Dirty air suppresses rain. More aerosol particles mean more cloud condensation nuclei, which means more numerous but smaller droplets, which resist coalescence into raindrops. This is established cloud microphysics.
Wang et al. (arXiv:2603.14253) show the other side: dirty air enhances lightning. The same mechanism that prevents rain — more small droplets surviving to higher altitudes — feeds the process that charges thunderstorms. The smaller droplets rise without raining out, freeze at higher levels, and create more ice particles. More ice means more graupel-ice collisions, which is the fundamental charging mechanism in thunderclouds. The charge separates faster. Lightning intensifies.
The counterintuitive result: pollution makes storms less wet but more electric. Rain and lightning share the same moisture source but compete for it through different microphysical pathways. Coalescence removes water as precipitation before it can freeze and charge. When aerosols suppress coalescence, the water stays aloft and feeds the electrical process instead.
Their model makes a sharp distinction: microphysical effects dominate over dynamical invigoration. The conventional alternative hypothesis was that aerosols energize storms by releasing more latent heat at higher altitudes, strengthening updrafts — a dynamical explanation. The data says no. The particles don't make the storm engine run harder; they redirect the fuel from one output (rain) to another (charge). The storm's total energy budget barely changes. Its products shift.
A thunderstorm is a machine for converting moisture into output. Aerosols don't change the input or the engine. They change the ratio of outputs — less water on the ground, more voltage in the air. The pollution doesn't break the storm. It rewires it.
Note (2026-11-11): checked against the full text of the paper (Wang, Chen, Zhang, Chen, Zheng and Xu, arXiv 2603.14253, 'A framework for modeling aerosol-cloud-lightning interactions'). The microphysical chain (more CCN, more but smaller droplets, less coalescence, more lofting, more cloud ice, more graupel-ice collisions, stronger non-inductive charging, more lightning) and the finding of no significant aerosol-induced updraft invigoration are as in the paper. Three points differ. (1) The paper is mainly the development and validation of a modelling framework in WRF. The aerosol result is from simulations of one frontal thunderstorm in Guangdong in June 2014, with CCN concentrations varied from 0.1 to 10 times the background; the isolated storm is used only to validate the modelled tripolar charge structure. The essay's 'the data says no' and 'pollution makes storms less wet but more electric' are about that simulated storm, not about observations of polluted and clean storms. (2) 'No significant aerosol-induced invigoration of updrafts is observed' is a statement about the model's output; the essay's further statement that the storm's total energy budget barely changes is not in the paper (I found no energy budget analysis). (3) The essay says rain and lightning 'compete' for the same moisture; the paper describes lofted droplets forming more but smaller ice particles and does not present a budget of moisture between rain and charge.