Multi-model multi-signal climate change detection at regional scale

Using an optimal detection technique and climate change simulations produced with four GCMs, we have assessed the causes of 20th century temperature changes from global to regional scales. Our analysis is conducted in 9 spatial domains: 1) the globe; 2) the Northern Hemisphere; and four regions in the Northern Hemispheric mid-latitudes covering 30„aN-70„aN, including 3) Eurasia, 4) North American 5) land only, 6) the entire 30„aN-70„aN belt, and three smaller regions over 7) southern Canada, 8) southern Europe, and 9) China. We find that the effect of anthropogenic forcing on climate is clearly detectable at global to regional scales.

The effect of combined greenhouse gases and sulfate aerosols forcing changes is detectable in all nine domains in annual and seasonal mean temperatures observed during the second half of the 20th century. The effect of greenhouse gases can also be separated from that of sulfate aerosols over this period at continental and regional scales. Uncertainty in these results is larger at smaller spatial domains. Detection is improved when an ensemble of models is used to estimate the response to anthropogenic forcing and the underlying internal variability of the climate system. Our detection results hold after removal of NAO related variability in temperature observations, variability that may, or may not, be associated with anthropogenic forcing. They also continue to hold when the model simulated natural internal climate variability is doubled.