEI.

2.2.2. In Situ Data for SDD Calibration/Validation

A total of 225 independent SDD measurements collected between 2005 and 2020 were compiled from three sources, including the National Field Scientific Observation and Research Station of the Jiaozhou Bay Marine Ecosystem in Shandong Province, the National Earth System Science Data Center, and relevant references [11]. All observations were quality-controlled; turbid outliers (>3× median absolute deviation) were discarded [11].

2.2.3. Pre-Processing Pipeline

All satellite and reanalysis layers were re-projected to WGS-84 geographic coordinates. MODIS 1 km pixels were first aggregated to an intermediate 0.01° grid using an arithmetic mean of valid fine pixels, then upscaled to the common 0.083° grid (approximately 9 km) via GDAL 3.4.0. NaN values were ignored during aggregation to minimize information loss.

Temporal gaps caused by cloud cover were left unfilled to preserve data integrity for trend analysis [11]. Temporal gaps caused by cloud cover, aerosols, and other low-quality conditions were left unfilled to preserve the integrity of the data for trend analysis [9]. Spatial or temporal interpolation was avoided due to the high spatio-temporal heterogeneity of the coastal environment, which can lead to interpolation artifacts and biased estimates, especially near fronts, estuaries, and in areas with strong seasonal variability (e.g., monsoon-influenced regions). To mitigate the impact of data gaps while maintaining data authenticity, our analysis relied on (1) the robustness of the non-parametric trend tests to missing data; (2) generating long-term (2003–2020) monthly and seasonal climatologies by averaging all valid daily values within each time period across all years; and (3) using native daily data for trend detection, which provides a sufficient number of observations over the 18-year period for robust statistical analysis at most pixels.

2.3. Remote-Sensing Algorithm for Water Transparency

We adopted the QAA-v6 semi-analytical algorithm proposed by Lee et al. [19] and applied the refinement proposed for optically complex waters by Chen et al. [11,20,21]. This refinement couples the ‘clear’ (QAA_clear) and ‘turbid’ (QAA_turbid) sub-models via a logistic function to simulate a continuous transition, thereby avoiding abrupt changes at a fixed threshold.

$$C_{clear}=1/(1+e^{-k(Z_{{sd}_{clear}}-x_0)})$$

$$Z_{sd}=C_{clear}\times Z_{{sd}_{clear}}+C_{turbid}\times {Z_{sd}}_{turbid}$$

where k and x₀ are empirically tuned for China’s coastal waters using in situ SDD. The model parameters (k = 11.84 and x₀ = 0.99) were directly adopted from Chen et al. [20,21], who originally calibrated them using 106 match-up pairs in Honghu Lake, a turbid inland water body. The core hypothesis of our study is that this model structure and its parameters are robust and transferable to the optically complex coastal waters of China. Model performance improved from R² = 0.63, RMSE = 1.49 m (original QAA-v6) to R² = 0.70, RMSE = 1.24 m after bias correction (regression slope = 1.04, intercept = 0.77 m). The final daily SDD product (2003–2020, 0.083°) was bias-adjusted using the derived linear model.

We validated the improved QAA model for coastal waters. The performance of the improved QAA model with the adopted parameters (k = 11.84, x₀ = 0.99) was rigorously evaluated against the independent in situ SDD dataset. A total of 225 pairs of independent in situ validation data were used for accuracy verification. The results demonstrate the effectiveness of the improved algorithm [21]. A time window of ±24 h and a spatial window of 3 × 3 pixels (approximately 9 km²) centered on each sampling station were applied to match satellite-derived $Z_{sd}$ with in situ measurements. Accuracy was assessed using the coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE) [11].

2.4. Spatio-Temporal Analysis

2.4.1. Trend Detection

Long-term trends (2003–2020) for each habitat driver were assessed at a pixel level using the non-parametric Mann–Kendall test (significance level α = 0.05) combined with Sen’s slope estimator [22]. The null hypothesis of no monotonic trend was rejected when |Z| > 1.96.

