Spatial Heterogeneity of Human–Land Interactions Under Coastal Development in a Port Region: Implications for Ecological Sustainability

Coastal development reshapes land cover, human activity, and ecological quality, yet a higher coordination metric does not necessarily denote greater ecological sustainability. Taking Qingdao West Coast New Area as a case study, we integrated remote-sensing and land-cover data for 2005, 2009, 2015, 2019, and 2025 on a common 500 m grid. Ecological quality was represented by the Remote Sensing Ecological Index (RSEI), human activity by the Composite Nighttime Light Index (CNLI), and their relative matching by the coupling coordination degree (CCD); multiscale geographically weighted regression (MGWR) was used to characterize spatial associations and variable-specific scales. From 2005 to 2025, 183.015 km2 of cropland was converted to impervious surface, while mean CNLI increased from 0.0081 to 0.0243. Mean RSEI declined from 0.6526 to 0.5256 by 2019 and was 0.5645 in 2025. Across 8228 fixed common-observation grids, including structural zeros, mean CCD decreased from 0.0851 in 2005 to 0.0808 in 2009 and then increased to 0.1775 in 2025, indicating that the later increase was not solely attributable to changes in sample coverage. MGWR showed that the spatial scales of association varied among years and variables. The mean coefficient for cropland proportion was negative in both the main and NTL-only specifications, although its magnitude was sensitive to how human activity was constructed. Because annual PCA loadings varied and 2024 NTL was used as a near-term proxy for the 2025 assessment, cross-year magnitudes and end-period results require cautious interpretation. The findings show that CCD should be interpreted jointly with the direction of ecological-quality change, structural zeros, and land function, whereas MGWR provides evidence on spatial associations and scale differences.