Abstract: Projecting vegetation dynamics in the Yellow River water conservation area under future climate change is essential for guiding ecological conservation and regional sustainable development. Utilizing MODIS NDVI data from 2000 to 2020, we examined the spatiotemporal variation of vegetation cover by integrating Theil-Sen trend analysis and the Mann-Kendall test. We further quantified the bidirectional causal relationships between NDVI changes and key climatic drivers using the geographic convergent cross mapping model, and projected the NDVI changes under multiple scenarios via the GeoSOS-FLUS model. The results showed that: (1) During 2000-2020, NDVI exhibited a significant fluctuating upward trend across the region, with an average growth rate of 0.025 per decade. Spatially, NDVI patterns were heterogeneous, and the extent of vegetation improvement substantially exceeded that of degradation. (2) NDVI variations showed the strongest causal relationships with temperature and precipitation. Specifically, temperature and precipitation exerted strong influences on NDVI, while NDVI had a notable feedback on precipitation. (3) Scenario-based projections to 2030 revealed that under the Natural Development Scenario and SSP2-4.5, vegetation improvement would be relatively pronounced, with combined areas of moderately high and high vegetation coverage expanding by 1.31 ×10⁴ km² and 1.29×10⁴ km² respectively, compared to that in 2020. In contrast, under the Economic Development Scenario and SSP5-8.5, vegetation improvement would be relatively small, with an expansion of only 0.94×10⁴ km² for both scenarios. The most significant vegetation improvement was observed under the Ecological Protection Scenario, where the corresponding area increased by 2.0×10⁴ km². These findings provide a scientific basis for ecological restoration strategies in the Yellow River water conservation area. Region-specific management measures are proposed accordingly.

Key words: NDVI, geographic convergent cross mapping model, GeoSOS-FLUS model, multi-scenario prediction, Yellow River water source conservation area