Optimal fingerprint studies have shown that recent observed surface temperature trends show a significant climate change as compared to estimates of internal climate variability, and agree best with coupled model simulations forced with greenhouse gases and sulfate aerosols. Such studies yield estimates of the amplitude of climate change patterns from the observations, which are subject to several uncertainties. One uncertainty is related to using climate change patterns and internal variability from coupled models. Uncertainty is also associated with the instrumental observations. First, instrumental surface temperature observations are subject to sampling uncertainties due to limited spatial coverage of gridboxes by station data and complete absence of data at times. The influence of these errors on estimated signal amplitudes is demonstrated.
A further uncertainty is highlighted by the difference in the temperature evolution over the previous decades at the surface and in the free atmosphere. Besides possible instrumental biases in observations (which are disregarded in this work), physical mechanisms cause differences in the timing of warming over the recent three decades at the surface and aloft. The influence of these differences on estimates of the magnitude of climate change patterns is discussed