|
DRI’s
veteran atmospheric modeler anticipates new cluster capabilities
|
Visualize this.
Dr. Darko Koracin with his “freezer box” SGI Origins
2000 high performance computer system that he has used to
develop and run models visualizing air pollution dispersion
for the planned Ivanpah Airport south of Las Vegas and for
Washoe County’s air quality management plan as well
as other environmental projects. (Photo by John Doherty) |
No field of science is
benefiting more from high-end computing’s ability to model
reality than the study of the fluid, invisible processes of the
atmosphere. DRI’s “veteran” atmospheric modeler
is Dr. Darko Koracin, an atmospheric physicist in DRI’s Division
of Atmospheric Sciences. Koracin has been the lead or co-lead researcher
on more than 30 projects employing modeling applications that have
brought several million dollars in research funding to DRI during
the past decade.
More than two years ago,
Koracin acquired a 16-processor SGI Origins 2000 “Super-MicroComputer”
to support his work. The SGI machine, state of the art when it was
new, uses a “shared memory” rather than a network communicating
among its components. Koracin shares its capacity with other DRI
scientists, but he is eagerly anticipating the development of the
new computing clusters at DRI that will enable him to build and
operate even bigger, more sophisticated models.
| 
If seeing is believing . . .
Koracin’s Lagrangian random particle dispersion model
can predict the movement of pollutant plumes from Sacramento
and the larger California-Nevada region in addition to the
Truckee Meadows. This graphic is part of an animation of the
Sacramento and Reno air plumes on April 21-22, 2001. It can
be viewed on DRI’s web site http://www.dri.edu/Projects/Modeling/realmm5.html |
“In principle,
performance of shared-memory computers is better than that of distributed-memory
computers (if all components are the same),” Koracin says,
“but they have scalability limitations, and are so expensive
while networking capabilities have come so far that today everyone,
including DRI, is turning to clusters for affordable and powerful
high performance computing.”
Recently, Koracin used
his SGI machine to simulate air quality conditions and predict pollutant
dispersion for Washoe County’s Air Quality Management Division.
The model incorporates much of California and Nevada in its predictions.
Prior to that, he constructed an air dispersion model for the planned
Ivanpah freight and charter airport terminal south of Las Vegas.
While continuing his
work on DRI computers to develop better models and apply them to
meteorological and air quality challenges, Koracin says he and his
colleagues are also able to link to the supercomputer at the National
Center for Atmospheric Research in Boulder, Colorado, for even greater
computing capacity. This capability is important for extremely large
projects that might involve federal researchers and scientists from
other universities.
With each advance in
computing capability, Koracin and his DRI colleagues can ask bigger,
more complex questions about environmental processes and how humanity
influences them. And with new and better technology, the answers
will come faster than ever.
|
