Subtropical Water Vapor as a Mediator of Rapid Climate Change

R. T. Pierrehumbert (University of Chicago, Dept. of Geophysical Sciences, Chicago, IL 60637; ph. 773-702-8101; fax 773-702-9505; Internet:

Water vapor has a dual role as a greenhouse gas and as a tracer responding rapidly to atmospheric dynamics and thermodynamics. In consequence, it has the potential to amplify the sensitivity of climate to almost any conceivable forcing mechanism, whether it be Milankovic forcing, changes in CO2, or switches in the mode of ocean circulation. It is hardly surprising, then, that attention among those interested in millenial scale variability has begun to settle on water vapor as a mediator of rapid global scale climate change. In the spirit of stimulating discussion on this important topic, I present an ideosyncratic review of the role of water vapor in climate change, focusing on the vast subtropical dry-air pools which cover most of the territory from 30N to 30S latitude. I show that, paradoxically, increasing the water vapor (or CO2) content of this region can actually have a cooling effect on the atmosphere (while at the same time having a warming effect on the surface). I will also present new evidence that the moisture in this region is governed by advective transport due to large scale transient eddies. This implies that water vapor changes could be mediated by discontinuous response of the transient eddy activity to the slow variation of controlling parameters. I also note that the important question of what determines the extent of the tropical convective region (which is moistened directly by convection) remains unresolved.

Are there any paleo-indicators of atmospheric humidity? In an intriguing paper, Broecker (Global Biogeochem. Cycles, 1997) argued that tropical snowline data together with oxygen isotope data for the Huascaran glacier (Thompson, et al., Science, 1995) imply that the atmosphere had substantially lower relative humidity during the last glacial maximum. Taking up the challenge posed in this paper, I present a re-analysis of the data, and come to rather different conclusions. I assume that: (1) the vertical temperature profile is governed by the moist adiabat, as at present, (2) The boundary layer relative humidity remains at around 80%, and (3) Snow deposited on the glacier forms at an average of 700 meters above its surface. Under these conditions, both the Holocene and glacial era snowline and isotopic data can be very nearly matched by assuming that the sea level temperature drops 3C during glacial times. The additional isotope fractionation during glacial times results primarily from a steepening of the lapse rate. I argue further that the humidity of the ambient free-tropospheric air cannot be inferred from the composition of glacial snow, as precipitation is not derived from the extremely arid air that makes up most of the tropical free troposphere.