2.4.2. Seasonal Cycle Extraction

Monthly and seasonal climatologies were computed by averaging the valid daily values within each calendar month and astronomical season over the 18-year record. Spatial composites and anomaly maps were generated in ArcGIS Pro 3.1 using bilinear interpolation for visualization purposes.

3. Results

3.1. The Annual Mean Spatial Distribution of Environmental Factor in China Sea

The results show the spatial distribution of the annual averages from 2003 to 2020 for key environmental factors in the China Sea, including SST, salinity, transparency, NPP, EV, and NV (Figure 2). SST exhibited a distinct north–south thermal gradient, ranging from 9 to 13 °C in the Bohai Sea to over 20 °C in the East China Sea (Figure 2a). Salinity varied up to no more than 34.62‰, with lower values (blue-toned) near major river mouths and higher values (green) offshore (Figure 2b). Transparency features red-dominated low values (less than 5 m) along turbid coasts such as Subei, where a distinct sediment plume forms, transitioning to high values (29.20 m) in clear offshore zones (Figure 2c). NPP displays intense green nearshore hotspots (3229.88 mg/m²) near estuaries, fading to low-NPP areas offshore (Figure 2d). Current dynamics show EV peaking at 0.69 m/s (red) along the coast and NV reaching 1.01 m/s (red) in the Kuroshio region, particularly to the east of Taiwan (Figure 2e,f). These factors collectively illustrate the complex and varied environmental conditions across the region.

3.2. Seasonal Variations in Environmental Factors in China Sea

The study reveals distinct seasonal patterns in environmental factors from 2003 to 2020 (Figure 3). SST ranges from 6.71 °C in winter to 18.92 °C in summer, with transitional values in spring (The average annual temperature is about 10.22 °C, the same below) and autumn (16.48 °C) (Figure 3a–d). Salinity peaks at 31.68‰ in winter, decreases to 30.67‰ in summer, and shows intermediate values in spring (31.66‰) and autumn (30.71‰) (Figure 3e–h). Transparency is lowest at 8.80 m in winter, rising to 12.54 m in summer, and showing intermediate values in spring (9.88 m) and autumn (11.20 m) (Figure 3i–l). NPP peaks in spring (1289.23 mg/m²), with similar values in summer (1245.89 mg/m²) and autumn (1267.89 mg/m²), and drops to 929.58 mg/m² in winter (Figure 3m–p). EV is strongest in summer (0.13 m/s) and spring (0.08 m/s), and weakest in autumn and winter (0.07 m/s) (Figure 3q–t). NV is highest in summer (0.13 m/s), followed by spring (0.09 m/s), and lowest in winter (0.08 m/s), with autumn showing a similar value (0.09 m/s) (Figure 3u–x).

Overall, SST follows the typical pattern of being highest in summer and lowest in winter, with transitional seasons in spring and autumn. Nearshore salinity decreases in summer and increases in winter, while offshore salinity remains relatively stable. Transparency is highest in summer, followed by spring and autumn, and lowest in winter. NPP is highest in spring and summer and lowest in winter. EV is strongest in spring and summer and weakest in autumn and winter, while NV is strongest in summer and weakest in winter. These factors collectively illustrate the complex seasonal dynamics across the region.

