The particles in question are cloud condensation nuclei, or CCN, the microscopic specks onto which water vapour condenses to form liquid cloud droplets. In most of the world, CCN are abundant enough that clouds form freely, but the polar atmosphere is so clean that CCN concentrations can at times become the limiting factor for cloud formation altogether. That makes even small changes in the aerosol budget disproportionately consequential. Higher CCN concentrations produce clouds with more, smaller droplets, which reflect more sunlight and persist longer, generally cooling the surface below. Lower CCN counts do the opposite, thinning and brightening clouds in ways that can either amplify or dampen Arctic warming depending on season and cloud altitude.

A central uncertainty in projecting future CCN levels is the fate of new particle formation, the process by which aerosol particles are born directly from gas-phase precursor vapours such as iodine- and sulfur-containing compounds. Newly formed particles are initially far too small to seed clouds, but they can grow over hours to days into the CCN-relevant size range. Warming could theoretically boost this process by increasing precursor emissions from melting sea ice and thawing landscapes, or by shifting meteorology in favourable directions. Yet there is a competing effect: more aerosol particles in the atmosphere also increase the condensation sink, the rate at which vapour molecules are scavenged by pre-existing particles, which starves nascent particles of the material they need to form and grow. Add to this the nonlinear temperature dependences of precursor volatility and reaction rates, and the sign of the feedback becomes genuinely difficult to predict.

To cut through that uncertainty, the research team, led by Samuel De Xun Chua and Katrianne Lehtipalo of the University of Helsinki, compiled a harmonized dataset of particle number size distributions covering 2003 to 2024, drawing on long-term observatories at Alert, Pallas, Tiksi, Utqiagvik, Varrio, Villum and Zeppelin, together with measurements from the Arctic Ocean 2018, MOSAiC and ARTofMELT research expeditions. In total, the dataset comprised 21,705 valid measurement days. The team applied recently developed algorithmic methods consistently across all sites, computing nucleation-mode, Aitken-mode and accumulation-mode particle concentrations, particle formation rates at 15 nanometres, growth rates between 15 and 20 nanometres, and the condensation sink, ensuring that differences between sites reflected real atmospheric variability rather than methodological artefacts.

The observational record revealed clear latitudinal and seasonal structure. From December to May, the Arctic atmosphere was dominated by long-range transport of accumulation-mode aerosols from lower latitudes, largely of anthropogenic origin, producing the well-known phenomenon of Arctic haze. During this period, particle concentrations were generally higher at the southern continental sites than at the northern ones, and significant declines in accumulation-mode concentrations during winter were observed at Alert, Pallas, Varrio and Zeppelin over the 2010 to 2019 window, plausibly reflecting reduced anthropogenic emissions across Eurasia and North America. In summer, by contrast, local processes took over: median Aitken-mode concentrations at the southern sites ranged from roughly 79 to 474 particles per cubic centimetre, compared with only 42 to 116 at the northern locations, and particle formation rates at 15 nanometres were nearly an order of magnitude higher in the south.

The real innovation came in translating these sparse ground-based observations into a pan-Arctic picture. The researchers developed a novel metric called Potential_NPF, which estimates the likelihood of new particle formation events on a scale from 0 to 100. It was built by training an eXtreme Gradient Boosting machine-learning algorithm on collocated observational data, using satellite-derived environmental proxies including ultraviolet radiation, water vapour content, aerosol loading, vegetation indices and sea-ice concentration, alongside reanalysis air temperatures. The model achieved a cross-validated coefficient of determination of 0.84 on training data and generalized well to the withheld Pallas validation site. Applied across the polar region from 65 degrees north to 83 degrees north, the metric produced monthly maps from 2006 to 2021 showing that new particle formation potential remained negligible through the polar night, surged after polar sunrise, and peaked earlier in the south than over the open oceans of the Barents and Greenland Seas.

