HPC Climate modelling
Simulations laboratory) at the University
of Central Florida (UCF) is concerned with
climate modelling. It is investigating coastal
hydroscience and the related issue of how
systems such as hurricanes bring water onto
land in and around the state
of Florida.
The CHAMPS Lab is currently working
on a model to define the extreme flooding
scenarios that could occur once every 100 to
500 years. The supercomputing system deals
with around 850,000 computation points
per simulation and maps the landscape every
second over a three- or five-day period, as
Scott Hagen, associate professor of civil
engineering at UCF, explains: ‘Because we
have such a small time step our matrices need can scale the problem by making it larger. the threat of severe weather (high wind,
to be solved so many times. Being able to Many weather applications, especially on the hail or tornadoes). The development and
break that model down into interconnecting research side, can do this.’ movement of storms, however, is often steered
subdomains means we can do the simulation As part of a multi-institutional by the larger-scale flows, so we need to get
in terms of hours instead of days, thanks to our collaboration, the NCAR developed the that correct as well. Forecasts of more than
HPC facilities.’ WRF model, which is in wide use for one or two hours require a large domain, as
The team uses high-resolution LIDAR operational weather forecasting and for individual storms and weather systems in the
images and puts that data into a finite element atmospheric simulation. WRF-generated active spring season can travel quite fast (up
model, which is made up of interconnected forecasts are, amongst other applications, to 30 m/s).’
triangles to map the water and land’s surface repackaged as commercial forecast products But the team is modelling to even higher
and elevation. The next step for these models and then seen on nightly television in more resolutions, as Brewster explains: ‘For
is to improve the resolution further, as Hagen than 200 markets around the US. NCAR research, we are modelling the creation
adds: ‘Our present models use 850,000 scientists are currently running simulations and development of tornadoes within
computational points and we are limited of processes within the hurricane eyewall at thunderstorms. For that we are using the
by the smallest size of around 40 metres for resolutions of 60 metres. model run at grid resolutions as small as 20
each triangle’s side. It would be great to get Michalakes adds: ‘But it’s important metres. These cannot be done in real-time
down to the metre range to bring down the to realise that, while the amount of but they are very computationally demanding
approximation but a scale of less than five computations possible in a given amount of – so, to get turnaround in a reasonable length
metres would increase the number of points to time on the system increases, the speed of the of time, days not months, they must run on
make this kind of work infeasible at present.’ run does not necessarily increase. So weather HPC resources and run efficiently.’
‘One area that does require improvement is and climate applications that have a time to
working out the extent of the inland flooding solution requirement – real-time forecasting Tip of the iceberg
in localised areas. For example, you might or very long climate simulations – may not While such simulations are, of course,
want to zoom into a small area and see how benefit from conventional petascale systems. helping us understand the planet’s climate,
it will be affected by such flooding – this is For these applications, faster processors, not there does seem to be something a little
where the five-metre mark would become just more of them, are needed.’ counterintuitive about using power-hungry
important.’ servers to try to calculate how draining the
But, as the need for more HPC hardware Oklahoma! Earth’s resources has affected its climate. Last
increases, John Michalakes, lead software The Center for Analysis and Prediction year, the UK’s Met Office spent £33m on a
developer for Weather Research and of Storms (CAPS) at the University of new supercomputer to calculate how climate
Forecast (WRF) model at National Center Oklahoma is developing thunderstorm-scale change will affect Britain – only to find the
for Atmospheric Research (NCAR), warns numerical weather predictions. That is, new machine has a giant carbon footprint of
that researchers must box clever: ‘Today, models using resolutions of 3km or less. its own, emitting around 14,400 tonnes of
numerical weather prediction (NWP) is Keith Brewster, senior research scientist CO a year, which is equivalent to the CO
2 2
a terascale (10
12
floating point operations and associate director at CAPS, explains: ‘In emitted by 2,400 homes, according to reports.
per second) application. The current trend 2008 we did some real-time forecasts at 2km, So, as climatologists demand increasingly
in HPC is to reach petascale (10
15
floating and some 1km grid scale forecasts locally in more power from their supercomputers, could
point operations per second) and beyond by near-real-time. The high resolution allows turning the machines off ultimately be the
essentially building larger and larger clusters, us to explicitly model the flow within the way to curb the very effects they are trying to
which is okay for applications where you thunderstorm; this allows us to better gauge understand?
36
SCIENTIFIC COMPUTING WORLD february/march 2009 www.scientific-computing.com
SCWfeb09 pp34-36 weather.indd 36 4/2/09 10:44:41
Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44