Bridging Observations, Chemistry, and AI: A Comprehensive Study of Gigantic Jets from Parent Thunderstorms to Mesospheric Chemical Impact

This study presents an integrated investigation of Gigantic Jets (GJs), encompassing an analysis of parent thunderstorm conditions and a quantitative assessment of their chemical impact on the middle atmosphere via a novel modelling approach. We focus on a remarkable sequence of five GJs observed within 7 minutes from an isolated thunderstorm over South China on 18 August 2022. Analysis reveals the event was associated with a high-altitude -10 ℃ isotherm, substantial convective available potential energy (~2158 J/kg), pronounced upper-level wind shear (~14.5 m/s), and dominant intracloud lightning activity preceded by narrow bipolar events.

To quantify the chemical perturbations, we developed the first one-dimensional plasma-chemical model that couples time-dependent electron kinetics with a comprehensive atmospheric reaction scheme. Simulations indicate that GJ discharges induce transient yet significant perturbations, most notably ozone depletion and nitrogen oxide (NOx) enhancement within the 40–50 km altitude range, driven by electron-impact ionization and subsequent ion-molecule chemistry. The model also captures the characteristic blue-to-red spectral transition in optical emissions, linking it to the excitation dynamics of N₂ states.

Addressing computational efficiency and parametric uncertainty in traditional models, this research innovatively integrates a Physics-Informed Neural Network (PINN) into the framework. The PINN, constrained by the underlying physicochemical equations, learns the mapping from background atmospheric parameters and electric fields to species concentrations. This hybrid approach enables rapid, physically consistent predictions of chemical perturbations and provides a robust tool for sensitivity analysis, highlighting the altitude-dependent sensitivity of key reaction pathways.

By synthesizing multi-platform observations, detailed plasma-chemical modelling, and advanced machine learning techniques, this work provides a comprehensive understanding of GJs, establishing a powerful and scalable framework for assessing the role of transient luminous events in middle atmospheric chemistry.