Preprints
https://doi.org/https://doi.org/10.21203/rs.3.rs-6426983/v2
https://doi.org/https://doi.org/10.21203/rs.3.rs-6426983/v2
06 Oct 2025
 | 06 Oct 2025
Status: this preprint is open for discussion and under review for Atmospheric Chemistry and Physics (ACP).

Causal inference for stratospheric chemistry: insights into tropical middle stratospheric ozone variability

Evgenia Galytska, Birgit Hassler, Carlo Arosio, Martyn P. Chipperfield, Sandip S. Dhomse, Kimberlee Dubé, Wuhu Feng, Fernando Iglesias-Suarez, and Jakob Runge

Abstract. This study investigates the coupling between chemical and dynamical processes driving tropical middle stratospheric ozone (O3) variability using a causal inference framework that combines causal discovery with causal effect estimation. This approach integrates qualitative physical knowledge through a causal graph applied to satellite observations and a chemistry-transport model (CTM) simulation. The analysis is split into two subperiods of monthly data: 2004–2011, characterized by an O3 decrease, and 2012–2018, when O3 increased in the tropical middle stratosphere. Causal inference identifies distinct processes governing O3 behaviour. During 2004–2011, a robust negative contemporaneous connection from N2O to NO2 emerged, while in 2012–2018 this shifted to a one-month lag. This slower response reduced NO2 production from N2O oxidation, limiting O3 loss via the NOx catalytic cycle. Further analysis across Quasi-Biennial Oscillation (QBO) regimes reveals regime-dependent differences in the causal links. The N2O to NO2 connection is weaker during westerly shear, associated with reduced upwelling, and stronger during easterly shear, reflecting enhanced upwelling.

Our study highlights the pivotal role that causal inference can play in disentangling complex chemical-dynamical influences on O3, complementing traditional statistical methods. This approach lays the foundation for broader applications in stratospheric chemistry, where many relations remain uncertain. By discovering and quantifying causal links, this methodology addresses open questions with environmental and societal relevance. Therefore, integrating causal reasoning into data-driven science enhances process understanding and strengthens the synergy between machine learning and statistical methods in Earth and environmental sciences.

Share
Evgenia Galytska, Birgit Hassler, Carlo Arosio, Martyn P. Chipperfield, Sandip S. Dhomse, Kimberlee Dubé, Wuhu Feng, Fernando Iglesias-Suarez, and Jakob Runge

Status: open (until 17 Nov 2025)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
Evgenia Galytska, Birgit Hassler, Carlo Arosio, Martyn P. Chipperfield, Sandip S. Dhomse, Kimberlee Dubé, Wuhu Feng, Fernando Iglesias-Suarez, and Jakob Runge
Evgenia Galytska, Birgit Hassler, Carlo Arosio, Martyn P. Chipperfield, Sandip S. Dhomse, Kimberlee Dubé, Wuhu Feng, Fernando Iglesias-Suarez, and Jakob Runge
Metrics will be available soon.
Latest update: 06 Oct 2025
Download
Short summary
We explored how chemical and dynamical processes shape ozone in the tropical middle stratosphere. Using a method that identifies cause and effect with satellite data and a chemistry-transport model, we found that from 2004–2011 nitrous oxide quickly affected nitrogen dioxide and ozone, while from 2012–2018 this effect was delayed, weakening ozone loss. Large-scale winds also influenced this link, clarifying how different mechanisms control ozone.
Share