If we ever want to be able to properly include radar measurements into numerical models, we absolutely need to:
Better characterize and describe the evolution of precipitation such as condensation/deposition of vapor, coalescence/aggregation of rain/snow particles, etc;
Find ways to convert reflectivity information from rain, snow, and melting snow into quantities that models can use (average number concentration, mass, other moments of the size distribution...);
Better parametrize how these model quantities vary as a result of these processes.
This task has both a data analysis component and a model design component. Long term observations of snow and rain made by radars and particle imagers are used on the analysis part. New parameterization and model coding (three-moment microphysics) is then needed to make use of that information. We work on all three facets of this activity, and there is much to do because not doing an adequate job on any will result in a failure in our attempts to use radar information in models. After having set the modelling framework (Szyrmer et al. 2005, Laroche et al. 2005), and studied melting snow (Heyraud et al. 2008 and companions), we are currently attacking the snow problem.
Snow microphysics work (theory and measurements)
The main objective of this work was to obtain a parameterization of snowflake size distributions that accurately describes the relationship between the moments of the particle size distribution (PSD). This is important as most (measurable) bulk quantities in precipitation microphysics are generally (almost) represented by PSD moments, and hence the importance of such a scheme for Numerical Weather Prediction and Data Assimilation.
For this purpose, ground measurements of snowflake size and velocity collected using an optical disdrometer during multiple precipitation events in the winter of 2005/2006 at the Centre for Atmospheric Research Experiments site as part of the CloudSat/CALIPSO Validation Project were analyzed.
First, an empirical relationship between snowflake terminal velocity and its size was derived, finding that the main factors determining this relation are the depth of the precipitation system and the surface temperature during the event. Also, a thorough study of the variability of the snowflake velocity measurements was performed. Then, combing the velocity-size relationship with the relations between the Best and Reynolds numbers available in the literature or derived from theoretical considerations, the velocity-mass relation was obtained. Finally, the measured snowflake size distributions were normalized using one or two moments of the distribution, and the validity of these formulations was investigated considering the uncertainty of the estimated moments.
References:
Zawadzki, I., E. Jung and G. Lee, 2010: Snow studies, Part I: A study of the natural variability of snow terminal velocity. Journal of Atmospheric Sciences, in press.
Szyrmer, W. and I. Zawadzki, 2010: Snow studies, Part II: Average relationship between snowflake mass and its velocity. Journal of Atmospheric Sciences, submitted.
Szyrmer, W. and I. Zawadzki, 2010: Snow studies, Part III: Parameterization of snowflake size distributions within a scaling normalization framework. Journal of Atmospheric Sciences, submitted.