This website works best with JavaScript enabled

logo olive

Understanding Climate Models
(Despite uncertainty, these computer programs show scientists the future)

By Cheryl Pellerin
Science Writer

Washington — Everything scientists know about the future of Earth’s climate comes from models — computer programs that use mathematical equations to describe how the atmosphere, water, land, ice, living things and energy from the sun affect each other and the climate.

In a climate model, the world is broken up into little boxes called grid cells. The cells extend from the North Pole to the South Pole and from the bottom of the ocean up through the stratosphere — so there are cells for the ocean, atmosphere, land and sea ice.

Inside each cell, millions of equations are being solved for things like temperature, precipitation, humidity, heat, winds, carbon dioxide and the effects of Earth’s orbit. Each cell constantly communicates with its neighbors, passing data back and forth, moving wind and sun and rain from cell to cell and around the planet to build a picture of the future.

“The climate model is really a state-of-the-art tool for interpreting and projecting climate change,” Michael Winton, a climatologist at the National Oceanic and Atmospheric Administration’s Geophysical Fluid Dynamics Laboratory (GFDL) in New Jersey, said at a recent briefing in Washington. “This is the way we apply the constraints of science, really — of chemistry and physics — to the global climate problem.”


In the worst of several scenarios for a changing climate system, the Intergovernmental Panel on Climate Change (IPCC) used results from more than 20 climate models to predict that by the end of the 21st century global average temperature could rise by 4 degrees Celsius and sea level could rise by nearly a meter.

How can scientists confidently predict climate conditions more than 100 years from now when local weather forecasts are relatively accurate for only a few days in advance? The answer is that weather and climate are closely related problems but have important distinctions.

“The weather forecasting problem is a matter of taking a careful observation of the atmosphere and trying to move the atmosphere forward in time with a numerical model that encompasses the governing physical laws,” Winton said. “As you go forward in time small errors in the observations and in your numerical calculations tend to grow and spoil your forecast, limiting weather forecast accuracy to a week or so.”

Climate modeling doesn’t rely on careful observations of an initial condition, he said. Instead, the basis of predictability for climate is the global energy balance ( ) — the balance between incoming energy from the sun and outgoing heat from the Earth — which is not important on the short timescales of weather forecasting.


A critical step in climate modeling is making sure the model produces realistic results. Scientists do this by comparing a model’s results with real-world observations and experiences. Models that study future climate change usually start with atmospheric conditions that existed more than 100 years ago.

Scientists let the model run to present day and check results against the current climate. Once a model shows that it can represent present-day properties of the atmosphere, it’s ready to predict future climates.

Climate models are constantly tested against observations and improved, but there is still uncertainty in the results. Aerosols — tiny particles suspended in the air — and the clouds that form around them are a major source of uncertainty. Scientists can’t yet simulate clouds in a climate model, so they do something called “parameterization.”

“This is a way of essentially guessing the cloud field based on the grid-resolved properties of the model — the grid-scale winds and humidities, for example,” Winton said. “It’s a guess that’s based on physical principles and constrained by observations, but it’s a guess nevertheless.”


Today, after some advances in the field, scientists around the world are starting to model the interactions of aerosols and clouds.

“If you have more aerosols, you’ve taken the cloud’s water and distributed it over more drops. It has more surface area, it reflects more radiation and it has a stronger cooling effect,” Winton said. “So for the first time at GFDL and other centers, we’re going to be modeling clouds from scratch. We’re going to be modeling the aerosols from emissions and also the combination of aerosol and water that makes up a cloud drop.”

Another recent advance in climate modeling involves developing regional climate models — models that are not just global in scale but that can focus on a region and give specific climate information for communities.

“The global view is that the world is warm,” Greg Holland, director of the Mesoscale and Microscale Meterology Division at the National Center for Atmospheric Research (NCAR) in Colorado, said at a recent briefing. “And the question you ask yourselves is, what happens to the rainfall in Boulder, Colorado, or what happens to a hurricane in New Orleans?

“Those questions really need to be answered,” he said, “because it’s not just enough to say the world is warming. You need to know how it’s going to affect you locally so you can make the rational planning to be able to do something about it.”

Scientists and modelers from the climate and weather communities are working together, embedding a weather model that now produces regional forecasts into NCAR’s global Community Climate System Model. The combination is beginning to produce credible results, Holland said.

A podcast with scientist Greg Holland on climate modeling ( ) and a photo gallery on climate modeling ( ) are available from

Want to do something about climate change? Join the global conversation on Facebook ( ).

(This is a product of the Bureau of International Information Programs, U.S. Department of State.  Web site:
#fc3424 #5835a1 #1975f2 #2fc86b #f_syc9 #eef77 #020614063440