Duke Engineers Creating ‘More Refined’ Global Climate Model

Frustrated by the limitations of present numerical models that simulate how Earth's climate will be altered by factors such as pollution and landscape modification, Duke University engineers are creating a new model incorporating previously-missing regional and local processes.

"The model we are developing is much more refined," said the project's leader, Roni Avissar, chairman of the Department of Civil and Environmental Engineering at the Pratt School of Engineering.

Unlike previous designs now used by the world's climatologists, Avissar said Duke's model will have a "telescoping capability" to zoom in from global conditions to more localized ones in areas as small as parts of individual states like North Carolina.

The Duke design can thus, for example, model the far-reaching impacts of individual thunderstorms. "These local storms are not very big in size but are extremely powerful in affecting the global atmosphere," he said in an interview. "The current climate models have no capability to simulate those things."

Avissar (http://www.cee.duke.edu/faculty/avissar_r/index.html) currently heads a scientific steering group in charge of advising federal agencies such as the National Science Foundation and the National Oceanic and Atmospheric Administration about research shortcomings in the area of the "global water cycle."

The global water cycle is the term scientists use to describe how water gets distributed around the planet through a cycle of evaporation, transport and precipitation. Pound for pound, water vapor is a more powerful heat-trapping "greenhouse gas" than the carbon dioxide emitted by human activities, according to experts.

Avissar, previously the chairman of Rutgers University's Department of
Environmental Sciences, and founding director of Rutgers' Center for
Environmental Prediction, has done extensive studies on the roles of
water and other environmental factors on climate in tropical forests such
as the Amazon.

"From a global water point of view, that's where the action is," he said of
the tropics. "You modify the water cycle there, and it going to affect the
entire planet."

In the tropics as well as in Earth's more temperate zones, thunderstorms
provide a key influence on water distribution and weather, Avissar said.
For instance, the alteration of worldwide rainfall patterns observed during
El Nino events are triggered by "an increase in thunderstorm activity as a
result of an unusual sea-surface temperature warming in the Pacific," he
said.

Yet thunderstorms are too small and localized to be included in current
global climate models, which work on scales so large that an entire state
is represented by just "one point" in huge worldwide grid, he noted.

By contrast, Duke's new Ocean-Land-Atmosphere Model -- abbreviated
OLAM -- works on multiple scales. "By using a numerical trick to modify
the grid that we use to simulate the planet, we have the capability to go to
a small grid to simulate those thunderstorms," he said. "And we can
understand globally their impact much better.

"So it has this telescoping capability from one scale to the other, to
represent the entire planet as well as have a focus on a given region. If
you want to work regionally, you can. If you want to work globally, you can
do that too. Or you can work with both of them simultaneously."

OLAM -- which also means "world" in the original language of the Old
Testament, Avissar said -- was designed by Robert Walko, a master
programmer and senior scientist at the Pratt School.

Both men were post-doctoral researchers at Colorado State University,
where Walko designed and developed the Regional Atmospheric
Modeling System, one of the most widely-used current models for regions
the size of the Southwestern or Northeastern United States. They later
worked together at Rutgers, and now at Duke.

Another key factor in OLAM's development is a powerful "Beowulf Cluster"
of computers -- a linked group of desktop computers that collectively can
serve as a substitute for a mainframe supercomputer. That cluster is
among several now working around the clock at the Pratt School and
elsewhere at Duke.

While the OLAM project is mostly a product of the Pratt School's civil and
environmental engineering department, other research groups are also
contributing to the model. For instance, a "vegetation dynamics" model
developed by a group now at Harvard, which simulates the growth and
senescence of vegetation communities and their interactions with soils,
water and climate, will soon be merged with "the fluid dynamics
components of the planetary model that we already have," he said.

The Pratt School project has also developed a partnership with ATMET, a
small private Colorado company formed by Avissar, Walko and another
researcher that does meteorological and climatological forecasting.

ATMET "is probably going to use this model for come commercial
applications that are cannot be performed in a university environment,"
Avissar added. "Let's say that you want to forecast how cold the next winter
will be because that affects the coffee market."