Solar radiation (SRA) is a key driver of photosynthesis, and photosynthetic light-response processes have been extensively studied at the leaf scale, providing fundamental insights into plant carbon assimilation and photoinhibition mechanisms. However, studies of these processes at the ecosystem scale remain limited. In this study, we innovatively applied a modified rectangular hyperbola model, originally developed for leaf-scale light responses, which incorporates a photoinhibition term to capture photosynthetic downregulation at high irradiance, to eddy-covariance flux data. Based on the FLUXNET2015 dataset, we examined the light-response of net ecosystem exchange of CO2 across biomes and showed that photoinhibition is a common phenomenon in terrestrial ecosystems. At the spatial scale, the light saturation point (Isat) ranged from 963 to 1475 mu mol photons m- 2 s-1, with the highest values observed in savannas, followed by forests. Random Forest analysis revealed that Isat was primarily driven by SRA, reflecting the acclimation of ecosystem photosynthesis to light. In contrast, the photoinhibition slope (Sph) was significantly higher in forests, with differences up to 46.2%, and it showed a stronger positive correlation with leaf area index than in other biomes. Both Isat and Sph exhibited clear seasonal dynamics across biomes, indicating that photoinhibition was concentrated in spring and reached its peak in summer. Over the long term, Isat significantly increased at a rate of 0.45-1.27% yr- 1, a trend likely associated with CO2 fertilization. Meanwhile, Sph decreased significantly in savannas and shrublands, implying improved photoprotection in arid regions. Overall, our study stresses that Earth system models should incorporate photoinhibition processes to improve predictions of ecosystem carbon fluxes under future climate scenarios.