Urban expansion and intensive land use have profoundly reshaped landscape patterns, imposing increasing pressure on regional ecological security. To address this challenge, this study couples ecosystem service (ES) assessment with machine learning–based resistance surface modelling to construct the Ecological Security Pattern (ESP) of Shaanxi Province. Using the Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) model, four key ecosystem services (ESs)—water yield, soil conservation, carbon storage, and habitat quality—were quantified to identify ecological sources. Subsequently, a multidimensional resistance factor system encompassing natural, climatic, and anthropogenic drivers was established, and resistance surfaces were constructed and compared using logistic regression, support vector machine, random forest, and XGBoost models. Furthermore, circuit theory was applied to extract ecological corridors and ecological nodes, thereby forming a comprehensive ESP. The results demonstrate that the random forest model achieved the highest performance across multiple accuracy metrics, significantly improving the accuracy and robustness of resistance surface estimation. The spatial distribution of ES importance in Shaanxi follows a “high in the south—moderate in the north—low in the central region” pattern, with the Qinba Mountains serving as a critical ecological barrier. The total ecological source area amounts to approximately 33,600 km 2 , of which core sources account for 47.9%. A total of 147 ecological corridors were identified, with a combined length of 3753 km, characterised by longitudinal connectivity (north-south) yet central obstruction. Ecological pinch points were primarily associated with forest ecosystems, while barrier points were concentrated in areas of farmland expansion and urban development. Based on these findings, this study proposes a “Two Axes, Three Corridors, Six Cores, Four Functional Areas” framework for the ESP, aiming to enhance regional ecological connectivity and system resilience. Overall, the combination of ESs with machine learning approaches proves effective for the scientific construction of ESP, providing critical support for land-use planning, ecological redline delineation, and the sustainable development of ecologically fragile areas. • Coupling ecosystem services with machine learning constructs objective ecological security patterns by eliminating subjective weighting bias. • Random Forest outperforms comparative models in resistance surface estimation by accurately quantifying non-linear ecological driver interactions. • The highly urbanized Guanzhong Plain is identified as a critical high-resistance barrier obstructing longitudinal landscape connectivity. • A “Two Axes, Three Corridors, Six Cores, Four Functional Areas” framework is proposed to optimize regional connectivity and enhance ecosystem resilience.
Qu et al. (Sun,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: