Comment on 'Quantitatively evaluating the effects of CO2 emission on temperature rise'

In a recent paper (Chen et al., 2013), fractional changes in temperature were correlated with fractional changes in anthropogenic emissions of carbon dioxide in order to estimate the amount of warming attributable to CO2 emissions. No justification was given for using the emissions rate rather than cumulative changes in atmospheric CO2 and the assumptions are not physically-based. This leads to counter-physical predictions, such as cases where increased heating causes the Earth to cool. Furthermore, a 10-year shift in start date alters the calculated elasticity by a factor of 2. A standard energy balance model is physically-based and outperforms the Chen et al. (2013) model, and its fitted transient climate response is consistent with changes in atmospheric CO2 causing warming equivalent to approximately 100% of the temperature change observed since 1960, consistent with formal attribution analyses. It is cautioned that purely statistical correlations are not able to demonstrate cause, and that they are particularly poor at attribution when there is no physical basis for the selection of variables and functional forms used in the correlation analysis.