3.3. Long-Term Trends of Environmental Factors in China Sea

The study identifies long-term trends in environmental factors from 2003 to 2020: SST is largely stable, with 9.78% of areas showing significant increases and 0.43% showing significant decreases (Figure 4a). Warming was observed in the northern Yellow Sea and the southern Shandong Peninsula, while cooling was observed in the central East China Sea. Of all of the areas studied, 56.74% showed increased salinity and 43.26% showed decreased salinity, with only 1.31% showing significant decreases, mainly in the central Bohai Sea and Haizhou Bay (Figure 4b). Offshore salinity remained stable. 65.14% of areas showed increased transparency, and 34.86% decreased, with 13.09% significantly increased and 2.59% significantly decreased, indicating clearer waters overall (Figure 4c). Significant increases in NPP were observed in the nearshore Bohai Sea, the central northern Yellow Sea, the central southern Yellow Sea, and the far offshore southern East China Sea. Of the areas studied, 2.15% showed significant increases, 27.3% showed significant decreases, and 70.55% remained stable (Figure 4d). Decreases were prominent in the Yangtze River estuary, the nearshore East China Sea, the southern Liaodong Peninsula, and around Taiwan Island. EV showed significant increases in 2.09% of areas and significant decreases in 4.06%, with 93.85% remaining stable (Figure 4e). Increases were noted near the Liaodong Bay, and decreases in the Bohai Sea, southern Yellow Sea, northeast of Taiwan Island, and the Kuroshio influence area. NV increased significantly in 3.54% of areas, decreased significantly in 1.49%, and remained stable in 94.97% (Figure 4f). Increases were observed near Qinhuangdao and the continental shelf break of the East China Sea, while decreases were observed in Liaodong Bay and to the east of the Yangtze River estuary. These trends reflect a dynamic marine environment with regional variations influenced by both natural and anthropogenic factors.

Table 1. Pixel statistics for Sen’s slope estimation and Mann–Kendall significance test.

Zsd Trends	Extent
$\mathsf{\beta }<0, \left|\mathrm{Z}\right|>2.58$	Very significantly reduced
$\mathsf{\beta }<0, 1.96<\left|\mathrm{Z}\right|\le 2.58$	Significantly reduced
$\mathsf{\beta }<0, \left|\mathrm{Z}\right|\le 1.96$	Not significantly reduced
$\mathsf{\beta }>0, \left|\mathrm{Z}\right|\le 1.96$	Not significantly increased
$\mathsf{\beta }>0, 1.96<\left|\mathrm{Z}\right|\le 2.58$	Significantly increased
$\mathsf{\beta }>0, \left|\mathrm{Z}\right|>2.58$	Very significantly increased

Table 1. Pixel statistics for Sen’s slope estimation and Mann–Kendall significance test.

Zsd Trends	Extent
$\mathsf{\beta }<0, \left|\mathrm{Z}\right|>2.58$	Very significantly reduced
$\mathsf{\beta }<0, 1.96<\left|\mathrm{Z}\right|\le 2.58$	Significantly reduced
$\mathsf{\beta }<0, \left|\mathrm{Z}\right|\le 1.96$	Not significantly reduced
$\mathsf{\beta }>0, \left|\mathrm{Z}\right|\le 1.96$	Not significantly increased
$\mathsf{\beta }>0, 1.96<\left|\mathrm{Z}\right|\le 2.58$	Significantly increased
$\mathsf{\beta }>0, \left|\mathrm{Z}\right|>2.58$	Very significantly increased

3.4. Interannual Variations in Environmental Factors in China Sea

This study examines the interannual changes in environmental factors across the China Sea, from 2003 to 2020. The East China Sea has the highest average SST (15.01 °C), followed by the Bohai Sea (10.90 °C) and the Yellow Sea (10.69 °C) (Figure 5a). Notably, both the East China Sea and the Yellow Sea exhibit a “decline then rise” pattern, with SST dropping from 2003 to 2011 and then gradually increasing. In contrast, the Bohai Sea shows a fluctuating upward trend (rate of decline 2003–2011 (r = 0.11) vs. rate of increase after 2011 (r = 0.06)), despite significant interannual variability (Figure 5a). Over the period 2003–2020, the East China Sea has a significantly higher average salinity (32.06‰) compared to the Yellow Sea (29.47‰) and the Bohai Sea (25.61‰) (Figure 5b). Transparency is also highest in the East China Sea (17.06 m), significantly above the Yellow Sea (7.06 m) and the Bohai Sea (3.99 m) (Figure 5c). NPP shows an inverse spatial pattern, with the Bohai Sea having the highest average (2033.82 mg/m²), followed by the Yellow Sea (1397.68 mg/m²), and the East China Sea having the lowest (923.17 mg/m²) (Figure 5d). The East China Sea has significantly higher current velocities in both east–west (0.13 m/s) and north–south (0.15 m/s) directions compared to the Yellow Sea (0.04 m/s and 0.03 m/s) and the Bohai Sea (0.03 m/s and 0.02 m/s) (Figure 5e,f). The East China Sea’s velocities exceed the study area’s average, while those of the Yellow Sea and the Bohai Sea are below the mean. Overall, while there are interannual fluctuations in these environmental indicators, the overall changes are relatively stable during the study period.

