New to Climate Change?
View the MIT Climate Primer
Explainer

Climate Models

Climate models are computer programs that simulate weather patterns over time. By running these simulations, climate models can estimate the Earth’s average weather patterns—the climate—under different conditions. Scientists use climate models to predict how the climate might change in the future, especially as human actions, like adding greenhouse gases to the atmosphere, change the basic conditions of our planet.

Under the hood of a climate model

To simulate weather, climate models must reflect real properties of the Earth’s climate, including physical laws like the conservation of energy and the ideal gas law. They also include variables like air pressure, temperature, and wind. All of these are expressed as equations that a climate model must solve. Solving the equations produces a three-dimensional picture that shows natural climate patterns in action, like rainfall, ocean currents, and the changing of seasons.

Climate models agree on many important facts about our climate. For instance, models reliably show that adding more greenhouse gases to the atmosphere will cause average temperatures to rise. Models also try to predict how climate change will affect rainfall, sea levels, ice cover, and other parts of the natural world.

 

Principles, variables, and parameterizations

The dynamics of the climate system are governed by seven physical principles:

  • conservation of air mass
  • conservation of water mass
  • conservation of energy
  • conservation of momentum of air in three directions
  • the ideal gas law applied to air

Climate models describe these principles as seven equations, which constrain seven variables:

  • air temperature

  • pressure

  • density

  • water vapor content

  • wind magnitude in three directions

By solving the equations, climate models can simulate all these variables in three dimensions and in time.

Other variables that affect the Earth’s climate are hard to model directly. Clouds are a good example: a cloud is much smaller than the smallest unit of distance in a typical climate model, so the model cannot “see” individual clouds, but taken together they have big effects on the Earth’s temperature.

For these factors, climate models use “parameterizations,” or simplified equations that behave roughly the same as the real thing. Rain, snow, and evaporation are other physical processes that have to be “parameterized” in climate models. These are important features of the Earth’s climate, so getting the parameterizations right is a huge part of designing a good climate model.

 

Global vs. regional models

Climate models can be global or regional. Global models cover the whole Earth. They usually have “resolutions” of hundreds of kilometers, meaning they can only show climate trends on a very large scale: for instance, they can model temperature changes in New England, but not in Rhode Island.

Regional climate models, which zoom in on specific areas, have much finer resolutions, usually a few tens of kilometers. This is much closer to the scale of real-world observations about topography, land cover and soil types, all of which affect the climate system. For this reason, regional climate models can use more real-life data than global models, and their simulations are generally more accurate. They are useful for studying natural variations in the Earth’s climate; studying how land use (like agriculture and deforestation) can affect regional weather patterns; and making more detailed predictions about how climate change will affect the places where people live.

In general, global climate models are useful for understanding the consequences of human actions across the whole world. For example, when the Intergovernmental Panel on Climate Change evaluates the actions needed to meet the worldwide climate targets set in the Paris Agreement, they use data from global climate models. Regional climate models are better suited to studying how climate change affects things important to us, like agriculture, diseases, and specific ecosystems, and for making plans to adapt to future climate change.

 

Resolution in Climate Models. Climate models represent large land areas as three-dimensional grids. Models with higher resolution have more “squares” in the grid, which makes them more accurate and precise. But there’s a tradeoff: because a climate model must repeatedly solve equations for every square in the grid, the resolution can only be so high before the model becomes unmanageably slow to run. Models that cover smaller regions of the globe can afford to have higher resolutions. This graphic shows how typical climate models “see” the world, compared to a real satellite image (top). On the left, a global climate model with a resolution of 2º of latitude and longitude. On the right, a regional climate model with a resolution of 50 kilometers.
A powerful tool

The Earth’s climate is too complex for even the most powerful computers to fully simulate. Just as modern weather models cannot tell us with certainty whether it will rain next week, climate models can only predict a likely range of outcomes.

Nonetheless, they are a crucial tool for understanding climate change, and are continually growing more detailed and accurate. New discoveries in climate science are improving our understanding of natural climate processes, and providing more real-world data about the Earth’s climate system, which allows for more accurate simulations of complex features like clouds and the water cycle. At the same time, advances in computer technology are making it possible to simulate weather patterns on a finer scale than ever before.

 

Published January 8, 2021.

Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International license (CC BY-NC-SA 4.0).
Photo Credit
Stuart Rankin via Flickr
Elfatih Eltahir
by Elfatih Eltahir, Professor of Civil and Environmental Engineering
Aaron Krol
with Aaron Krol, MIT Climate Portal Writing Team
Related MIT Groups
More Resources for Learning
MIT Climate Primer: "Predicting climate"
Eltahir Research Group (Interactive Maps, Videos, and Scientific Papers)
Carbon Brief: "How do climate models work?" (Q&A)
MIT Department of Earth, Atmospheric and Planetary Sciences: "Historical climate models accurately projected global warming" (News Story)
National Center for Atmospheric Research: The Very Simple Climate Model (Teaching Tool)
Intergovernmental Panel on Climate Change: "Evaluation of Climate Models" (Report)
MIT OpenCourseWare: "Introduction to computational thinking" (Online Class)
TILclimate Educator Guide: Climate Models and Uncertainty
TILclimate Educator Guide: The Changing Ocean and Climate Change
Topics
Climate Modeling

Want to learn more?

