This work addresses a significant challenge in atmospheric science, specifically the computational constraints faced by large-scale models. With thousands of chemical compounds interacting and outputs influenced by numerous environmental parameters, current air quality and climate models often rely on simplified chemistry schemes. AI holds promise in enhancing our comprehension of atmospheric chemistry.
Surrogate AI for Chemistry
Modelling atmospheric chemistry is complex and computationally intense. We propose ChemNNE, an Attention-based Neural Network Emulator that can model atmospheric chemistry as a neural ODE process, using sinusoidal time embeddings and the Fourier neural operator. Details: arxiv.org/abs/2408.01829.
AI for chemical modelling
Large-scale air pollution models are computationally intensive due to numerically solving stiff systems of differential equations. Replacing numerical solvers with neural network-based solvers can significantly reduce costs. Our work focuses on estimating reaction parameters with neural networks for stiff chemical kinetics.
AI for air pollution
Air pollution modeling traditionally relies on complex differential equations that are computationally expensive. Neural networks reduce computational costs while improving prediction accuracy. Our research targets predicting pollutant concentrations across regions and estimating chemical reaction parameters despite limited experimental data.