Simulation-assisted machine learning for operational digital twins

In the environmental sciences, there are ongoing efforts to combine multiple models assist analysis of complex systems. Combining process-based models, which have encoded domain knowledge, with machine learning can flexibly adapt input data, improve modeling capabilities. However, both types data limitations. We propose a methodology overcome these issues by using model generate aggregating them lower resolution mimic real situations, and developing fraction inputs. showcase this method case study pasture nitrogen response rate prediction. train different scales test in sampled unsampled location experiments assess their practicality terms accuracy generalization. The resulting provide accurate predictions generalize well, showing usefulness proposed for tactical decision support.