Simple ice phase parameterization

Abstract

A two variable ice parameterization was developed for use in three-dimensional models of cumulonimbus clouds and mesoscale squall lines. Bulk water techniques were employed to simulate the growth and decay of snow crystals and of graupel in order to keep the use of computer resources to a minimum. An externally specified concentration of ice crystals was used to initiate snow. Graupel was assumed to follow the Marshall-Palmer distribution with a constant total concentration. Microphysical growth processes for snow included initiation from the vapor at liquid water saturation, riming, melting, vapor deposition and conversion of rimed crystals into graupel. The graupel microphysical processes that were modeled included raindrop freezing by contact with snow crystals, accretion of raindrops, vapor deposition, riming of cloud droplets and melting. Both types of ice were allowed to precipitate. Sensitivity tests and internal consistency checks on the parameterization were done using a one-dimensional, time-dependent cloud model. Results suggested that the parameterization should simulate adequately the ice phase evolution in higher dimensional models. The parameterization is most suitable for modeling studies in which the major emphasis is on exploring the dynamic consequences of the ice phase rather than exploratory studies in cloud microphysics. Several deficiencies of the parameterization were commented on, specifically: the use of an externally specified snow concentration and its influence on the conversion of snow into graupel. Comments were also made on how local changes in the snow concentration brought about by seeding, ice multiplication and aggregation could be handled in higher dimensional models.