+ Home > Virtual Snowflake Project

Virtual Snowflake Project

Virtual Snowflake Project

SSSO is working with investigator Dr. Kwo-Sen Kuo to model the growth of snowflakes as part of his research into retrieval of precipitation and cloud microphysical properties. Although the parameterized model for the growth of snowflakes is of relatively modest complexity, the high spatial resolution (1000x1000x250) and large number of time steps (> 100,000) require high-end computing to reduce the time-to-solution to practical levels. Performance limitations have led other researchers to restrict their simulations to fully symmetric crystals which reduces the workload by a factor of 24, but prevents study of a much wider variety of crystals. In addition to parallelizing the basic snowflake model, SSSO is assisting Dr. Kuo to partially automate a survey of parameter space for crystal growth. This survey will produce a database of representative snowflakes which will then be used as input to another software package to synthesize realistic instrument retrieval data.

The image above shows a 'snowfake' during the middle of its growth phase.