Vegetation dynamics along an urban–rural gradient in the Yangtze River Delta: patterns, mechanisms, and scenario projections using a mixed-cell cellular automata (MCCA) framework

Understanding how the urban–rural gradient modulates vegetation dynamics under climate change and rapid urbanization is crucial for precise ecological management. However, existing research often uses coarse gradient classifications, while prevailing projection models are either limited by a lack of geospatial process simulation or constrained by the discrete-cell assumption, hindering their ability to capture the continuous nature of vegetation change. To address these challenges, this study adopts a six-zone urban–rural gradient classification (Natural—Agricultural—Rural—Suburban—Urban Expansion—Urban Core). Based on a systematic investigation of the spatiotemporal evolution and driving mechanisms of NDVI in the Yangtze River Delta, the mixed-cell cellular automata (MCCA) framework is migrated and extended from discrete land use simulation to continuous NDVI simulation. For this purpose, a development probability calculation strategy that couples historical trends with the natural background is employed to project future NDVI trends. The results indicate that: (1) From 2000 to 2022, the overall NDVI in the Yangtze River Delta increased but exhibited a complex “U-shaped” spatial pattern along the urban–rural gradient, characterized by significant greening in the urban core and severe degradation in the urban expansion and suburban zones. Notably, the rural zone constituted the most extensive hotspot of vegetation degradation. (2) This pattern is shaped by a systematic shift in dominant drivers along the gradient and the dual role of socioeconomic factors, with the latter being closely linked to the planning orientations and land tenure systems in each gradient zone. (3) The adaptively modified MCCA model demonstrates its applicability, outperforming mainstream machine learning methods in both numerical accuracy and trend consistency, and is shown to be suitable for medium-to-short-term (approx. 10–12 years in this study) numerical projections and long-term trend analysis. (4) Future scenario projections for 2023–2035 show that, compared to the unstable SSP5–8.5 pathway, the SSP2–4.5 pathway, despite higher localized degradation risks, promises more robust ecological recovery, better aligning with the region's sustainability goals. This study not only deepens the understanding of vegetation dynamics in complex urban–rural systems but also validates the feasibility of the MCCA framework as an effective tool for NDVI simulation, offering a new perspective for the predictive simulation of medium-to-short-term vegetation NDVI. The findings provide critical scientific support for formulating spatially differentiated and precise ecological governance strategies in rapidly urbanizing regions.