3.5. Correlation Among Coastal Environmental Factors

To explore the relationships among environmental factors, this study further analyzed the correlations between these factors and their long-term trends (

Table 2. The correlation between environmental factors.

	SST	Salinity	Transparency	NPP	EV	NV
SST	1	0.95	0.76	0.62	0.64	0.56
Salinity	0.95	1	0.71	0.68	0.57	0.49
Transparency	0.76	0.71	1	0.07	0.65	0.58
NPP	0.62	0.68	0.07	1	0.17	0.12
EV	0.64	0.57	0.65	0.17	1	0.76
NV	0.56	0.49	0.58	0.12	0.76	1

and

Table 3. The correlation between long-term trends in environmental factors.

	SST	Salinity	Transparency	NPP	EV	NV
SST	1	−0.26	0.17	0.05	0.12	0.08
Salinity	−0.26	1	0.2	−0.36	−0.13	0
Transparency	0.17	0.2	1	−0.6	−0.05	−0.07
NPP	0.05	−0.36	−0.6	1	−0.09	0.04
EV	0.12	−0.13	−0.05	−0.09	1	0.16
NV	0.08	0	−0.07	0.04	0.16	1

). The results indicate varying degrees of correlation among the environmental factors. Specifically, SST shows a strong positive correlation with salinity (r = 0.95) and moderate positive correlations with transparency (r = 0.76), NPP (r = 0.62), EV (r = 0.64), and NV (r = 0.56). Salinity also exhibits moderate positive correlations with transparency (r = 0.71), NPP (r = 0.68), and EV (r = 0.57), but a weaker correlation with NV (r = 0.49). Additionally, transparency has a very weak correlation with NPP (r = 0.07), yet moderate positive correlations with EV (r = 0.65) and NV (r = 0.58). The correlations between NPP and the components of current velocity are also weak (EV: r = 0.17, NV: r = 0.12), while there is a moderate positive correlation between EV and NV (r = 0.76) (Table 2). The correlations of long-term trend changes among environmental factors indicate that, aside from a moderate negative correlation between transparency and NPP (r = −0.6), the correlations among other environmental factors in terms of long-term trend changes are generally weak (r < 0.5) (Table 3). This suggests that although environmental factors exhibit coordinated spatial variations, their correlations in terms of long-term evolutionary trends are significantly different.

4. Discussion

This study provides the first comprehensive, high-resolution assessment of six key habitat drivers in China’s coastal waters over an 18-year period (2003–2020), addressing critical gaps in EBFM. This advances beyond previous studies which were often limited to single drivers, shorter time frames, or coarser spatial scales. Building on technological advances in remote sensing and reanalysis products, our findings reveal intricate spatio-temporal patterns that refine our understanding of coastal ecosystem dynamics. Below, we interpret these results in the context of prior research, discuss implications for sustainable management under the UN Sustainable Development Goals (SDGs), and outline future research directions.

