With an incoming U.S. President vowing to seriously address climate change, and his cabinet filling with outspoken advocates for such action, the United States, its economy, and its approach to the climate issue are poised to change in profound ways.
The science motivating this political momentum is based largely on global-scale climate models, so these research tools have the very real potential to change the world they are designed to describe. In some instances, they already are doing so.
Yet to most people, including politicians and policy makers acting on the messages these models convey, climate models remain a black box of sorts. Given their importance, though, those who teach, cover, or otherwise act as explainers of climate science to policymakers and regular folks will do well to better understand how global climate models work, and how confident we can be in what they tell us.
Intertwined and stacked cubes blanketing the globe
In simplest terms, a modern global climate model is a computerized grid of millions of mathematically intertwined and stacked cubes blanketing the globe, with each representing a specific spot around the globe. Cubes represent the ocean, land, sea ice, and the atmosphere, and modelers essentially create separate models for each of them and then tie them together to project how they interact to govern the planet’s climate.
Researchers work to program models to calculate as accurately as possible what will happen to temperatures, winds, water currents, and other parameters as appropriate in each cube under various scenarios, for instance a doubling of the current atmospheric carbon dioxide concentration. (The realclimate.org website in early January posted an informative “FAQ on climate models“.) Analyzing and combing data from such calculations allows a model to combine the results for all the individual cubes to project the larger picture of what might happen to the global climate.
Each cube is really just a collection of formulas mathematically describing processes within that area that are relevant to climate. Many of these formulas are based simply on the rudimentary physics that govern motion on a rotating sphere. The laws of gravity, Newton’s laws of motion, and other basics are all tapped to answer such questions as how hypothetical winds will move through cubes, affecting temperatures as they go.
Other formulas needed to calculate a modeled cube’s environment address things Newton likely never pondered. Equations for how sunlight will reflect off a chunk of Arctic ice, warming the air in the cube above it and others nearby, for instance, are based on more recent laboratory experiments. Some important factors, such as how trees might slow a cooling wind, have to be addressed as net effects because attempting to compute the impacts of every individual tree on the planet obviously is impossible.
Clouds, Aerosols Among Modeling Challenges
Much of what will govern the climate at a given point can be well addressed, but other components remain difficult to model. The impacts of clouds as they trap heat and reflect sunlight away from Earth, for instance, are notoriously complex, and clouds’ impacts remain a source of modeling uncertainty. Pollution particles, or aerosols, in the atmosphere can have important cooling effects, but accurately incorporating these impacts remains another modeling challenge. One of the main sources of variation between models from different research groups involves how they incorporate these more difficult components.
Most current models, including the 23 used for the 2007 Intergovernmental Panel on Climate Change report, do not calculate how ocean plankton, trees, and other forms of life process carbon dioxide and other gases to control their fate – whether they end up in the atmosphere providing warming effects or are buried in the ocean or elsewhere. Instead, based on research into how gas cycling occurs, modelers convert greenhouse gas emissions into atmospheric gas concentrations that vary appropriately over time according to the emissions scenario used. The modelers then plug this information into their models.
Earth System Models … ‘The Next Generation’?
Researchers currently are focusing on building new “Earth System Models,” some of which are likely to be used for the next IPCC report. These incorporate biology to generate their own projections for how greenhouse gases will cycle and build up in the atmosphere under a given emissions scenario. Such models are in their infancy, but as they improve they might ultimately reveal important and previously missed controls on climate, such as an unforeseen reduction in carbon dioxide uptake by plankton that would lead to increased atmospheric concentrations.
Modelers also are working to improve the resolution of their models, the number of cubes in the grid that compose a model’s virtual planet and atmosphere. More cubes, just like more pixels in a digital photo, make for a clearer picture. Improved resolution alone does not guarantee a better understanding of climate, but higher resolution, such as incorporating biology, offers the potential to reveal important processes that might otherwise remain unknown.
Studying a Virtual Planet
Once all the equations for the cubes are set, researchers “spin up” a model to run various types of experiments, such as projecting potential future climates based on plausible scenarios specifying factors such as carbon dioxide emissions from human activities. In the case of models used by the IPCC, these scenarios are determined not by modelers, but by social scientists based on studies of population, technology, and other trends. It’s safe to assume the laws of physics will remain constant, but it’s impossible to say what humans are going to do in the future, so filling in these blanks is one of the largest sources of uncertainty in climate modeling.
A model run might begin at some point in the past, say 1860, and run well into the future, often to 2100, calculating a cube’s parameters at specified intervals. Researchers do have to set certain boundaries such as future carbon dioxide concentrations, but they do not plug current and past temperature and other information to fit a model run to the climate that has already been observed. Instead, only initial data for such parameters is set, and the model is then allowed to run on its own. (See article from The Yale Forum.)
Because models are not tweaked to match observations, a key way researchers quantify accuracy involves comparing the trends a model reveals through various runs to what has actually happened. Some subjectivity is involved, for instance in which parameters to compare, but good matches are considered a strong indicator of a model’s reliability in projecting a future climate if a given scenario for human inputs plays out.
Researchers also run modeling experiments where they change some aspect of the past. For instance, they might run a model holding greenhouse gases constant at their 1860 levels to study the degree to which natural factors such as volcanic gas emissions and variations in the sun’s heat output might control climate without any consideration of increased carbon dioxide emissions or concentrations. Work along these lines has repeatedly suggested that observed warming and other climatic trends can be explained only if the human greenhouse gas emissions are included.
One of the main reasons researchers’ confidence in models is increasing, and the IPCC reports are expressing more confidence in model projections despite remaining uncertainties, is that models have grown better and better at matching observed climate. The projections of models that take different approaches to addressing areas of uncertainty are also telling ever more similar stories about the likely future impacts of large greenhouse gas emissions.
Modeling can never be a perfect science, but as many of those involved have pointed out, unless we figure out a way to build planets identical to Earth on which to perform experiments, the virtual planets they describe will remain the best available laboratories for studying future climate change.
Mark Schrope is a freelance science writer living in Melbourne, FL.