When the Hunga Tonga volcano exploded on January 15, 2022, it did something no eruption had done before in the satellite record: it drove roughly 150 teragrams of water vapor — about 10% of all background moisture in the stratosphere — deep into the upper atmosphere. That made scientists uneasy. Stratospheric water vapor traps heat, and early estimates suggested the eruption could nudge global temperatures upward.
What actually happened to Earth’s energy balance turned out to be a more complicated story — one that a coordinated team of researchers and five climate models has now worked to untangle.
An eruption unlike any other in the satellite record
The January 15, 2022 explosion of Hunga Tonga–Hunga Ha’apai sent a plume reaching up to 55 kilometers into the atmosphere — far above the altitude of typical volcanic eruptions. That exceptional height helped drive an injection of roughly 146 teragrams of water vapor into the stratosphere, representing about 10% of all background stratospheric moisture. Nothing like it had been observed in the satellite era.
What made the eruption scientifically disorienting was the mismatch between its two main outputs. The water vapor injection was unprecedented. But the sulfur dioxide released — estimated between 0.41 and over 1.0 teragrams depending on the instrument — was modest by historical standards. Mt. Pinatubo, which erupted in 1991, injected somewhere between 10 and 20 teragrams of SO2. Hunga’s sulfur load was at least 20 times smaller.
That contrast set up a genuine scientific dispute. Early estimates suggested the water vapor alone could produce a small net warming, while other analyses argued the sulfate aerosols forming from the SO2 would dominate and cool the planet. Without a historical precedent, there was no obvious way to know who was right.
What five climate models agreed on
To resolve the disagreement, researchers organized the HTHH-MOC Project — a coordinated multi-model effort involving five Earth system models. Each ran simulations under nudged conditions, meaning the models were constrained to follow observed atmospheric circulation, which allowed scientists to isolate the volcanic signal from the background noise of natural variability.
The result was clear. All five models independently showed a significant negative — that is, cooling — radiative forcing at the top of the atmosphere, concentrated in the Southern Hemisphere. The multi-model mean global instantaneous radiative forcing, averaged over 2022–2023, came in at −0.19 ± 0.06 W m⁻² at the top of the atmosphere and −0.16 ± 0.06 W m⁻² at the surface. Both numbers point unambiguously toward cooling.
The models also agreed on what was not responsible. Simulations that injected only water vapor produced negligible radiative forcing. The cooling signal came almost entirely from sulfate aerosols formed by the SO2 injection — not from the record-breaking water vapor plume.
Water vapor’s hidden role: accelerating aerosol growth
That does not mean the water vapor was irrelevant. Rather than trapping heat directly, the elevated stratospheric moisture accelerated the growth of sulfate aerosol particles. In co-injection simulations — where both SO2 and water vapor were released together — aerosols reached an effective radius of about 0.4 micrometers in under a month. After Pinatubo, the same growth process took roughly four months.
The practical consequence was stronger initial cooling. During the first six months after the eruption, simulations that included water vapor produced over 10% stronger aerosol radiative forcing in the Southern Hemisphere compared to SO2-only runs: −0.39 W m⁻² versus −0.34 W m⁻². This mechanism also helps explain why Hunga’s stratospheric aerosol optical depth was only about 10 times lower than Pinatubo’s — not 20 times lower, as the SO2 difference alone would predict. The water vapor appears to have amplified aerosol optical efficiency, partially compensating for the smaller sulfur load.
Long-term forcing and the complication of ENSO
Free-running model simulations extended the analysis out to 2031. The global mean effective radiative forcing at the top of the atmosphere averaged −0.14 ± 0.10 W m⁻² across the first two years, then declined to an average of −0.09 W m⁻² through 2027. The Southern Hemisphere consistently bore the brunt of the impact — restricting the analysis to that hemisphere roughly doubled the forcing values compared to global averages.
When the ocean was fully coupled in the models, the picture became noisier. The eruption appeared to modulate El Niño–Southern Oscillation variability, triggering a La Niña-like response in 2022–2023 and an El Niño-like response around 2025 — adding variability that made the volcanic signal harder to isolate.
There is also a caveat on magnitude. The model protocol used an SO2 injection of 0.5 teragrams. Newer retrieval estimates from the Infrared Atmospheric Sounding Interferometer suggest the actual injection may have been closer to 1.0 teragrams or higher. If confirmed, the true cooling effect could be underestimated by as much as 50%.
Why this matters beyond one eruption
The HTHH-MOC study provides the most comprehensive multi-model quantification of Hunga’s radiative impact to date — relevant not just for this eruption, but for the broader challenge of attributing recent climate changes to individual causes. The finding that a moderate eruption far smaller than Pinatubo could still leave a measurable imprint on Earth’s energy budget reinforces the case for including such events in climate models.
The study also identified a methodological issue worth noting: estimating aerosol and gas radiative contributions separately produces different values than estimating them together using a double radiation call. The two approaches are not interchangeable.
Perhaps the most consequential open question involves 2023’s record-warm global surface temperatures. If Hunga’s true SO2 injection was larger than assumed, its cooling effect was also larger — making that warmth even harder to explain through volcanic forcing alone. The eruption once feared as a warming agent may have been masking heat rather than adding to it.