4.1. Interpretation of Key Findings in Relation to Previous Studies

Our 18-year, high-resolution (0.083°) assessment provides a new perspective on key ecological and environmental factors in the China Sea, building on and extending earlier findings derived from localized or coarser-scale studies. The pronounced spatial gradients, such as the north–south thermocline in SST, estuarine salinity minima, and coast-to-offshore transparency transitions, align with existing regional observations, but also reveal systematic patterns that were previously undetectable at this spatial scale. For example, the low transparency (<5 m) at Subei Shoal and the high clarity (>29 m) offshore validate and generalize the algorithmic refinements by Chen et al. [23] for optically complex waters. This confirms the robustness of our improved QAA-v6 model (R² = 0.70, RMSE = 1.24 m).

Seasonal dynamics also reflect the well-documented phenology of phytoplankton [9], with SST and NPP peaks in summer and spring, respectively. However, our data reveal notable stability in offshore salinity, contrasting with estuarine seasonality, and highlight current velocity maxima in summer. This underscores the role of monsoon forcing in the Kuroshio-influenced East China Sea [24].

Long-term trends show clear regional differences and are influenced by multiple factors. While the northern Yellow Sea warmed significantly, the central East China Sea cooled—a pattern consistent with the negative phase of the Pacific Decadal Oscillation since 2008 [25,26,27] and enhanced Kuroshio-mediated upwelling [28,29]. The Bohai Sea sustained the highest NPP (2033.82 ± 189 mg/m²), likely due to intensified nutrient inputs from the Yellow River. Conversely, NPP near the Yangtze Estuary declined by 27.3%, exceeding earlier estimates [30] and reflecting reduced sediment and nutrient fluxes from upstream dam retention [31]. This decline aligns with documented reductions in fishery biomass, suggesting bottom-up control of the ecosystem [32,33].

Transparency increased significantly across 13.09% of the study area, which is a trend three times higher than previous estimates [34,35]. This increase is likely due to large-scale sediment management, such as the operations of the Three Gorges Dam, although the impact of changes in the climate of winds and waves and in river discharge cannot be ruled out [36,37,38].

Finally, correlation analyses confirm strong spatial covariation among drivers (e.g., SST–salinity, r = 0.95), but weak coupling in long-term trends (most |r| <0.5). This decoupling highlights the heterogeneous responses of environmental drivers to localized forcings, challenging the assumption of uniform ecosystem change and emphasizing the need for regionally specific management strategies.

4.2. Implications for Ecosystem-Based Fisheries Management and Policy

For ecosystem-based management, the observed spatiotemporal trend implies that short-term actions, such as seasonal fishing closures, can exploit the strong spatiotemporal co-variation patterns evident in Table 2 to identify current habitat hotspots. However, long-term strategies must recognize that the underlying drivers are evolving independently. Therefore, they require parameter-specific thresholds, with particular vigilance in decoupling hotspots such as the Yangtze Estuary. Consequently, regional management systems should incorporate real-time trend tracking for all six drivers in order to anticipate and respond to potential future shifts, rather than assuming synchronized change.

Our spatially explicit findings directly support SDG 14 (“Life below water”) by enabling dynamic habitat management. For example, the identification of transparency “hotspots” in the Bohai Sea could guide real-time closures for sensitive species like Chinese sturgeon during critical life stages. Similarly, declining NPP trends in the Yangtze River estuary (27.3% significant decreases) warrant adaptive zoning of marine protected areas to buffer eutrophication impacts, aligning with China’s “Ecological Redline” policy [39]. Seasonal current velocity maps could enhance early-warning systems for habitat degradation, such as algal blooms triggered by summer stratification.

Under SDG 13 (“Climate action”), the regional warming/cooling dichotomy underscores the need for localized adaptation strategies. Warming in the Yellow Sea may necessitate thermal refuge designations, while cooling in the East China Sea could inform climate-resilient fishery quotas [40]. The weak trend correlations highlight that EBFM must account for driver-specific responses, avoiding one-size-fits-all approaches.

