Exploring the severe winter haze in Beijingthe impact of synoptic weather, regional transport and heterogeneous reactions

Same factories. Same traffic. Same coal burned for heat. Nothing changed overnight, and yet, on the morning of January 12th, 2013, air quality monitors in Beijing recorded a single-hour swing of 778 micrograms of fine particles per cubic meter. That number should be impossible if emissions are the cause. It demands a different explanation entirely. That explanation is what Zheng and colleagues set out to find. Their study of the January 2013 Beijing haze is a forensic reconstruction of hourly observations at Tsinghua University combined with targeted atmospheric model runs. What it reveals overturns the intuitive story about urban air pollution in three separate ways. Let’s start with the scale of what happened. Over the course of January 2013, the monthly average of PM2.5, which refers to fine particulate matter smaller than 2.5 micrometers and the fraction that penetrates deepest into the lungs, sat at 121 micrograms per cubic meter at the Tsinghua site. Hourly readings peaked at 855 micrograms per cubic meter. The introduction cites values approaching 900 across the region. For reference, the Chinese standard for "heavily polluted" air begins at 350 micrograms per cubic meter. Visibility dropped to around 100 meters during the worst stretches. The month contained four distinct multi-day haze episodes, and the worst episode averaged 245 micrograms per cubic meter across its entire duration. What made the event scientifically strange was not just the peak values but the swings. More than 40 individual hours recorded PM2.5 changes exceeding 100 micrograms per cubic meter. Single hours saw increases of nearly 352 micrograms and decreases of nearly 218. If local emissions were responsible, those rates of chemical production would be physically extraordinary. So Zheng and colleagues asked the first, most basic question: was this actually an emissions problem? The answer, clearly, was no. Year-on-year changes in emissions across the Beijing-Tianjin-Hebei region were tiny. Primary PM2.5 differed by just 2.1 percent between January 2012 and January 2013, sulfur dioxide by 1.5 percent, and nitrogen oxides by 2.5 percent. These are rounding errors, not causes of a catastrophe. To make the case rigorously, the team ran scenario simulations using the WRF-CMAQ modeling system, swapping 2012 and 2013 emissions and meteorology independently. Changing only the emissions produced PM2.5 differences of around 10 micrograms per cubic meter — negligible. Changing only the meteorology produced increases of 10 to 40 micrograms per cubic meter in Beijing and up to 120 micrograms per cubic meter across the North China Plain. The two meteorology scenarios produced temporal PM2.5 patterns that correlated at an R-squared value of 0.97. Meteorology explained the haze. Emissions did not. The meteorological pattern that mattered was stable and stagnant. During polluted periods, a weak high-pressure center sat northeast of Beijing, suppressing winds and blocking the cold northwesterly air that normally scours the region clean. The planetary boundary layer, which is the atmospheric layer where surface emissions mix, compressed toward the ground. With less vertical room to mix into, pollutants concentrated rapidly. Clean periods aligned with a much stronger Siberian anticyclone pushing pressure up to around 1046 hectopascals, generating the kind of wind and convective mixing that resets the atmosphere. Without that flushing, the North China Plain became a basin. Now here is where the story gets chemically interesting. Inside that meteorological basin, something unexpected was happening to the particles themselves. The natural assumption is that smog gets worse partly because sunlight drives chemistry: ultraviolet radiation breaks down ozone and other molecules, generating hydroxyl radicals, which are the atmosphere's primary oxidizing agent. These then react with sulfur dioxide and nitrogen oxides to form sulfate and nitrate particles. More sunlight leads to more oxidants, resulting in more secondary aerosol. But the January 2013 haze was dense enough to block its own sunlight. Zheng and colleagues quantify the dimming precisely. Solar radiation at the surface fell to as low as 2.77 megajoules per square meter per day on polluted days, against a clean-day average of 9.36. Ozone dropped by roughly 80 percent, from above 50 down to below 10 micrograms per cubic meter equivalent. Modeled hydroxyl radical concentrations fell from the range of 0.004 to 0.020 parts per trillion by volume down to about 0.004. When the team ran their model without heterogeneous reactions, relying only on gas-phase photochemistry, the