Where the model holds
The peer-reviewed study published in Science Advances on 10 July 2026 turns marine cloud brightening away from long-term warming and toward a seasonal-to-multiyear phenomenon. The team first exploits the opportunistic experiment provided by the 2019-2020 Australian wildfires: the natural cloud brightening those fires produced, and the La Nina-like response that followed, are reproduced by simulating the intervention in the southeast Pacific. It is then tried on the 1997-1998 and 2015-2016 El Nino events.[1]
The mechanism is named plainly. An intervention started during El Nino's growth phase disrupts the Bjerknes feedbacks that normally amplify El Nino conditions. That means cutting the loop which grows the event rather than suppressing a temperature difference directly, and it is where the lever finds its grip. The hardware scale Phys.org reports is a separate constraint: roughly 2,400 vessels equipped with sea-salt nozzles are needed, and current nozzles cannot yet achieve the required volume.[1]
When the lever is released
The study's most striking finding concerns persistence. After the intervention is terminated the effects weaken, and only the earliest and longest interventions restore neutral ENSO conditions and weaken teleconnections. Among the unintended consequences is an earlier La Nina following the targeted El Nino, and the paper notes that early and short interventions may counter that effect.[1]
This changes the accounting for an intervention. Gain and cost do not close inside the same event; the gain sits in the targeted El Nino while part of the cost appears in the phase that follows. An early and long intervention restores neutral conditions and at the same time raises the ship count and the duration, while an early and short one balances the La Nina effect. Timing, then, is a setting pulled between two aims rather than one with a single best value. A more cautious reading is also available: this behaviour may be specific to how this model represents ENSO and could come out differently in another.[1]
What the model did not look at
Wan says the team did not look at long-term impacts. The study that would make such a lever weighable is exactly that: the intervention's full balance across several successive ENSO cycles. What exists now is a gain, a weakening and a shift measured over two historical events, and how they behave when stacked remains open. The signal to watch is a multi-cycle simulation published together with regional temperature and precipitation differences.[1]