4.3. Limitations and Uncertainties

Despite advances, our study inherits limitations from remote-sensing data. Cloud-induced temporal gaps (~15% of daily records) may bias trend estimates, though our unfilled-gap approach preserved integrity. A key uncertainty stems from cloud-induced data gaps, which are estimated to have affected 15–20% of potential daily observations. Although our trend analysis method is robust in the face of random missing data, the uneven seasonal distribution of clouds—with a higher frequency during the summer monsoon—could potentially introduce bias if environmental conditions during cloudy days are non-random (e.g., if they are associated with higher wind speeds and sediment resuspension). If the probability of obtaining a valid satellite observation is systematically lower during certain seasons or under specific environmental conditions (e.g., high turbidity events coinciding with cloud cover), this could theoretically introduce bias into long-term trend estimates. However, we assume that such missingness is largely random over the 18-year timeframe. Future studies could benefit from developing more sophisticated gap-filling models tailored specifically to the complex dynamics of coastal waters, in order to address this potential issue more effectively.

Algorithm uncertainties persist in highly turbid zones, where SDD retrievals showed residual errors (RMSE = 1.24 m). Although our datasets’ 0.083° resolution is ideal for identifying large-scale patterns and trends, it is insufficient for resolving the fine-scale processes that are critical to fisheries management in nearshore environments. Sub-mesoscale dynamics, such as localized eddies and fine-grained frontal features within estuaries and bays, are not captured. Consequently, our findings are most applicable to broad-scale habitat suitability and environmental change assessments. Future efforts should integrate higher-resolution sensors (e.g., Sentinel-2) [41,42], validate these with expanded in situ networks, and develop management strategies for critical habitats such as estuarine nurseries or migratory corridors. Future assessments would benefit from the inclusion of additional drivers, such as dissolved oxygen and pH, as reliable, high-resolution data products for these variables to become available.

4.4. Future Research Directions

Building on this foundation, we recommend the following: (1) incorporating biotic variables (e.g., fish abundance data) to model species–habitat relationships dynamically; (2) extending analyses to include acidification and deoxygenation drivers for holistic SDG assessments; and (3) developing machine-learning frameworks to predict habitat shifts under IPCC scenarios, enhancing proactive EBFM. Cross-border collaborations could expand coverage to the South China Sea, addressing transboundary management challenges.

In summary, this 18-year high-resolution of environmental factors marks a paradigm shift in coastal monitoring, providing the empirical backbone for spatially adaptive conservation. By bridging technological innovations with policy frameworks, we advance toward sustainable ocean governance in China’s rapidly evolving seascape.

5. Conclusions

This study leverages high-resolution MODIS-Aqua and CMEMS GLORYS12V1 data to assess six key environmental factors in the China Sea from 2003 to 2020. It reveals significant spatial and seasonal gradients, with SST ranging from 9 to 13 °C in the Bohai Sea to over 20 °C in the East China Sea, and NPP peaking at 2033.82 mg/m² in the Bohai Sea. Seasonal dynamics show summer peaks in SST and transparency, and spring peaks in NPP. Long-term trends indicate regional warming in the northern Yellow Sea and cooling in the central East China Sea, with overall improved transparency—possibly linked to changes in sediment supply, although direct evidence is still lacking—and declining NPP in some estuaries. Correlation analysis shows strong spatial coupling but weak trend coupling, except for a moderate negative correlation between transparency and NPP.

These findings inform the zoning of marine protected areas, optimize fishing quotas based on SST trends, and enhance algal bloom warning systems. Limitations include data gaps from cloud cover and unresolved sub-mesoscale processes. Future work should integrate multi-source satellites and artificial intelligence techniques, and couple biological data to build dynamic species–habitat models. This 18-year assessment lays an empirical foundation for sustainable coastal governance, transitioning China’s nearshore areas to climate-adaptive management by integrating remote sensing into EBFM, balancing ecological integrity with socio-economic needs.