Abstract

Anthropogenic activities have dramatically altered the chemical composition of the atmosphere. The focus of this study is on the composition of the troposphere, mainly associated with ozone which acts as a greenhouse gas, is damaging to living organisms, and co-determines the oxidative capacity of the atmosphere. A coupled tropospheric chemistry - general circulation model (ECHAM) has been applied to the simulation of tropospheric ozone distributions, using emissions of ozone precursors (NOx, CO, higher hydrocarbons) as boundary conditions. The model has been extended with detailed parameterizations for dry deposition of trace species, for the lower stratospheric ozone concentration which is used as boundary condition, and for the treatment of higher hydrocarbon species. The model has been extensively evaluated by comparison with observed long-term climatological data and with in-situ measurements from specific measurement campaigns. A proper representation of all ozone sources and sinks is prerequisite to an accurate estimate of the anthropogenic ozone increase in the troposphere. The representativity of stratosphere-troposphere exchange, which forms a major source for ozone in the troposphere, and its contribution to tropospheric ozone levels has been studied. Simulations have been performed using pre-industrial, present-day and future emission scenarios as boundary conditions, and the radiative forcing associated with the ozone increases has been estimated. The annually averaged global tropospheric ozone contents from these simulations are 190 Tg O3, 271 Tg O3, and 332 Tg O3 in 2025, corresponding to a global annual net radiative forcing at the tropopause of 0.42 W m-2 between the pre-industrial and the present-day simulations, and of 0.31 W m-2 between the present and future simulations. A second focus of the study is the simulation of the sulfur cycle. The model was part of a model intercomparison exercise, that aimed to document the present status of global sulfur cycle models and to identify major uncertainties in process parameterizations.