How do Scientists Model the Climate?

Earth’s climate system is complex. The system comprises interactions between Earth’s landmasses, the oceans, the ecosystems, the atmosphere, and even the sun. Despite those complexities, one alarming fact remains clear: the climate is changing, and is projected to continue changing. Projections suggest the melting of the polar ice caps, rising sea levels, and extremes in rainfall patterns. But how are scientists able to make these projections? How do scientists even begin to approach the daunting task of modeling and making predictions about such a large, complex system like Earth’s climate? 

As with any large task, they break it up into smaller tasks. Climate models divide the Earth into a grid, typically 100 kilometers on either side horizontally (about the size of Jamaica), and 1 kilometer vertically. Next, scientists input measurements of important factors into each grid; things like temperature, wind, and humidity.. Finally, scientists use the known laws of physics to solve for how the totality of these factors alter the climate system with time. Since such a task is much too complicated for any human to do with pencil and paper, scientists instead write computer programs on some of the most powerful supercomputers in the world–these programs then crunch the numbers. For instance, the climate science supercomputer at NASA, named Discover, can perform 6,800 trillion calculations per second.

But like with any model, climate models come with uncertainties and room for improvement. Breaking up the Earth into grids makes the model fuzzy, and blurs relevant effects that occur on scales smaller than the size of grid units. For example, cumulus clouds are known to impact the atmosphere, but they typically have a size on the scale of 1 kilometer, meaning a grid size of 100 kilometers is too low a resolution to represent them. To help alleviate this issue, scientists consider quantities that are simpler to represent, like temperature and humidity, and use these to hopefully reconstruct how clouds impact the atmosphere. In this way, the effects of clouds get encoded into the calculations, without clouds explicitly being included in the model. Climate scientists call this indirectly measured effect a ‘parameterized effect’, because it is calculated in terms of other parameters. It is thought that better understanding of how to best represent parameterized variables, as well as decreasing grid size and access to more powerful supercomputers, can reduce the uncertainties in global climate models.

These kinds of climate models allow us to think in terms of probabilities, to estimate trends in the climate system, and to calculate the risks inherent to global climate change. If you had a bag full of different colored marbles, then you can be quite sure that by adding, say, blue marbles to the bag, you increase the chance of picking out a blue marble, even though you can’t know for sure what color marble you would pick out on any given draw. Climate models work similarly, in that they make it clear that adding more greenhouse gases to the atmosphere and heating the planet increases the risk of catastrophic climate change, without having to know with certainty what the climate will be like in any given year. In this way, climate models represent our best efforts to inform leaders and the public of the risks we run by continuing to emit greenhouse gases. 

***

For further reading on climate modelling:

https://www.climate.gov/maps-data/primer/climate-models

For further reading on parameterized effects, see the first few pages of:

https://curry.eas.gatech.edu/Courses/6140/ency/Chapter8/Ency_Atmos/Parameterization_Phys_Processes_Clouds.pdf

For further reading on grid size: 

http://eo.ucar.edu/staff/rrussell/climate/modeling/climate_model_resolution.html

For further reading on Klaus Hasslemann and Syukuro Manabe, the two scientists who were awarded the 2021 Nobel Prize in Physics for their work on climate models: https://www.nobelprize.org/uploads/2021/10/popular-physicsprize2021.pdf

***