Dehydration of the upper troposphere and lower stratosphere by subvisible cirrus clouds near the tropical tropopause

Abstract

The extreme dryness of the lower stratosphere is believed to be caused by freeze‐drying of air as it enters the stratosphere through the cold tropical tropopause. Previous investigations have been focused on dehydration occurring at the tops of deep convective cloud systems. However, recent observations of a ubiquitous stratiform cirrus cloud layer near the tropical tropopause suggest the possibility of dehydration as air is slowly lifted by large‐scale motions. In this study, we have evaluated this possibility using a detailed ice cloud model. Simulations of ice cloud formation in the temperature minima of gravity waves (wave periods of 1–2 hours) indicate that large numbers of ice crystals will likely form due to the low temperatures and rapid cooling. As a result, the crystals do not grow larger than about 10 µm, fallspeeds are no greater than a few cm‐s−1, and little or no precipitation or dehydration occurs. However, ice clouds formed by large‐scale vertical motions (with lifetimes of a day or more) should have fewer crystals and more time for crystal sedimentation to occur, resulting in water vapor depletions as large as 1 ppmv near the tropopause. We suggest that gradual lifting near the tropical tropopause, accompanied by formation of thin cirrus, may account for the dehydration.

A conceptual model of the dehydration of air due to freeze‐drying by optically thin, laminar cirrus rising slowly across the tropical tropopause

“…The ice number densities predicted in these simulations are much lower than those reported in our earlier modeling study [Jensen et al, 1996]. The difference arises from two factors: First, in the previous study we were using an ice nucleation parameterization which predicted much lower values of t•Hlnuc, resulting in higher ice number densities (see Figure 3).…”

“…[Koop et al, 1998]. The formulation predicts aerosol freezing at ice supersaturations of about 60% at typical tropical tropopause temperatures (5 -75øC) ß The primary difference between the model used here and that used in our earlier study [Jensen et al, 1996] is that we previously used an ice nucleation model that predicted sulfate aerosol freezing at much smaller ice supersaturations (<10%).…”

“…The efficiency of dehydration by cloud formation depends upon several coupled factors including the number of ice crystals nucleated, how large the crystals grow, the fall speed of the ice crystals, and cloud lifetimes [Jensen et al, 1996]. A range of scenarios are possible: If the ice crystals do not grow large enough to fall relative to the rising air, then they will simply sublimate as the air rises into warmer regions in the stratosphere, and no net dehydration will occur.…”

“…In a previous study, we modeled the potential for dehydration due to ice cloud formation driven by gravity waves at the tropopause [Jensen et al, 1996]. In this study, we use a microphysical-dynamical model to evaluate the potential for freeze-drying of air rising slowly across the tropical tropopause.…”

Aircraft observations of thin cirrus clouds near the tropical tropopause

“…The temperature variations were several degrees peak-to-peak, implying a more than 2 ppmv variation in the saturation mixing ratio; this, in turn is .•70% of the typical value of the water vapor mixing ratio at the tropopause in northern hemisphere winter. The wave periods, varying from 4 to 40 hours for cases isolated by the analysis, are long enough to allow large (10 microns or more) ice particles to form and fall out, thus effectively dehydrating the tropopause region [Jensen et al, 1996]. If these waves are widespread in the tropics (and observations [e.g., Karoly et al, 1996] indicate that they are) the implication is that they could effectively "ratchet down" the water vapor content in the tropopause region to values below those implied by synoptic-scale analyses at the tropopause or, for that matter, averages of tropopause cold point saturation mixing ratios from radiosondes [Dessler, 1998].…”