Overlaying these maps on a satellite-derived dataset of cloud condensation nuclei concentrations and applying a spatial clustering algorithm, the team demarcated three aerosol regimes. The High Arctic, broadly following the summer sea-ice extent and snow-covered lands such as Greenland and the Canadian Archipelago, showed low formation potential but a positive relationship between new particle formation and CCN, confirming that local particle birth feeds directly into the regional cloud-seeding budget. The Maritime Arctic, covering the open Barents and Norwegian Seas, exhibited the highest formation potential of the three regimes and likewise a positive coupling, consistent with marine biological activity and sea-ice processes supplying precursor vapours. The Continental Arctic, a broad belt from Finland to the Lena Basin, showed a near-zero slope, indicating that its CCN population is sustained mainly by transported accumulation-mode aerosols and local anthropogenic emissions rather than by local particle formation, which is further suppressed by the large condensation sink from those transported particles.

The divergent futures of these regimes emerged when the team applied empirically derived temperature sensitivities to projected warming under the intermediate SSP2-4.5 emissions scenario, using an ensemble of five CMIP6 climate models. By 2090 to 2100, new particle formation potential was estimated to rise in the High Arctic, where sub-zero temperatures prevail for most of the year and warming enhances precursor emissions from sea-ice melt at the marginal ice zone, a trend already visible in the observed increases in nucleation-mode particles at Alert and Villum. Extrapolating the observed CCN-formation relationship forward, High Arctic CCN concentrations could increase by up to 11 percent, with a confidence interval spanning 3 to 16 percent, a substantial shift in a CCN-limited environment. In the Maritime Arctic, however, higher temperatures were associated with declining formation potential, possibly through increased precursor volatility or the suppressive effects of higher water vapour on nucleation and growth, translating into an estimated CCN decline of as much as 30 percent, with a confidence interval of 15 to 31 percent. The Continental Arctic showed almost no temperature-driven change, because its CCN depend more on long-range transport, wildfire activity and industrial emissions trajectories than on local particle formation.

The findings carry an important caution for climate modellers. Earlier studies, often anchored to a single monitoring site, reached seemingly contradictory conclusions precisely because no single location is representative of the Arctic as a whole. The authors emphasize that simple narratives, such as the assumption that warming will uniformly increase new particle formation and cloud condensation nuclei across the region, risk being actively misleading. They also acknowledge limitations: satellite retrievals are too coarse to capture short-lived local events, direct measurements of particles smaller than 12 nanometres and of gaseous precursors remain scarce, and the temperature dependence of formation potential itself may evolve as emission sources shift in the rapidly changing marginal ice zone. Addressing these gaps will require nested networks of long-term observations embedded within satellite and model frameworks. What the study delivers, however, is a framework: by recognizing the Arctic as a mosaic of distinct and sometimes opposing aerosol-cloud relationships, scientists can now build more nuanced projections of how the region’s clouds, and its climate, will respond to the warming that is already underway.

Subject of Research: Aerosol-cloud interactions and new particle formation feedbacks in a warming Arctic

Article Title: Contrasting aerosol–cloud feedbacks across a warming Arctic

Article References: Chua, S. D. X., Hartl, H. G., Boyer, M., Pernov, J. B., Krejci, R., Tunved, P., Zieger, P., Kojoj, J., Massling, A., Skov, H., Russell, L. M., Saha, S., Chan, T., Asmi, E., Vakkari, V., Backman, J., Hyvärinen, A.-P., Williamson, C. J., Jaakkola, I., … Lehtipalo, K. (2026). Contrasting aerosol–cloud feedbacks across a warming Arctic. Nature Geoscience. https://doi.org/10.1038/s41561-026-02114-x

Image Credits: AI Generated

DOI: 10.1038/s41561-026-02114-x

Keywords: Arctic, aerosols, cloud condensation nuclei, new particle formation, climate feedback, sea ice, Arctic haze, satellite remote sensing, Nature Geoscience, Arctic warming, marine aerosol, climate modelling