Artificial oasis regions in arid northwestern China are highly sensitive to landscape disturbance and ecological degradation, yet long-term assessments of their ecological risks remain limited. In this study, 259 Landsat images from 1990 to 2019 were synthesized using an annual maximum NDVI composite approach to overcome cloud contamination and phenological inconsistencies, enabling the construction of a time-continuous and high-quality land-use dataset. An object-oriented classification method and a landscape ecological risk index were used to quantify the spatiotemporal patterns of ecological risks, while the CA-Markov model was applied to simulate future risk scenarios for 2050 and 2080. Results showed that the overall ecological risk in the Alar Reclamation Area decreased significantly over the past three decades, driven primarily by the conversion of fragmented unused land into cohesive cultivated land, which enhanced vegetation cover, landscape connectivity, and ecological stability. Significant clusters of high-risk areas were concentrated in the northwest and along the Tarim River in 1990, but these clusters gradually shrank as reclamation progressed. A clear ecological risk mutation point occurred around 1995, marking the transition from unstable to more resilient ecosystem conditions. Future projections, however, indicate that under continued land reclamation and water-resource pressure, ecological risks will rise again by 2080, with high-risk zones expanding outward from reclaimed cropland. These findings highlight the nonlinear response of artificial oasis ecosystems-where moderate reclamation initially improves ecological security, but excessive expansion surpasses environmental carrying capacity and increases risk. The study provides a new long-term, data-rich framework for ecological risk assessment in oasis systems and offers guidance for sustainable land management in fragile arid regions.