MOZART
A 3D Global Chemistry Simulation

"This model implies the absolute necessity for a Mass Storage System of some sort, and is impractical now on anything but a supercomputer."
--Stacy Walters, MOZART software developer.

Trace gases:
To what complicated chemical fate?

NCAR's Atmospheric Chemistry Division has developed a new, highly refined chemical transport model called MOZART (Model of Ozone Research in the Troposphere). This model simulates the three-dimensional behavior and seasonal evolution of 41 trace gases, including the most important species that determine the oxidation capacity of the atmosphere. MOZART helps researchers analyze the complicated chemical fate of these trace gases in response to surface emissions, long-range transport, convection, washout, and dry deposition.

Researchers are comparing simulation results with observed data from the surface, aircraft, and satellites; differences between model output and observed data indicate gaps in the current understanding of atmospheric chemistry. Researchers are also using MOZART to help plan and support field campaigns.

Because perturbations in the chemical composition of the atmosphere are a potential cause of global changes in the earth system, the ultimate goal of scientists using MOZART is to understand how natural and anthropogenic processes will affect the climate of the future.

For Supercomputers Only

MOZART is a highly vectorized, multitasked code that typically runs at 95% efficiency. It runs on all the CRAY platforms at NCAR: the Y-MP8/864, the Y-MP/8I, and the J9s.

A unique characteristic of the model is that it integrates a vast amount of information. Because it is a fully diurnal, three-dimensional, off-line simulation, it requires input of dynamic variables such as winds, temperatures, and precipitation many of which are output from NCARs Community Climate Model 2 (CCM2). A full year's simulation reads approximately 2.5 gigabytes of input data; thus, a 100-year simulation would input 250 terabytes of data. This requires a powerful Mass Storage System.

With 41 trace gases, the chemical transport model is more compute intensive than climate models such as NCARs CCM2. Running on the CRAY Y-MP8 in parallel mode at over one gigaflop, MOZART takes 200 CPU hours to complete a one-year simulation for a T42 spatial grid encompassing 144,000 points for the 41 trace gases; three-year simulations are typical. This makes the model impractical on anything but a supercomputer.

Visualize!

Building a sophisticated atmospheric chemistry model is only the first step in a project of this magnitude. Researchers must also verify that the code runs correctly and produces reasonable output (e.g., are all chemical processes included and modeled accurately?). This cannot be done by poring over hundreds of gigabytes of printed output. The only way to obtain a global understanding of the models output data is with visualization tools.

Accordingly, ACD has developed a strong and flexible set of visualization tools for MOZART. With these tools, scientists can visualize data in MOZART's 3D field in as many as ten different ways, retrieving specific information from model output stored on MSS tapes.

Return To Exhibit Home Page