Decoding Socio-Ecological Dynamics for Urban Resilience: A 30-Year Study of Ecosystem Health and Its Drivers in the Guanzhong–Tianshui Economic Zone, China

Densely populated urban regions are dual focal points of vulnerability and innovation, where socio-ecological dynamics fundamentally shape regional resilience to global environmental transformation. To decipher this dynamic process, this study adopts ecosystem health (EH) as the core lens to conduct a 30-year (1990–2020) empirical analysis of China's Guanzhong–Tianshui Economic Zone (GTEZ) — a region serving as both a key corridor of the Belt and Road Initiative and a typical area of urban expansion. Its spatial structure, bordered by the ecologically sensitive Loess Plateau to the north, sheltered by the Qinling Mountains ecological barrier to the south, and containing the densely populated Guanzhong Plain in the center, makes it an ideal case for investigating the response mechanisms of human-environment systems. The study period spans three critical transformative phases: rapid industrialization, the gradual establishment of an environmental regulatory framework, and the widespread awakening of ecological conservation awareness.

This research integrates multi-source remote sensing and statistical data within a “Vigor–Organization–Elasticity–Services” assessment framework to systematically characterize the spatiotemporal evolution of EH. It further synthesizes natural drivers (temperature, precipitation, downward longwave radiation) and anthropogenic drivers (PM₂.₅, population density) to reveal the underlying mechanisms. By comparing multiple machine learning models, the CatBoost model with superior performance was selected and combined with the SHAP method for attribution analysis. The main findings are: (1) EH changes followed a clear “deterioration-to-improvement” trajectory. The initial decline was linked to rapid industrialization and a lack of ecological protection, while subsequent improvement benefited from the refinement of environmental regulations and increased public ecological awareness. (2) The dominant drivers shifted significantly from socio-economic factors to natural factors, indicating that after initial containment of anthropogenic pressures, the influence of natural processes like climate change on regional environmental health has become increasingly prominent. (3) EH exhibited significant spatial heterogeneity, with high-value areas consistently distributed in the southern ecological barrier zone, while low-value areas were concentrated in the western and central basin regions, reflecting a spatial gradient of human disturbance intensity.

By employing explainable artificial intelligence methods, this study deepens the understanding of the dynamics within complex urban socio-ecological systems and provides a methodological reference for related monitoring and modeling research. The results not only offer a scientific basis for climate-adaptive spatial planning and ecological risk management in similar urbanizing regions but also help identify key intervention points for resilience building. Ultimately, this research provides empirical insights into how cities and their surrounding areas can proactively adapt to and shape sustainable socio-environmental transformation pathways through collaborative governance and systematic planning. It contributes to translating global sustainable development goals into localized, actionable implementation strategies and offers context-specific guidance for coordinating development and conservation in comparable regions.