Spatiotemporal evolution and drivers of carbon stock in arid oases: Multi-source assessment and scenario prediction in Wei-Ku oasis, China

Accurately assessing carbon stock (CS) in arid oasis ecosystems is vital for the global carbon cycle and regional "dual carbon" goals. This study integrates field surveys, satellite imagery, climate data, machine learning, In VEST model, and the optimal parameters-based geographical detector model to analyze the Wei-Ku Oasis. The results indicate that: (1) carbon density ranked as follows: garden land (27,00.62 gm −2 ) > forest-grassland (2,386.31 gm −2 ) > arable land (1,940.43 gm −2 ) > construction land (987.93 gm −2 ) > unused land (869.86 gm −2 ) > water bodies. (2) Total CS fluctuated from 2010 to 2022, with a net increase of 1.301 × 10 6 t. By 2028, only the ecological protection scenario will increase CS. (3) Vegetation cover ( q ¯ = 0.356) is the dominant driver of CS spatial heterogeneity, with interactions among factors enhancing explanatory power. This study provides an integrated assessment framework and scientific evidence for land use optimization and ecological strategy development in arid regions.