New! Sign up for our email newsletter on Substack.
For every three solar panels’ worth of electricity that came online somewhere in the world between 2017 and 2023, the equivalent of one panel’s worth of power quietly vanished. Not stolen by faulty wiring or cloudy weather. Taken by the air itself, specifically the the microscopic particles drifting out of the coal-fired power plants that solar was supposed to be replacing. The numbers are, when you sit with them, a little vertiginous. Globally in 2023, aerosol pollution shaved 5.8% off solar photovoltaic output: 111 terawatt-hours, enough to supply the entire annual generation of 18 medium-sized coal plants, gone before it reached the grid. And most of this effect was invisible in the models governments and energy planners had been using to track the clean energy transition.
That gap between what solar produces and what it could produce is the subject of a new study in Nature Sustainability, led by researchers at the University of Oxford and University College London. The team mapped more than 140,000 solar PV installations worldwide using satellite imagery and machine learning, crossing that facility-level geography with atmospheric data to calculate losses site by site: the most granular picture yet of how fossil fuel infrastructure is degrading the very renewables built to replace it.
The study’s central finding cuts at something that gets underappreciated in the relentless optimism surrounding the solar boom: renewable capacity and fossil fuel capacity have not been trading places. They have been growing together. China built solar installations at a pace that made it the world’s largest solar producer, generating 793.5 terawatt-hours in 2023, some 41.5% of the global total. It also kept building coal plants, with new construction reaching a near-decade high in 2024. The consequence shows up with uncomfortable precision in the data. In China, aerosols cut solar output by 7.7% in 2023, and the aerosol losses from existing installations ran at 38.4% of the energy added by new capacity on average. In some years, that ratio exceeded 50%; it peaked at 62.1% in 2021. In other words, for stretches of three consecutive years, more than half the electricity China was adding via new solar panels was simultaneously being eaten away by the pollution its coal plants were producing.
“We are seeing rapid global expansion of renewable energy, but the effectiveness of that transition is lower than often assumed,” said lead author Dr Rui Song, of the University of Oxford’s Department of Physics and the Mullard Space Science Laboratory at UCL. “As coal and solar expand in parallel, emissions alter the radiation environment, directly undermining the performance of solar generation.”
The mechanism itself is not new physics. Coal combustion emits sulfur dioxide, which oxidises in the atmosphere into sulfate aerosols, fine particles that scatter and absorb incoming sunlight before it can reach a panel. In China’s case, sulfate aerosols accounted for 46.2% of all aerosol-related PV losses, which made them the dominant factor by some margin. Carbonaceous particles (also partly from coal) added another 18.4%. Dust, though spatially concentrated in arid western regions, contributed roughly a third of losses, but even in those desert zones, where you might expect coal’s influence to be minimal, the data showed sulfate aerosol losses remaining high. The co-location of solar farms and coal plants in China, it turns out, extends far beyond the North China Plain into the regions that policy documents have tended to describe as dedicated renewable hubs.
The contrast with the United States is instructive. American solar installations suffered only 3.1% aerosol-induced losses in 2023, and the loss-to-growth ratio reached a minimum of 12.8% by year’s end. The researchers traced this partly to atmospheric conditions, but also to something simpler: in the US, solar and coal are geographically separated. Large-scale solar has concentrated in the low-pollution West, while coal plants cluster in the East, and there is essentially no spatial correlation between coal capacity and solar losses across American grid cells. The upshot, as the paper notes, is that aerosol-induced performance loss is not an inherent property of solar panels. It is what happens when you build them next to coal plants.
There is, however, a piece of good news embedded in the China data, and it is somewhat paradoxical. Despite suffering the worst aerosol losses of any major economy, China is also the only country where those losses have been falling consistently, at an average rate of 1.4% per year since 2013. The US and Europe, meanwhile, saw losses trend upward by 1.5% and 1.3% per year respectively. The improvement in China came not from closing coal plants but from retrofitting them with ultra-low-emission technology, which accounted for 91% of SO2 reductions from the coal power sector. Retirements contributed only 9%. China’s coal fleet is getting cleaner without getting smaller.
