Code and Data for Sturm and Silva (2024) A nudge to the truth: atom conservation as a hard constraint in models of atmospheric composition using an uncertainty-weighted correction

This record contains the Julia photochemical model (https://doi.org/10.5281/zenodo.13385541) output in csv format used for training XGBoost in ProjectionConservationRF.py to emulate ozone photochemical formation. Nonphysical predictions that violate conservation of atoms are corrected using a closed-form, constrained least-squares approach that factors in uncertainty and scale using species-level weights. The file ozoneNOx_visualization.py contains an example and visualization for a smaller system, the primary photolytic cycle from which the Leighton relationship can be derived.

The corresponding preprint is available here: https://doi.org/10.48550/arXiv.2408.16109