Differences in the water vapor distribution among available datasets demonstrate that large uncertainties exist in assessment of the mean distribution. A numeric study of radiative sensitivities to perturbations in water vapor has been carried out using realistic clouds. At both the TOA and the surface, net radiative sensitivity is dominated by the longwave spectrum and is largest in the dry zones of the subtropics and eastern Pacific Ocean where measurement uncertainty is also large. The results underscore the important role the arid tropical and subtropical regions play in the climate’s radiative balance.

A strong relationship between lateral gradients in sensitivity and clouds is identified. In regions of frequent high clouds, sensitivity at TOA is significantly reduced from the cloud-free and low-cloud environments. Similarly, surface radiative fluxes in regions dominated by frequent low clouds are less sensitive than in regions where low clouds are rare. Longwave sensitivity is more than twice as large at the surface than at TOA and is therefore the dominant component of total columnar heating sensitivity. At both the surface and TOA, solar flux sensitivities are of opposite sign and are substantially reduced from those in the longwave. Sensitivity of TOA fluxes to natural variability and measurement uncertainty is substantially greater in the middle and upper troposphere than in its lower layers. In contrast, surface fluxes are more strongly impacted by the lower troposphere and low level variations are therefore critical for coupled simulations.