Research
Using a deep learning U-net model to estimate NOₓ emissions

Using a deep learning U-net model to estimate NOₓ emissions

Introduction

Starting in the spring of 2025, I worked with Prof. Dylan Jones to use a deep-learning U-net model to estimate NOₓ emissions. The code I am developing for this project is contained within the Python Package unox. Documentation for this package including an installation guide, example usage, and complete API reference is available on Read the Docs. The code is available on the repository’s GitHub page.

The U-net model used in this project is based on Tailong He’s repository for Chinese NOₓ emissions1. The initial transition from the China region to North America was done by Evelyn MacDonald. The initial adaptation to make estimates for CO was done by Daniel Sequeira.

References

  1. He, T.-L., Jones, D., Miyazaki, K et al. (2022) “Inverse modeling of Chinese NOₓ emissions using deep learning: Integrating in situ observations with a satellite-based chemical reanalysis”, Atmospheric Chemistry and Physics, 22(21):14059-14074, doi:10.5194/acp-22-14059-2022 ↩︎