Ecosystem service assessment provides a critical basis for optimizing regional ecological management and promoting sustainable development. From the water-land-carbon coupling perspective, this study established a technical framework for quantifying individual services, coupling a composite index, and analyzing multidimensional driving mechanisms. The InVEST model was applied to quantify three core ecosystem services: water yield, habitat quality, and carbon storage. A Composite Ecosystem Service Index (CESI) was constructed through normalization and weighted summation. Multidimensional driving factors were identified using the Optimal Parameter-Based Geographical Detector. Taking Ningxia during 2004-2024 as the study area, the results showed that the CESI exhibited a fluctuating upward trend with significant spatial heterogeneity, characterized by a south-high and north-low pattern. Land use transitions were dominated by bidirectional conversions between cropland and grassland, while impervious area expanded rapidly and barren land decreased overall. The spatial differentiation of CESI was jointly controlled by natural and anthropogenic factors, with land use type, precipitation, and digital elevation model showing the strongest explanatory power, and all two-factor interactions displaying pronounced enhancement effects. This study provides a reproducible framework for ecosystem service assessment in arid and semi-arid regions, supporting ecological restoration, land use optimization, and the coordinated development of ecology and economy under water-land-carbon synergy.