Stratospheric cooling and the troposphere - Reply

The success of our method for reconstructing tropospheric temperature trends is reinforced by Gillett et al., who show that our method is robust for reconstructing the tropospheric temperature trends, and that the statistical relationships between our T 4, T 2 and T 850-300 estimates are in agreement with those independently derived from climate-model output. But Tett and Thorne use different data sets in the tropical region and suggest that our approach produces tropospheric temperature trends that are biased to warm and that it overfits the data. We argue that the differences in tropical tropospheric temperature trends between our estimate of T (T fjws) and the T 850-300 of Tett and Thorne do not invalidate our method. We also question their interpretation of the comparison between global climate model results and satellite observations. © 2004 Nature Publishing Group.