Observational data are employed in a radiative transfer model to simulate the mean variation in normalized greenhouse effect (NGE) between January and July. This is performed at a variety of locations, and the mean local rate of change in NGE with surface temperature is determined. The result is 1.5 times larger than the variation of NGE with surface temperature obtained by spatially correlating the aggregated data. This disagreement is ascribed to systematic differences between the two approaches and is interpreted as indicating the significant role that large-scale circulations as well as surface temperatures have on determining local thermal and humidity structures. The separate effects of water vapor and lapse rate variations are estimated, by simulating the January--July changes in NGE with each process in turn held constant: beyond the tropics the lapse rate feedback is found to dominate over the water vapor feedback, particularly over land; in the inter-tropics, lapse rate variations account for about a third of the change in greenhouse trapping, contributing substantially to the ''super-greenhouse effect.'' Utilizing a radiative-convective model, the possible effects on climate change of both lapse rate changes and water vapor feedback are compared: a global mean model climate is perturbed by a doubling of atmospheric carbon dioxide and equilibrium surface temperatures obtained for a variety lapse rates. If, under conditions of climate change, the global mean lapse rate varies with surface temperature in the same manner as the present-day mean seasonal cycle (increasing the lapse rate magnitude by 6%), then the lapse rate feedback amplifies the modeled water vapor feedback by 40%; conversely, a 12% reduction in the magnitude of the lapse rate completely nullifies the water vapor feedback. ¿ American Geophysical Union 1995