simulated PM2.5 to elemental carbon ratio fell from 16.05 in clean conditions to 11.72 under heavy pollution. Secondary sulfate relative to elemental carbon dropped 52 percent, nitrate dropped 29 percent, and secondary organic carbon dropped 53 percent. Gas-phase chemistry should have made the haze less chemically productive as it got worse. But the observations showed the opposite. Sulfate relative to elemental carbon rose from 3.03 in clean periods to 6.35 in heavily polluted ones. Nitrate rose from 3.33 to 5.89. The secondary oxidation ratios, or SOR for sulfur and NOR for nitrogen, which measure what fraction of the sulfur and nitrogen has been chemically transformed, were essentially flat under dry conditions. Then, as relative humidity climbed above 50 percent, both jumped. At relative humidity levels of 70 to 80 percent, SOR averaged 0.34 and NOR averaged 0.28. This is the humidity signature of chemistry happening not in the open air but inside particles. Those are heterogeneous reactions. The term means chemical transformations occurring on or within existing particle surfaces — specifically, in the liquid water that coats hygroscopic particles at high relative humidity. This aqueous layer concentrates dissolved precursors. It provides enormous reaction surface area. And critically, the rate of sulfate production through this pathway scales with the volume of that aerosol liquid water, not with the concentration of gas-phase oxidants alone. Zheng and colleagues express this as a rate equation. In words: the rate of sulfate formation is proportional to how much sulfite is dissolved in the aerosol water, how many oxidants are present in that water, and how much total aqueous particle volume is available for the reaction. You multiply those three factors together. One of them, the aerosol volume, exploded during the January haze. PM2.5 rose roughly 25 times from clean to heavily polluted conditions, from about 18 to 450 micrograms per cubic meter, while relative humidity climbed from around 20 percent to around 70 percent, swelling each particle with water. Even though aqueous oxidants dropped by up to 90 percent, the surge in particle volume more than compensated. The heterogeneous pathway accelerated precisely when photochemical pathways shut down. And the pathway was selective. It preferentially produced sulfate and nitrate — the inorganic secondary species — over secondary organic carbon. Secondary organic carbon indicators stayed roughly flat with increasing relative humidity and actually declined with worsening pollution. When the revised WRF-CMAQ model was run with enhanced heterogeneous chemistry included, it reproduced observed PM2.5 with a normalized mean bias of just 0.4 percent. This selectivity explains why the chemical fingerprint of the worst haze days was so dominated by inorganics — not because organic precursors vanished, but because the chemistry that formed inorganics had found an accelerant in the particles themselves. The fourth finding reframes what those sharp hourly spikes in Beijing actually represent. The team's analysis of concurrent measurements across the region shows that aerosol accumulation didn't begin in Beijing. It built successively from cities to the southeast, under the same stable synoptic conditions, before arriving in the capital. The spike on January 12th is the clearest example: a weak cold air intrusion briefly cleared Beijing's air, then the already polluted regional air mass reoccupied the city as the intrusion weakened. What looked like rapid local chemical production was actually a fast recovery — pollution resuming its advance after a brief pause. This matters enormously for how you think about the problem. If Beijing's worst hours are driven by regional accumulation rather than local production, then reducing emissions within Beijing's city limits addresses only part of the source. The pollutant reservoir spans the North China Plain. The accumulation process begins in surrounding cities. Regional emission controls, the paper argues, are not supplementary; they are central. The study's combination of dense hourly observations and carefully designed model scenarios, which involve swapping meteorology and emissions independently and adding or removing heterogeneous chemistry, gives it a diagnostic clarity that single measurements or single-scenario models cannot. It doesn't just identify what happened in January 2013; it provides a framework for decomposing any future haze episode into its meteorological drivers, its chemical pathways, and its spatial origins. That decomposition is what makes it useful — not just as a record of one bad winter, but as a template for understanding the next one. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.