Featured Researcher
UC Los Angeles: Verica Savic-Jovcic
Dynamics of Precipitating Stratocumulus
Computational resources of SDSC and SCD at NCAR are utilized in a research of an impact of precipitation on dynamics of the stratocumulus-topped boundary layer (STBL) and organization of stratocumulus clouds. The work is performed in the close collaboration with my advisor, Professor Bjorn Stevens at UCLA.
Stratocumulus play an important role in determining planetary albedo. They cover large areas in the subtropical regions and are quite bright. Our recent work suggests that precipitation (and hence factors which influence it) might be an impetus in formation of the patches of almost clearings within these vast cloudy decks. Figure 1 presents a planform view of cloud albedo as represented by large-eddy simulations for various drizzle strengths, as well as no drizzle at all. We see that in absence of drizzle, cloud has quite high value of albedo, while with increasing strength of drizzle, the albedo values are reducing.
Figure 1: Albedo at the end of the 6th hour for three simulations with various strengths of
precipitation from the stratocumulus.
Large-eddy simulations provide a valuable laboratory for evaluating theoretical ideas and discovering new phenomena. However, they are very demanding in computer resources. For instance, our work is based on the simulations on a grid with 516 x 516 x 97 points, which corresponds to the domain of 25 x 25 x 1.5 km3 . We chose such a large domain to allow for the development of the larger scales we assumed to form when additional forcing (in our case precipitation) is introduced, and to, therefore, allow for calculation of stable statistics. With time step of only a half a second, necessary for the numerically stable simulations, the calculations require parallel computer systems with large number of processors to provide the results for 6-h long simulations in any reasonable time frame. DataStar at SDSC and bluesky at NCAR are two such computers we are able to use, and for each experiment we utilize 64 processors on 2 nodes to perform the simulation. Using these two supercomputers gives us an opportunity to explore the dynamics in more details in shorter time frame. The ability to shorten the time of the experimental phase of my PhD research is very important for me and I am grateful for the given opportunity. SDSC is in particular significant in my work, because it is a part of UC system and it grants me an access to its resources as long as I am a graduate student at UCLA.
In addition to DataStar, at SDSC we also use HPSS to store the large amount of data our simulations produce. The data storage is very important as only one of our simulations produces more than 100 GB of data we are analyzing. Ability to take just pieces of data to analyze and keep the rest in a secure, but easily accessible place is of great importance for us, especially as we are planing to do more experiments to study the how drizzle introduce the changes in cloud structure presented in the Fig. 1.
More on my PhD work investigating the dynamics of precipitating stratocumulus can be found at www.atmos.ucla.edu/~verica/research/.