Song was unambiguous about the limits of what cleaning up existing coal can achieve. “Air pollution doesn’t just block sunlight – it also changes clouds, which can cut solar power even further,” he said. “That means the real impact is likely to be bigger than we’ve measured, so we may be overestimating how much solar power can contribute to reducing emissions if we do not get pollution from coal power under control.” The study explicitly notes that its loss estimates are conservative, capturing only the direct radiative effects of aerosols on sunlight, not the additional suppression that comes when aerosol particles alter cloud formation and reflectivity. What’s already measurable is large enough, arguably, but the actual toll could be higher still.
Professor Myles Allen of Oxford, who founded Oxford Net Zero and was not involved in the research, offered an observation that puts the economics in a different frame. “All scenarios that meet the goals of the Paris Agreement show a rapid transition away from unabated coal, which isn’t happening,” he said. “The reason is that coal power is still remarkably cheap – as this study shows, that’s because the real costs are hidden.”
Those hidden costs have a way of showing up elsewhere. Dr Chenchen Huang of the University of Bath, a co-author, pointed to the risk that standard projections will systematically overestimate how much clean energy the world is actually delivering. “Our findings send a clear warning to the Sustainable Development Goals: overlooking pollution-induced solar energy losses can lead to a systematic overestimation of renewable energy output by governments, businesses and the broader community,” she said. “To stay on track, policies must account for this hidden drag and shift fossil-fuel subsidies away from coal.”
The study’s methodology is likely to sharpen in coming years. Professor Jan-Peter Muller of UCL’s Mullard Space Science Laboratory noted that the satellite infrastructure for tracking these effects in near-real-time is already emerging. “In the near future, we will be able to observe the impacts of dust and smoke particles on reducing solar energy at the Earth’s surface in real-time every 10 minutes from geostationary satellites spanning the Earth,” he said. What that improved surveillance reveals about the interaction between coal and solar, in a decade when both are still expanding across large parts of the world, remains one of the more consequential questions in energy policy. The panels are going up. So is the smoke. The question is which one we’re really counting.
Read the study: Coal plants persist as a large barrier to the global solar energy transition, Nature Sustainability (2026).
Why would building more solar panels not automatically reduce coal’s share of the grid?
Because renewables and fossil fuels have mostly been expanding in parallel rather than trading places. When energy demand grows fast enough, new solar capacity can come online without displacing existing coal generation. This study adds a physical twist to that economic problem: coal pollution actively degrades the output of nearby solar panels, meaning the net electricity gained from new installations is smaller than the nameplate capacity suggests.
How much does coal pollution actually reduce what a solar panel produces?
Globally, aerosol pollution cut solar PV output by 5.8% in 2023, equivalent to losing the entire output of 18 medium-sized coal plants. In China, where solar and coal plants are often built near each other, the figure was 7.7%. In some years, the energy lost to aerosols was more than half the energy gained from new solar installations in China, a loss rate that largely goes unaccounted for in official projections.
Is cleaning up coal plant emissions enough to fix the problem, or do the plants need to close?
Cleaning up helps: China’s aggressive ultra-low-emission retrofit programme has reduced aerosol-related solar losses by about 1.4% per year since 2013, even as coal capacity has continued to grow. But retrofits cut particulate and sulfur emissions, not carbon dioxide, so they don’t address the climate dimension. The researchers are also clear that their loss estimates are conservative; aerosol effects on clouds add further suppression that wasn’t captured, which means full coal retirement would deliver substantially more benefit than cleaning up existing plants.
Does this problem affect solar panels in the UK and Europe?
Europe’s aerosol losses are moderate and closer to the global average, but the trend is moving in the wrong direction: aerosol-induced PV losses in Europe rose by about 1.3% per year over the study period, compared with China’s sustained decline. In the UK specifically, cloud cover is a bigger factor than pollution in limiting solar output. Improved geostationary satellites are already making it possible to forecast cloud movements more accurately, which helps grid operators manage fluctuations in solar generation.
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.
This site uses Akismet to reduce spam. Learn how your comment data is processed.
