From Nature Geoscience
It is the regional and seasonal expression of climate change that determines the effect of greenhouse warming on ecosystems and society1. Whereas anthropogenic influences on European temperatures have been detected over the twentieth century2, 3, it has been suggested that the impact of external influences on European temperatures before 1900 is negligible4.
Here we use reconstructions of seasonal European land temperature5, 6 and simulations with three global climate models7, 8, 9 to show that external influences on climate—such as the concentrations of stratospheric volcanic aerosols or greenhouse gases, other anthropogenic effects and possibly changes in total solar irradiance—have had a discernible influence on European temperatures throughout the past five centuries. In particular, we find that external forcing contributes significantly (p<5%) to the reconstructed long-term variability of winter and spring temperatures and that it is responsible for a best guess of 75% of the observed winter warming since the late seventeenth century.
This warming is largely attributable to greenhouse-gas forcing. Summer temperatures show detectable (p<5%) interdecadal variations in response to external forcing before 1900 only. Finally, throughout the record we detect highly significant summer cooling and significant winter warming following volcanic eruptions.
See the: Supplementary Information (995KB)
Based on the multiregression result for European seasonal temperature described in the body of the paper, there is some evidence for solar forcing being detectable in summer, but the result was sensitive to the analysis period. However, our knowledge on forcing and response is not equally robust between forcings. For example, greenhouse gas
forcing is far better constrained than solar forcing, and volcanic forcing could be argued
to be more robust than solar as well (e.g. ref. 23).
To further investigate to what extent our results are robust to first identifying the more robust forcing responses, we used a stepwise regression, which first estimates the response to better constrained forcings, such as greenhouse gases and volcanism, and estimates solar forcing from the residual. To enhance power, we also use the spatial pattern.
This is equivalent to using a solar forcing fingerprint in time that is orthogonalized to that of anthropogenic forcing (see figure SI6; note that this essentially removes the anthropogenic component from the timeseries and resulting spatial pattern prior
to a regression on the solar timeseries).
This approach is consistent with the greater confidence in the shape and size of the anthropogenic forcing. Note that if data up to 1950 are used, the only difference between the orthogonal and original regressor is that the longterm trend is no longer visible (Figure SI6). The resulting regression pattern was compared with patterns obtained by regression of a random time series of the same autocorrelation as the solar response time series onto the reconstruction data (see figure SI6 for some examples).
If the summer reconstruction over the period 1500 to 1950 is used, then the response
pattern to solar forcing is significantly warmer in the area average than that obtained from random time series in the reconstruction. If the same is done to model data, then the result is warming, which is not significant, but its pattern (Figure SI7) projects more strongly on that from the reconstruction than 90% of the cases where a random timeseries was regressed on both models and reconstruction. Similar results are obtained if the timeseries is analyzed until 1996 rather than 1950. In contrast, using the shorter period back to 1675 provides insignificant results.
However, if the solar pattern is orthogonalized to both the anthropogenic and volcanic EBM fingerprints (effectively removing the contribution by both forcings to the reconstruction prior to analysis) the response pattern is no longer detectable. This raises concerns
that despite a low correlation between the EBM response to solar forcing and volcanism (0.11), some degeneracy may remain between both, particularly given that that both solar forcing and volcanism tend to cool the period termed the Little Ice Age. Thus, the detection of a solar response in European summer temperatures remains uncertain.
Regressions of the solar forcing timeseries on European temperatures in other seasons than summer (JJA) show no evidence for a detectable solar signal, and larger samples and model simulations with individual forcings are needed to assess the possibility of a dynamical response to solar forcing in the cold season as discussed in the literature.
h/t to Dr. Leif Svalgaard