讲座内容简介 | The time consumption for air quality predictions is severely constrained by the high computational demands of chemical transport models (CTM) for simulating complex atmospheric reactions. With the rise of AI-based weather prediction models, it has been demonstrated that AI models derived solely from reanalysis data can be highly accurate and efficient. However, the realm of atmospheric chemistry presents greater complexity in terms of dimensions and interactions compared to weather models. To address these challenges, we have developed an AI-based air quality model that incorporates concentrations, emissions, and meteorological factors. This model delivers accurate predictions within a trivial fraction of the time required by traditional CTM models. It then enables rapid large-scale ensemble predictions and emissions inversions, enhancing the feasibility of comprehensive and timely air quality assessments. |
