Analysis of Future Green Space Changes in Cities and Counties of Chungcheongbuk-do Using AI-based Machine Learning

This study analyzes changes in green space area across 11 cities and counties in Chungcheongbuk-do and develops future projection data under Shared Socioeconomic Pathways (SSP)1-2.6and SSP5-8.5 climate scenarios using machine learning and explainable artificial intelligence techniques.Ordinary least squares (OLS) regression, Random Forest, and XGBoost models were employed to analyze and predict green space change, while shapely additive explanations-based explainable artificial intelligence was applied to evaluate the relative importance of explanatory variables.The results indicate that green space changes are explained more strongly by population density and land-use structure than by climate variables.Compared with the OLS model, both Random Forest and XGBoost achieved substantially higher predictive performance and consistently identified population density and land-use variables as major determinants of future green space dynamics.Future scenario analysis further revealed considerable spatial heterogeneity among cities and counties, with several local governments showing relatively larger reductions in green space under the SSP5-8.5 scenario.The generated data provide historical trends and future projections of green space dynamics at the local government level and can support future studies on climate change adaptation, urban planning, environmental management, and sustainable land-use policy.By providing scientifically validated projection data and interpretable evidence on the structural drivers of green space change, this study contributes to the expansion of reusable geospatial information for research and policy applications.