1. Introduction

Laboratory study of atmospheric phenomena provides a controlled, steady environment which is usually difficult to obtain in the turbulent atmosphere. Boundary conditions for cloud formation such as homogeneous ice nucleation of water droplets near --40C and cloud droplet formation on hygroscopic particles at less than a few percent above water saturation were first appreciated by careful laboratory experiments many years ago. These ideas are foundation-stones of our understanding of atmospheric clouds. Such ideas came from careful laboratory studies of the behavior of particles under well controlled conditions.

There are different approaches to improve our understanding of things which go on in the atmosphere, and there are different ways of deciding when problems exist and when they do not. We may define `important' problems as those to which other processes are sensitive. Insights come by showing that some things are unimportant as well as showing that some are. Laboratory cloud physics and cloud chemistry are perhaps unique in atmospheric science in that direct, controlled experiments are possible. If we can specify a system well enough, it may be recreated in the laboratory, as a controlled experiment. It may provide an experimental challenge to do so, but that is where the experimentalist's skill becomes paramount. Three types of laboratory experiment are practical: a) Where a specific process is simulated in the laboratory. b) To explore basic chemistry and physics of the process---using a range of parameters well outside of those in the atmosphere. Results are to be set within a physical or numerical model, to provide appropriate curve-fits and boundary conditions for a larger scale context. As an example, the growth of a snow crystal may be simulated in a system reproducing temperature, supersaturation, fall velocity, and air pressure. Physical processes may be examined by growth under zero air velocity and at velocities well above terminal; gases may be added to influence surface structure, scratches or deformation to introduce defects. The radiative field may be modified to simulate the atmospheric environment. c) To physically model the process by, for example, an electrostatic analogy using dimensional arguments or by using other substances.

The overall rationale for the laboratory study lies in the high variability of real systems in the atmosphere, as any cursory look at a cirrus covered sky will confirm. Application of such techniques to the atmosphere evidently needs to be pursued with caution and insight. Some processes are structure insensitive in the sense that cloud droplet nucleation and growth rate depend on nucleus mass and gross thermodynamic properties (vapor pressure, latent heat). Others are structure sensitive---ice nucleation depends on specific sites on a molecular scale which are much more difficult to know about, let alone reproduce. The laboratory can readily simulate conditions---temperature, saturation ratio, radiative flux; particles in that environment grow and interact with each other as in the atmosphere and new phenomena are discovered which may change our view of things overnight. The experimental challenge lies in choice of variables and choosing methods to quantify them together with designing experimental procedures to isolate these variables.

The broader implications of microphysical processes lies in their integrated effect---as precipitation flux, as latent heat release, as scatter and absorption of radiation of both solar and thermal wavelength, resulting in motion and mixing in a turbulent environment. Thus the reality of chemical and physical processes enters the world of the numerical model of cloud processes, in practice as curve-fits to particle spectra. These predict gross features of the radiation budget which drive the global circulation. The reality of the numerical prediction necessarily depends on the input of the dynamics of the process; it depends critically on the microphysical and microchemical input at an early stage. If all cloud droplets froze at 0C, or condensation failed to occur until the water vapor saturation ratio exceeded 400%, the atmosphere would be a very different place.

This review examines progress in laboratory study of aerosol, cloud and precipitation with emphasis on the role of such studies in the broader picture, where gross properties of the atmosphere are sensitive to detail of the microprocesses, both chemical and physical.