关键词: 数据-模型整合, 全球变化, 控制实验, 陆面过程模型, 生态系统碳循环, 模型不确定性

Abstract: Anthropogenic global change profoundly affects terrestrial ecosystem structure and function, creating an urgent and persistent need to accurately predict future ecosystem states. Field-based manipulative experiments provide critical mechanistic insights into these impacts but are inherently limited in spatio-temporal scope. Conversely, process-based models can extrapolate to broader scales but often contain simplified or unrealistic mechanisms that lead to uncertain projections. Data-model integration has emerged as an essential approach to bridging this gap, testing model assumptions against empirical evidence and guiding experimental design via model-based hypotheses. This review synthesized progress in integrating manipulative experiments with process-based models across three key global change drivers: elevated CO₂, climate change (warming and altered rainfall) and nutrient manipulation. We demonstrated how this integration reduced key uncertainties in processes such as photosynthesis, carbon-nutrient coupling and soil biogeochemistry, whilst exposing persistent gaps in plant hydraulics, microbial dynamics and multifactorial stresses. These advances were most pronounced in representing CO₂ fertilization effects, including improved stomatal optimization theory, dynamic carbon allocation schemes and coupled carbon-nitrogen-phosphorus cycling. By contrast, its application to warming, rainfall change and multi-nutrient interactions remained underdeveloped. To catalyze future progress, we propose specific strategies to foster a more synergistic cycle of knowledge co-production. These include prioritizing the quantification of mechanism-specific data to develop dynamic model formulations, systematically using multi-site experimental networks to benchmark and refine model processes across scales, and strategically employing models to design targeted experiments. Ultimately, these strategies are indispensable for developing more realistic models and achieving predictive understanding of ecosystem responses to global change.