The article discusses the development of a deep learning transport emulator to improve air quality forecasting. It is noted that poor air quality negatively impacts human health, and with recent increases in extreme air quality events, a need for finer resolution air quality forecasting is seen. The current model of the National Oceanic and Atmospheric Administration (NOAA) is based on a 15km spatial resolution, but the goal is to reach a 3km resolution. The authors propose a deep learning method that reduces computations while maintaining comparable skill with the existing model. The method shows potential for operational use, especially in the face of extreme air quality events.
Publication date: 22 Feb 2024
Project Page: https://arxiv.org/abs/2402.14806
Paper: https://arxiv.org/pdf/2402.14806
