An ensemble of six simulations of HadAM3 were driven with observed Sea-Surface Temperatures, well-mixed greenhouse gases, tropospheric and stratospheric ozone changes, land-surface changes, and the direct and indirect effects of sulphate aerosol. HadAM3 has 19 vertical levels and a resolution of 3.75 by 2.5 degrees. We analyzed the period 1952 to 2002 from these simulations. Simulated channel 4 (Lower Stratospheric LS) temperatures were computed from the simulations. We find that simulated global-average LS temperature is almost entirely controlled by surface temperature and "stratospheric forcing". Comparison of these simulations with a L64 version of the model and line-by-line calculations show that this result in unlikely to be an artifact arising from low vertical resolution.

Having established that the simulated LS temperatures, and likely nature, are controlled by surface temperatures and stratospheric forcing we examine the difference between the HadAM3 ensemble and satellite observations for the period 1979-2002. The difference is too much warming at the time of major volcanic eruptions with a exponential damping timescale of about five years. We conclude that the difference is due to stratospheric water vapor which enters the stratosphere after major volcanic eruptions.