"The physics in regional models is much better than in the global climate models," says Dr. Gregory S. Jenkins, assistant professor of meteorology at Penn State. "Regional climate models might provide more useful information for planners."
Global climate models use about 30 data points for the continental United States, while regional models use about 800 data points. Regional models represent mountains better than the global models. While mountains are important for climate and weather, global models smear them, making something like the Rocky Mountain chain insignificant. Global models also have difficulty defining coastlines, but the regional models can resolve the coastal area.
"In the Northeast, major storms come up the Atlantic coast and it is important both for weather forecasting and for climate studies," says Jenkins.
The Penn State researchers are using a global climate model called Genesis, developed by the National Center for Atmospheric Research, and a regional climate model developed by Penn State and NCAR to investigate the climate change differences found when modeled on a planetary or regional scale.
"We want to see the changes we get with elevated greenhouse gases in variables like precipitation, soil moisture and snow depth," said Jenkins today (Feb. 17) at the annual meeting of the American Association for the Advancement of Science. "We also want to understand how the variables in the two models differ over various regions within the United States."
The regional model Jenkins uses was originally designed for weather forecasting over short periods but has been modified to investigate long-term climate. For climate modeling over seasons and years, vegetation and local variations become important.
"We seem to get the same results with some variables, like temperature, with both models," says Jenkins. "But soil moisture, for example, increases in the global model, but not in the regional model."
The researchers, who also included Giang Nong, graduate student in meteorology and John Haynes, undergraduate in meteorology, analyzed the global and regional models for twice today's carbon dioxide levels. Overall, the global models produced consistently high precipitation.
However, the regional model does a better job at producing rainfall amounts similar to observation. The two-times carbon dioxide regional climate simulation for the Northeastern United States produced warmer temperatures, leading to more rain and less snowfall during winter. Less snow means less snowpack for spring melting.
"This type of result is important for water balance," says Jenkins. "The snowpack is important because snow melts slowly, rather than running off immediately."
The regional models rely on the global models for input. Problems arise when interpolating to increase the number of data points 30 times.
One way the researchers tested the models was to try to simulate present-day extreme events, like El Nino or the drought of 1988.
"The Genesis model, run with observed sea surface temperatures over 10 years, produced some correlation with real events for the 1986-87 El Nino, but precipitation for both models was, in some places, incorrect," says Jenkins. "Precipitation looks like a simple variable, but it is really complex."
The simulation also had difficulty with modeling drought. Both models incorrectly simulated the 1988 summer drought conditions over the central United States.
"These coupled global and regional models can not give the best results yet because coupled atmosphere-ocean models which provide boundary conditions for the regional climate model are still early in their development," says Jenkins. "But eventually the coupled atmosphere-ocean models will show us how extreme events like El Nino can moderate or exacerbate the effects of global warming. This will modulate our understanding along with regional climate model results."
For example, normally the Southwestern United States has relatively dry winters, but during an El Nino, winters become much wetter. If the current climate models predict that greenhouse warming increases Southwestern aridity, then the models of the future, which can predict El Nino events, might show an El Nino year as a welcome occurrence, mitigating the effects of the warming trend.
EDITORS: Dr. Jenkins may be reached at (814) 865-0479 or osei@essc.psu.edu by email.