Uncertainty Quantification in Reduced‐Order Gas‐Phase Atmospheric Chemistry Modeling Using Ensemble SINDy

• Published in: Journal of geophysical research. Machine learning and computation Vol. 1; no. 4
• Main Authors: Guo, Lin, Yang, Xiaokai, Zheng, Zhonghua, Riemer, Nicole, Tessum, Christopher W.
• Format: Journal Article
• Language: English
• Published: Wiley 01.12.2024
• ISSN: 2993-5210; 2993-5210
• DOI: 10.1029/2024JH000358

Uncertainty quantification during atmospheric chemistry modeling is computationally expensive as it typically requires a large number of simulations using complex models. As such, large‐scale modeling is typically performed with simplified chemical mechanisms for computational tractability. Here, we describe a probabilistic surrogate modeling method using principal components analysis (PCA) and Ensemble Sparse Identification of Nonlinear Dynamics (E‐SINDy) to both automatically simplify a gas‐phase chemistry mechanism and to quantify the uncertainty introduced when doing so. We demonstrate the application of this method on a small photochemical box model for ozone formation. With 100 ensemble members, the calibration R $R$‐squared value is 0.86 among the three latent species on average and 0.98 for ozone, demonstrating that predicted model uncertainty aligns well with actual model error. In addition to uncertainty quantification, this probabilistic method also improves accuracy as compared to an equivalent deterministic version, by ∼ ${\sim} $ 64% for the ensemble prediction mean or ∼ ${\sim} $ 45% for deterministic prediction by the best‐performing single ensemble member. Overall, the ozone testing root mean square error (RMSE) is 20% of its root mean square (RMS) concentration. Although our probabilistic ensemble simulation ends up being slower than the reference model it emulates, we hypothesize that use of a more complex reference model in future work will result in additional opportunities for acceleration. Versions of this approach applied to full‐scale chemical mechanisms may result in improved uncertainty quantification in models of atmospheric composition, leading to enhanced atmospheric understanding and improved support for air quality control and regulation. Plain Language Summary To quantify the uncertainty that originates from simplifying complex atmospheric gas phase chemical mechanisms, we apply a probabilistic machine‐learning framework (E‐SINDy) to build a surrogate model that consists of multiple models trained with different subsets of data and possible equation terms. As demonstrated on a simple photochemical mechanism, this method can effectively and reliably quantify the uncertainty in its predictions and shows promise toward scaling to more complicated atmospheric models. Compared to an equivalent deterministic approach, E‐SINDy is not only more robust but also more accurate when predicting the levels of various substances in the atmosphere under different environmental conditions. With a full‐scale reference mechanism, this method could greatly improve uncertainty quantification in atmospheric modeling, enhancing scientific ability to understand atmospheric change and supporting air quality control. Key Points Using a simple tropospheric ozone chemistry model, we quantify uncertainty caused by simplifying the model with machine learning Compared to the deterministic simplification method, the probabilistic model also reduces error Full‐scale use could improve uncertainty quantification in atmospheric modeling, improving atmospheric insight and air quality control