Natural aerosol variability can reverse the apparent air-quality response to abrupt emission reductions

JIANI YANG, Tao Li, Haolin Wang, Mohammad Saleh Ali-Taleshi, Lin Tan, King-Fai Li, Yaowei Li, Meng Gao, Philippe Ciais

Abrupt societal disruptions, including economic shocks, pandemic lockdowns, and armed conflicts, are widely used as natural experiments for attributing changes in air quality to changes in emissions. Such attribution commonly rests on observed concentrations: combustion tracers fall, and the resulting improvement is credited to reduced emissions. Here we show that coincident natural aerosol variability can mask, and in this case reverse, the expected concentration response, so that observation-only attribution can recover the wrong sign. We demonstrate this with the 2026 Iran conflict (28 February to 31 March 2026), combining satellite columns, Tehran ground stations, two aerosol reanalyses, back-trajectory analysis, and GEOS-Chem factor-separation simulations. TROPOMI columns and ground stations confirm substantial reductions in SO2, NO2, and CO, the signature ordinarily read as an air-quality improvement. Over the same period and domain, surface PM2.5 instead rose by 67.7 μg m−3. Factor separation attributes the increase to mineral dust (+74.8 μg m−3, exceeding the net anomaly itself), which overwhelms the emission-driven reduction (−2.7 μg m−3) and a small meteorological term (−0.8 μg m−3). Friction-velocity diagnostics place the driver in the tail of the wind distribution rather than in the mean: extreme wind-stress events over the Iraq–Kuwait source region became roughly three times more frequent, a signal that cancels in region-mean statistics but is amplified by the nonlinear threshold response of dust emission. The inferred sign is also scale-dependent, since urban stations in Tehran record cleaner air while the surrounding region deteriorates. We conclude that attribution during abrupt disruptions requires explicit separation of natural and anthropogenic aerosol contributions, and that combustion-tracer columns alone are not sufficient evidence of an air-quality benefit.