Surface ozone pollution is a critical global environmental challenge driven by the complex, nonlinear photochemical cycling of RO_x radicals (OH + HO₂ + RO₂). Oxygenated volatile organic compounds (OVOCs) are central to these cycles as both radical sources and sinks, yet their quantitative impact on regional radical budgets remains poorly understood due to historical limitations in ambient measurements. This knowledge gap hinders the accurate prediction of persistent ozone exceedances. Here we show that constraining atmospheric models with a broad suite of 23 OVOCs—specifically reactive dicarbonyls—is essential for the accurate simulation of radical chemistry in southern China's background air through comprehensive field observations and photochemical modeling. We find that models constrained with only the three most common OVOCs (formaldehyde, acetaldehyde, and acetone) overestimate hydroxyl radical concentrations by 50%–100%, whereas comprehensive constraints align simulations with observations. This discrepancy is caused by complex offsetting errors, including the severe overestimation of isoprene-derived intermediates and the significant underestimation of secondary biacetyl production. Our results reveal that photolysis of the measured OVOCs contributes 49–61% of total RO_x production, with species such as methylglyoxal and biacetyl playing unexpectedly dominant roles in driving ozone formation. These findings highlight critical deficiencies in current chemical mechanisms and demonstrate that high-resolution monitoring of reactive OVOC intermediates is vital for developing effective emission control strategies to mitigate persistent regional ozone pollution.

Predicting wind flow statistics in urban areas is important for various environmental and engineering applications. Currently, building-resolved computational fluid dynamics (CFD) simulations are the most commonly used and reliable methods to simulate urban wind flows but they are time-consuming which limits their use in real applications. Therefore, our objective is to develop a surrogate model based on deep learning (DL), which can be used as a faster alternative to CFD methods for urban flows. The proposed model hypothesis is that the spatial distributions of the time-averaged flow quantities within urban canopies are highly correlated to the local urban geometries. To test this hypothesis, we developed a model to predict the flow in uniform urban street canyons by constructing a geometry reading filter to convert local urban geometry information around the targeted locations into a numerical array as DL model inputs. A standard feedforward DL model is then trained using large-eddy simulation (LES) results to predict the mean wind and turbulence within uniform street canyons. Our results show that the model can give fast and accurate predictions compared to LES results. The prediction errors are found to range from 5.8% to 36%, and the normalized mean bias magnitudes range from 6.6×10⁻³ to 1.6×10⁻¹ for the different flow quantities. The DL model is also found to predict the flow patterns reasonably well, consistent with experimental data similar to the results of coarse-resolution LESs. This model has the potential to be further developed into a robust and practical tool for fast urban flow predictions.