PerspectiveATMOSPHERE

When Dry Air Is Too Humid

See allHide authors and affiliations

Science  01 Dec 2006:
Vol. 314, Issue 5804, pp. 1399-1402
DOI: 10.1126/science.1135199

As moist air rises to colder regions in the atmosphere, the humidity rises above its equilibrium value over ice. To relax this metastability, the air releases its water vapor via ice cloud formation. Such atmospheric ice clouds form in two steps: First, ice nucleates in or on existing aerosol particles; second, these ice particles grow through condensation of supersaturated water vapor onto the ice surfaces. Recent field observations (13) call into question the basic principles underpinning the current understanding of ice cloud formation and alter the assessment of water distribution in the upper troposphere.

The governing quantity for nucleation and growth is the excess activity relative to the equilibrium humidity over ice, also called ice supersaturation (expressed as a percentage). The equilibrium humidity decreases strongly with falling temperature. Hence, when an ascending air mass cools, it can become supersaturated with respect to ice. Ice nucleation requires a supersaturation above a critical threshold value. Nucleation can occur homogeneously from aqueous solution droplets, or heterogeneously on particles known as ice nuclei. At upper-troposphere temperatures, homogeneous freezing sets in at a supersaturation of ∼60% (4); lower supersaturations are sufficient for heterogeneous nucleation. After nucleation, vapor molecules condense onto the ice particles, causing them to grow and the gas phase to become depleted in water until equilibrium is reached (see the green curves in the figure).

Large-scale regions of persistent supersaturation up to 60% outside ice clouds are not unexpected in the absence of ice nuclei. Yet values even above 100% have been observed in cloud-free regions (1) (red curves in the figure). These values are far above the critical value for homogeneous ice nucleation (5) or cloud chamber data (6).

At least as puzzling are supersaturations of 30% reported to persist inside ice clouds and contrails (2) for at least 1 hour of aircraft measurement time (3) (orange curves in the figure). Such large supersaturations are expected to relax rapidly as a result of fast vapor condensation (7) unless continuous cooling remains sufficiently strong. To achieve such cooling, the clouds would have to rise by several kilometers in the measurement time, which contradicts the observations.

Measuring water in the upper troposphere is difficult. A major international effort to assess water vapor measurements in the upper troposphere and stratosphere concluded that, on the basis of laboratory calibrations, typical mean accuracies of aircraft and balloon instruments were on the order of 10% (8). However, direct comparisons in the upper troposphere suggest that differences between various instruments on aircraft and balloons often exceed 25%, especially when temperatures are very low (9). Also, balloon-borne instruments appear to yield mostly lower supersaturations than do aircraft instruments (10). Nonetheless, large supersaturations were observed during all recent aircraft and balloon campaigns; these studies used a range of instruments based on different measurement principles (11). Hence, only a fraction of the observed supersaturations can be ascribed to instrumental inaccuracies.

The theoretical assumptions underlying modeling of ice cloud formation should also be reassessed. How can we explain ice nucleation in light of the extreme supersaturations observed in cloud-free regions, and ice growth in light of the persistent supersaturations observed within dense ice clouds?

Outside ice clouds, the supersaturation at which ice nucleation occurs depends on the equilibrium vapor pressure of supercooled water. Measured data only exist down to ∼235 K and must be extrapolated to lower temperatures (12), possibly leading to errors of up to 20%. Moreover, although air masses appear to always contain sufficient numbers of aerosol particles for cloud formation, the composition of these aerosols might inhibit ice nucleation. The homogeneous ice nucleation threshold of 60% was established for atmospherically relevant salt solutions and sulfuric acid, but only for a few organic species. Cloud chamber data indicate that aerosols containing only organic and elemental carbon may be almost completely unable to nucleate ice (13). Alternatively, if water-rich aerosols were fully covered with organic surfactants, nucleation might be suppressed if it started preferentially at the surface (14). In field experiments, the presence of organic pollutants has indeed been associated with impeded ice particle formation (15). However, laboratory data of surface nucleation and field data on particle composition and surface morphology of upper tropospheric aerosols are too limited to allow any conclusions to be drawn.

Inside sufficiently dense ice clouds, condensation should rapidly reduce vapor pressures in excess of the equilibrium vapor pressure of ice. However, errors in this equilibrium value might lead to perceived supersaturation. Laboratory data show (16) that below 200 K, cubic ice (a metastable form of ice) nucleates first and might persist in clouds. The equilibrium vapor pressure for cubic ice is ∼10% higher than that for stable hexagonal ice (17).

But even after an ice crystal has nucleated and transformed into hexagonal ice, surface effects might hinder its growth. It is usually assumed that most water molecules hitting the crystal are built into the lattice, but recent laboratory data indicate that this is the case for fewer than 10% (18) or even as few as 0.4% (19) of water molecules. Furthermore, gas-phase species such as nitric acid may selectively block the growth sites on ice crystals (2), although this hypothesis needs support from laboratory studies.

Rising air and ice cloud formation.

The top three panels sketch three scenarios for the formation of ice clouds along an ascending air parcel trajectory; the bottom panel sketches the effect of these scenarios on supersaturation. According to conventional understanding, ice particles nucleate (star), grow, and reduce the supersaturation (green curves). Recent observations suggest suppressed nucleation (red curves) or suppressd growth (orange curves) in large parts of the atmosphere.

Further reasons for the observed supersaturations may be related not to the properties of individual ice particles, but to effects on larger scales. The presence of ice nuclei in cloud-free air may initiate ice nucleation below the homogeneous-nucleation threshold, leading to clouds with low ice particle number densities, in which supersaturations might be sustained for relatively long periods. Also, conditions within clouds might fluctuate faster than aircraft-borne instruments can resolve, causing apparent supersaturation by averaging over glaciated and ice-free supersaturated patches. Verification of such effects will have to await higher-resolution instrumentation.

None of these hypotheses is likely to be the sole explanation for the observed high supersaturations outside and inside ice clouds. The uncertainties in the expressions used to retrieve the supersaturation from the data must be resolved and their usage clarified (12). Uncertainties in the aircraft and balloon data must be determined accurately. In addition, mesoscale meteorological fluctuations must be characterized, theories of ice nucleation and growth must be reassessed, and state-of-the-art numerical models must be compared. Six years after a major effort to characterize the distribution of upper atmospheric water (8), the issue is again open, and, because of its climatic importance, more pressing than ever.

References and Notes

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
View Abstract

Stay Connected to Science

Navigate This Article