Convection in the tropics turns over the upper troposphere at rates (0.08 d⁻¹) comparable to photochemical processes controlling the absolute abundance of HO_x (OH + HO₂) and the abundance of NO_x (NO + NO₂) relative to HNO₃. Here we identify convection of boundary-layer CH₃OOH as a primary source of HO_x to the upper troposphere. Turnover of NO_x leads to NO/HNO₃ ratios much higher than predicted for local photochemical steady-state. Through convective transport the upper troposphere is more photochemically active in producing O₃, an important greenhouse gas.

The distribution of ²²²Rn over North America is simulated with a 3-d chemical tracer model (CTM) based on the meteorology of the GISS general circulation model (GCM). The GISS GCM has been used extensively for studies of climate change and global transport of chemical tracers. Simulation of ²²²Rn (e-folding lifetime 5.5 days) tests the ability of the model to describe the transport of pollutants in the boundary layer and the exchange of mass between the boundary layer and the free troposphere. Model results are compared to surface observations from 5 sites in the United States. It is found that the ²²²Rn concentrations are regulated primarily by dry convection. At night, the model underpredicts observations because it does not resolve the sharp²²²Rn concentration gradient which forms near the surface. In daytime, the predicted and observed concentrations are usually in good agreement, indicating that vertical mixing of surface air is reasonably simulated. Inspection of seasonal trends reveals, however, several significant discrepancies which are traced to anomalies in the GCM meteorology. In particular, the simulated²²²Rn concentrations over the northeastern United States are too high in the spring, because of excessive rainfall which suppresses dry convection, and too low in the fall, because of a severe drought which allows intense dry convection. Ventilation of ²²²Rn to the free troposphere is most efficient in the western half of the North American continent, due to intense dry convection, and is followed by rapid eastward advection of ²²²Rn in the upper westerlies. This transport mechanism produces a layer of high ²²²Rn concentrations in the upper troposphere over the eastern United States and over the western Atlantic Ocean in summer.

A one-dimensional photochemical model for cloud-topped boundary layers is developed which includes detailed descriptions of gas-phase and aqueous-phase chemistry, and of the radiation field in and below cloud. The model is used to interpret the accumulation of pollutants observed over Bakersfield, California, during a wintertime stagnation episode with low stratus. The main features of the observations are well simulated; in particular, sulfate accumulates progressively over the course of the episode due to sustained aqueous-phase oxidation of SO₂ in the stratus cloud. The major source of sulfate is the reaction S(IV) + Fe(III), provided that this reaction proceeds by a non radical mechanism in which Fe(III) is not reduced. A radical mechanism with SO₃ ⁻ and Fe(II) as immediate products would quench sulfate production because of depletion of Fe(III). The model results suggest that the non radical mechanism is more consistent with observations, although this result follows from the absence of a rapid Fe(II) oxidation pathway in the model. Even with the non-radical mechanism, most of the soluble iron is present as Fe(II) because Fe(III) is rapidly reduced by O₂ ⁻. The S(IV) + Fe(III) reaction provides the principal source of H₂O₂ in the model; photochemical production of H₂O₂ from HO₂ or O₂(−I) is slow because HO₂ is depleted by high levels of NO_x. The aqueous-phase reaction S(IV) + OH initiates a radical-assisted S(IV) oxidation chain but we find that the chain is not propagated due to efficient termination by SO₄ ⁻ + Cl⁻ followed by Cl + H₂O. A major uncertainty attached to that result is that the reactivities of S(IV)-carbonyl adducts with radical oxidants are unknown. The chain could be efficiently propagated, with high sulfate yields, if the S(IV)-carbonyl adducts were involved in chain propagation. A remarkable feature of the observations, which is well reproduced by the model, is the close balance between total atmospheric concentrations of acids and bases. We argue that this balance reflects the control of sulfate production by NH₃, which follows from the pH dependence of the S(IV) + Fe(III) reaction. Such a balance should be a general characteristic of polluted environments where aqueous-phase oxidation of SO₂ is the main source of acidity. At night, the acidity of the cloud approaches a steady state between NH₃ emissions and H₂SO₄ production by the S(IV) + Fe(III) reaction. A steady state analysis suggests that [H⁺] at night should be proportional to (E_{SO 2}/E_{NH 3})^1/2 where E_{SO 2} and E_{NH 3} are emission rates of SO₂ and NH₃, respectively. From this analysis it appears that cloud water pH values below 3 are unlikely to occur in the Bakersfield atmosphere during the nighttime hours. Very high acidities could, however, be achieved in the daytime because of photochemical acid production by the gas-phase reactions NO₂ + OH and SO₂ + OH.

A three-dimensional chemical tracer model for the troposphere is used to simulate the global distribution of ⁸⁵Kr, a long-lived radioisotope released at northern mid-latitudes by nuclear industry. Simulated distributions for the period 1980–1983 are in excellent agreement with data from six latitudinal profiles measured over the Atlantic. High concentrations of ⁸⁵Kr are predicted over the Arctic in winter, advected from European sources, and somewhat smaller enhancements arising from the same sources are predicted over the tropical Atlantic in summer. Latitudinal gradients are steepest in the northern tropics, with distinctly different seasonal variations over the Pacific, as compared to the Atlantic. The global inventory of ⁸⁵Kr is reconstructed for the period 1980–1983 by combining the concentrations measured over the Atlantic with the global distributions predicted by the model. The magnitude of the Soviet source is derived. The interhemispheric exchange time is calculated as 1.1 years, with little seasonal dependence.