Satellite measurements of the clear-sky greenhouse effect from tropospheric ozone

Radiative forcing from anthropogenic ozone in the troposphere is an important factor in climate change, with an average value of 0.35 W m according to the Intergovernmental Panel for Climate Change (IPCC). IPCC model results range from 0.25 to 0.65 W m, owing to uncertainties in the estimates of pre-industrial concentrations of tropospheric ozone, and in the present spatial and temporal distributions of tropospheric ozone, which are much more variable than those of longer-lived greenhouse gases such as carbon dioxide. Here, we analyse spectrally resolved measurements of infrared radiance from the Tropospheric Emission Spectrometer on board the NASA Aura satellite, as well as corresponding estimates of atmospheric ozone and water vapour, to obtain the reduction in clear-sky outgoing long-wave radiation due to ozone in the upper troposphere over the oceans. Accounting for sea surface temperature, we calculate an average reduction in clear-sky outgoing long-wave radiation for the year 2006 of 0.48 ± 0.14 W m between 45 S and 45 N. This estimate of the clear-sky greenhouse effect from tropospheric ozone provides a critical observational constraint for ozone radiative forcing used in climate model predictions. The Tropospheric Emission Spectrometer (TES) is an infrared Fourier-transform spectrometer on board the NASA (National Aeronautics and Space Administration) Earth Observing System Aura platform. Launched in July, 2004, Aura is in a near-polar, sun-synchronous orbit with equator crossing times of 13:40 and 2:29 local mean solar time for ascending and descending orbit paths, respectively. TES is predominantly nadir viewing and measures radiance spectra at frequencies between 650 and 2,250 cm of the Earth’s surface and atmosphere. TES was designed with sufficiently fine spectral resolution (0.06 cm, unapodized) to measure the pressure-broadened infrared absorption lines of ozone in the troposphere. Along with ozone profiles, atmospheric temperature, concentrations of water vapour, deuterated water vapour, carbon monoxide and methane, effective cloud pressure and optical depth, surface temperature and land emissivity are derived from TES radiance spectra. The TES forward model used for computing spectral radiances in atmospheric retrievals is based on a line-by-line radiative transfer model, which has been used extensively for calculating atmospheric heating and cooling rates. Radiometric calibration, retrieval algorithms and error characterization have been described previously. TES radiances have been validated using the Atmospheric Infrared Sounder (AIRS) spectrometer on Aqua, which is∼15min ahead of Aura. For the ozone absorption band near 9.6 μm, TES V002 data have a 0.12 K cold bias with respect to AIRS (ref. 17), which is within the accuracy of AIRS radiance measurements, 0.2 K (ref. 18). TES ozone retrievals have been compared with ozonesondes and have a consistent high bias, ∼10 p.p.b. in the troposphere, which is accounted for in this analysis. The vertical resolution for ozone profiles is 6–7 km, and vertical sensitivity, as quantified by degrees of freedom, is ∼1.5 degrees of freedom in the troposphere for clear-sky tropics and subtropics. Unlike water vapour, the bulk of ozone absorption in the infrared region is confined to the spectral range around 9.6 μm. To compute the top-of-atmosphere (TOA) flux associated with infrared ozone absorption, TES radiance spectra are integrated and converted to flux in Wm (see the Methods section). To remove the largest sources of variability in nadir radiance spectra (clouds and land), we select clear-sky ocean scenes (see Supplementary Information, Fig. S1) using TES spectra and atmospheric retrievals with corresponding cloud effective optical depth <0.05. For clear-sky tropical (30 S to 30 N) ocean scenes, Huang et al. compute an average flux of 18Wm from 1970 Infrared Interferometer Spectrometer (IRIS) spectra for the 985–1,080 cm ozone band. Over the same latitudes and frequency range used with the IRIS spectra, we obtain 18.6± 0.15Wm (uncertainty from radiometric accuracy and anisotropy assumptions) with a standard deviation of 0.8Wm for the annual average (December 2005 to November 2006) clear-sky, ocean flux from TES spectra, consistent with the IRIS data. Previous studies have used satellite radiance spectra to demonstrate decadal greenhouse gas changes and to test whether general circulation and climate model predictions reproduce the sources of variability present in the spectra. These studies have shown how spectral resolution enables characterization of the parameters that drive outgoing long-wave radiation (OLR) variability beyond what can be obtained with broadband OLR measurements such as those from the Clouds and the Earth’s