Formation of ice in the atmosphere has long been recognised to be a topic of great importance due to its key role in the precipitation process. However, progress in elucidating this role has been plagued by a host of complex issues such as the precise mode of action of ice nuclei (IN) (e.g., Cooper, 1980), in situ modification of IN activity by various substrate coatings, including residual ice (Borys, 1989; Curry et al., 1990; Rosinski and Morgan, 1991; Beard, 1992; Vali, 1992), secondary ice production (e.g., Beard, 1992) and a lack of consistency in measurement techniques (cf., Bigg, 1990; Vali, 1991; Rogers, 1993; Pruppacher and Klett, 1997).
Because of these issues, it is premature to quantitatively assess the impact of ice formation on indirect forcing. Instead, the potential importance of such formation is given a preliminary assessment by addressing a set of four fundamental and serial questions whose answers would, in principle, yield the desired assessment. A summary of what is known with respect to these questions is given below.
Does ice formation have an impact on radiative forcing?
In principle the phase partitioning of water in clouds should have a substantial impact on cloud radiative forcing, first because the ice hydrometeors will tend to be much larger than cloud drops and thus increase precipitation, and second because the size of hydrometeors (determined by both ice/vapour and ice/liquid partitioning) can have a significant impact on radiative balances. Several GCM sensitivity studies have supported these expectations (e.g., Senior and Mitchell, 1993; Fowler and Randall, 1996) by demonstrating that the fraction of supercooled water in the models which is converted to ice has a significant impact on the global radiative balance. A simple sensitivity study with the ECHAM model (cf., Lohmann and Feichter, 1997) conducted for this assessment revealed a very large difference in globally averaged cloud forcing of +16.9 Wm-2 induced by allowing only ice in clouds with temperatures below 0°C as compared with allowing only water in clouds with temperatures above -35°C. The liquid-water path change in these experiments (160 gm-2 to 54.9 gm-2) was larger than the 60% uncertainty in this quantity from measurements (Greenwald et al., 1993; Weng and Grody, 1994). However, this certainly demonstrates that even small changes in ice formation could have a significant impact on the indirect climate forcing due to aerosols.
Is formation of the ice phase modulated by aerosols?
The relative roles of different types of ice nucleation in cirrus clouds is very complex (e.g., Sassen and Dodd, 1988; Heymsfield and Miloshevich, 1993, 1995; DeMott et al., 1997, 1998; Strom et al., 1997; Xu et al., 1998; Martin, 1998; Koop et al., 1999). Presumably there is a temperature-dependent transition from predominantly heterogeneous nucleation (i.e., initiated at a phase boundary with a substrate - the heterogeneous nucleus) to homogeneous nucleation (i.e., within the liquid phase alone - no substrate required) that depends on the chemistry and size of the precursor haze drops of the homogeneous process as well as on the chemistry and concentration of the heterogeneous nuclei. Supersaturations with respect to ice in excess of 40 to 50% are necessary to freeze sulphate haze drops, even at quite low temperatures. Far lower supersaturations will be adequate if hetero-geneous IN are present. There is thus a large supersaturation range in which heterogeneous IN could have a significant impact (Fahey et al., 1999). However, in both heterogeneous and homogeneous cases, aerosols play an important role in glaciation.
For lower level, warmer (though still supercooled) clouds, in principle the answer to our query is necessarily positive. In this vast, liquid-water reservoir, temperatures are simply too warm for homogeneous freezing of cloud drops to occur and aerosol surfaces of some sort must provide the substrate for ice initiation. However, prolific secondary ice formation due to such processes as the Hallett-Mossop mechanism renders the establishment of a clear relationship between measured IN concentrations and ice particle concentrations quite difficult (cf., Beard, 1992; Rangno and Hobbs, 1994). Nevertheless, in some instances relationships between IN concentrations and ice formation in lower-tropospheric clouds have been found (e.g., Stith et al., 1994; DeMott et al., 1996). This lends credibility to the view that the actual ice initiation process must be modulated by aerosols.
For both upper and lower level clouds in the troposphere, it seems clear that the ice initiation process is dependent on aerosols, though the nucleation process can proceed via different pathways and from a variety of different nucleating chemical species.
Continues on next page
Other reports in this collection