Authors' response

— In their reply to our Commentary1, Forster et al.2 mainly criticize us for points which we did not make regarding climate sensitivity and future global warming and fail to come to grips with our central point, namely that in assessing the skill of climate models by their ability to reproduce warming over the twentieth century, the latest report from the IPCC3 may give a false sense of their predictive capability. Increasing the spread of the modelled twentieth century climate change may have implications for estimates of climate sensitivity and its uncertainty and for projections of future climate change, but those implications cannot be determined unless the consequences of uncertainties in forcing are systematically examined.

Additionally, we have several specific points in response to Forster et al. as follows:

  • Changes in forcing over time that are scaled by a constant factor would, for a linear relationship between forcing and response, result in a relative change in surface temperature of the same factor irrespective of the time period over which the temperature change was measured. There would thus seem to be little justification for relying on the modelled temperature change not responding linearly to the applied forcing or the temperature change not being in equilibrium with the forcing as the reason for reduction in uncertainty in temperature change relative to that of forcing in the absence of model experiments explicitly examining these points.

  • No comfort should be taken in the fact that different models used in the study may exhibit different rates of ocean heat uptake; rather the goal of stringent examination of model performance should be to rule out, or improve, those models that poorly represent the process being simulated.

  • We did not dispute that forced simulations are necessary for models to exhibit the observed increases in surface temperature.

Climate models are essential for examining prospective future climate change that would result from changes in atmospheric composition and other forcings. We anticipate that much progress would be made in understanding and characterizing the present suite of climate models if the IPCC and modelling groups that participated in the exercise leading to Fig. SPM-4 (ref. 3) were to systematically examine the implications of the uncertainty in forcing. Only by constraining the forcing will it be possible to rule out models that skilfully reproduce twentieth century warming for the wrong reason.