On the Relative Humidity of the Atmosphere

Water is a nearly miraculous molecule that enters into the operation of the climate system in a remarkable variety of ways. A wide-ranging overview can be found in Pierrehumbert (2002). In the present chapter, we will be concerned only with the radia-tive effects of atmospheric water vapor on climate, and with the kinematics of how the water vapor distribution is maintained. These effects are important because feedbacks due to changes in atmospheric water vapor amplify the climate system's response to virtually all climate forcings, including anthropogenic and natural changes in CO 2 , changes in solar luminosity, and changes in orbital parameters. In contrast to cloud feedbacks, which differ greatly amongst general circulation models, clear-sky water vapor feedback is quite consistent from one model to another. Essentially all general circulation models (GCMs) yield water vapor feedback consistent with that which would result from holding relative humidity approximately fixed as climate changes (Held and Soden 2000; Colman and McAvaney 1997). This is an emergent property of the simulated climate system; fixed relative humidity is not in any way built into the model physics, and the models offer ample means by which relative humidity could change. Why should models with such diverse representations of moist processes yield such similar results when it comes to water vapor feedback? The answer to this question has a considerable bearing on the extent to which one can trust models to faithfully reproduce the water vapor feedback occurring in nature, and in particular in climates that haven't yet been directly observed. There is much indirect evidence that the water vapor feedback in models is correct, and indeed no compelling reason has emerged to doubt it. Nonetheless, it has proved difficult to articulate cleanly and convincingly from basic principles exactly why one should have confidence in this aspect of the models. If the atmosphere were saturated at all levels, understanding water vapor feedback would offer few challenges. The difficulty arises from the prevalence of highly unsaturated air in the atmosphere. To make progress, one needs better conceptual models of the factors