Beyond green: Unveiling the drivers of gross ecosystem product in high-density urban green spaces — A case study of central Beijing

Rapid population growth and intensive urbanization continue to threaten the supply of ecosystem services (ES) in cities. In high-density urban areas, the planning and management of urban green spaces (UGS) under complex landuse constraints are vital for sustainable urban renewal. Gross ecosystem product (GEP) provides a practical framework for valuing ES and guiding sustainability policies. However, UGS are often fragmented by buildings and roads into isolated ecological units, while existing studies remain largely at macro or administrative scales, overlooking patch-scale GEP variations. As a bridge between internal ecological processes and urban spatial patterns, the patch scale is essential for refined and comparable GEP accounting. This study developed a patch-scale framework to enhance the ecological value of UGS in dense cities. By integrating multisource semantic data, an object-based approach, and random forest modeling, we quantified GEP and identified the driving mechanisms across UGS types in central Beijing. In 2023, total GEP reached 160.86 billion CNY, averaging 143.27 CNY/m²; parks showed the highest values, followed by subsidiary and other green spaces. The magnitude and direction of influencing factors varied across UGS types: management dominated GEP in parks, while vegetation features played major roles in subsidiary and other green spaces. The variables year and greenspace components exhibited opposite marginal effects between parks and the other two types. This study highlights the patch scale as the smallest unit for policy implementation and spatial regulation, providing quantitative and managerial insights for optimizing green spaces and advancing standardized, comparable urban ecosystem service assessments.