It is widely assumed that global mean surface air temperature response is proportional to the global mean radiative forcing on the tropopause. Studies with different climate models showed that the coefficient of proportionality (climate sensitivity) depends little on the nature of the forcing. However forcings used in those studies, such as change in solar constant, carbon dioxide or sulfate aerosol concentration, have one common feature. Namely, all these changes produce forcing on the surface of the same sign as on the tropopause. On the other hand, increase in the concentration of black carbon produces positive forcing on the tropopause and negative on the surface. In this study climate response to the forcing of such kind is studied by means of numerical simulations with a climate model of intermediate complexity. Our simulation show that such a forcing can lead to either warming or cooling of the surface depended on the stratification of the forcing. Heating of the low troposphere leads to the warming of the surface in spite of direct surface forcing being negative. When heating is concentrated in middle or upper troposphere surface air temperature decreases. This difference in climate response is associated, to large extent, with difference in the changes in clouds. Contributions of other feedbacks, such as temperature, water vapor and surface albedo feedbacks, are also estimated.

Simulations with the versions of the model with different sensitivities show, however, that sensitivity to changes in CO2 concentration provide a reasonably well measure of the model sensitivity to other forcings. At the same time results of those simulations show that if the main reason for the difference in models’ sensitivities is a difference in the strength of cloud feedback the range of models’ responses to the forcings leading to the cooling of the surface will be more narrow that range of models’ responses to the forcings leading to the warming. This is explained by cloud feedback being less efficient in the case of increasing sea ice extent and snow cover.