Attribution of historical ozone forcing to anthropogenic emissions

Anthropogenic ozone radiative forcing is traditionally separately attributed to tropospheric and stratospheric changes assuming that these have distinct causes1. Using the interactive composition–climate model GISS-E2-R we find that this assumption is not justified. Our simulations show that changes in emissions of tropospheric ozone precursors have substantial effects on ozone in both regions, as do anthropogenic halocarbon emissions. On the basis of our results, further simulations with the NCAR-CAM3.5 model2, and published studies3,4, we estimate industrial era (1850–2005) whole-atmosphere ozone forcing of ∼0.5 W m−2 due to anthropogenic tropospheric precursors and about−0.2 W m−2 due to halocarbons. The net troposphere plus stratosphere forcing is similar to the net halocarbon plus precursor ozone forcing, but the latter provides a more useful perspective. The halocarbon-induced ozone forcing is roughly two-thirds the magnitude of the halocarbon direct forcing but opposite in sign, yielding a net forcing of only ∼0.1 W m−2. Thus, the net effect of halocarbons has been smaller, and the effect of tropospheric ozone precursors has been greater, than generally recognized. Observations of long-lived greenhouse-gas concentrations are typically used for estimating anthropogenic radiative forcing (RF; defined here as the net radiative imbalance at the tropopause after allowing stratospheric temperatures to respond to an imposed perturbation1). Very few reliable historical observations are available for short-lived ozone, however, and thus models are used to estimate changing tropospheric concentrations. Observations extend back several decades in the stratosphere, covering the period when anthropogenic ozone changes are believed to have been greatest. Assessments of RF based on observed ozone depletion in the stratosphere and multi-model simulations of tropospheric changes provide industrial-era ozone RF estimates of ∼0.35Wm−2 (range 0.25–0.65Wm−2) due to tropospheric ozone increases and −0.05 ± 0.10Wm−2 due to stratospheric ozone loss1. Similar results were recently obtained using a merged data set based on models in the troposphere and stratospheric observations5. The assumption implicit in separating these regions is that the tropopause is a strong barrier to transport, allowing the areas to evolve independently. However, it is widely recognized that there is significant transport across the tropopause and therefore, prescribing observed changes in one region and using modelled values in the other can lead to inconsistencies. In simulations using the Goddard Institute for Space Studies GISS-E2-R coupled model, we have included interactive whole-