Listen to this episode of MIT's "Today I Learned: Climate" podcast on uncertainty and predicting climate change.

KE: [00:00:00] [00:00:00] If we have a temperature increase of about 6 degrees Centigrade it may be catastrophic. The probability is not large, but it's not zero, and it's not tiny either. And so you have to think about that.

LHF: [00:00:16] Welcome to TIL Climate, the show where you learn about climate change from real scientists. My name is Laur Hesse Fisher, and today we’re talking about … risk and uncertainty.

If you’re anything like me, then you want an easy answer about climate change. How will it affect my family, my business, my country? When will this happen? How much do I really need to change?

When we interviewed MIT Prof. Kerry Emanuel about hurricanes, we also spoke about where the uncertainty is in climate change, and he told us how he speaks with business leaders and politicians about risk.

To really get into this, there’s two things you should [00:01:00] know. The first is -- the Earth’s climate is complicated.

KE: [00:01:06] We're dealing with a very complex system. Many many interacting components: the transmission of radiation through the atmosphere, wind, the coupling with the ocean, ocean circulation, the land surface, the whole biosphere, which interacts with all of that…

I would say it rivals say the human body in complexity. We have all these different subsystems, you know, like we have hearts, and livers, and intestines.

And if those parts start to function differently or, God forbid fail, it's going to affect the whole organism in ways that even today in some ways medical sciences doesn't perfectly understand.

Just as the earth's climate has these subsystems, and how they interact with each other's complex.

LHF: [00:01:54] The second thing you should know is, that it’s hard to map that complexity.

KE: [00:01:59] When we [00:02:00] talk about a climate model, we’re really talking about an algorithm, which is a way of solving a very complex set of equations that govern the behavior of the system. And those equations are not just pulled out of thin air. They're actually the equations that we know govern the behavior of physics.

LHF: [00:02:20] If we wanted to create a perfect model how every part of the human body works, you’d have to know what’s happening on the nano level, like, the subatomic level, for everywhere in your body. It’s kind of the same for our climate system:

KE: [00:02:35] To make a perfect model of the system, you would have to be able to calculate things as small as a cubic centimeter or so or maybe less to do that. And we have nowhere near the computational firepower to do that.

LHF: [00:02:51] A good example of this is clouds. As we know from our previous episode with Prof. Dan Cziczo, the particles that help clouds [00:03:00] form behave very differently depending on where they are in the world. But, as powerful as computer climate models are, even clouds are, for the moment, just too small for the models to take into consideration.

KE: [00:03:12] Clouds are very very important. But the clouds may be 10 kilometers across you can't resolve that with today's climate models. They’re too small, and yet if the climate model didn't have some representation of them, it wouldn't work. And so we have to tell the climate model that they're there.

LHF: [00:03:33] You had written that there are roughly 40 climate models used by different organizations around the world, and they all give somewhat different predictions on climate change. Why do they differ from each other?

KE: [00:03:44] They all make different assumptions about what's happening on scales that are too small for them to actually compute. And it is a way of dealing with, it's not by any means a perfect way, but it is a way of dealing [00:04:00] with uncertainty. So you have different groups making different assumptions about how to do this, running different models, and comparing them.

LHF: [00:04:09] What’s neat is that this is an essential part of science. If we don’t know something for sure, then we want scientists to take different assumptions of what could be, and run them through their models, so we can see what the most common outcome would be. It’s like getting quotes from different contractors, or advice from multiple consultants: you hear what each of them say, compare them, and then use that to build a picture of what to do. This is like what the scientific community does, and they’re really transparent about it.

KE: [00:04:40] One of the most fascinating and interesting and useful parts of science is actually quantifying our own ignorance, quantifying the level to which we’re uncertain.

Let me take an everyday example: if I were to tell you as an atmospheric scientist that the temperature tomorrow, the high temperature, in Boston would be 50 [00:05:00] degrees, but it might be as warm as 53 or as cold as 47, most people understand that the you know you can't make a perfect weather forecast, that there's uncertainty in it. And that doesn't mean that we don't know right? It will be somewhere in that range.

LHF: [00:05:17] As a side note, Prof. Emanuel isn’t saying that climate change is like weather. Weather is like your mood, while climate is like your personality; you might generally have a sunny disposition, but you’re going to feel grouchy sometimes. In the same way, weather may change day to day, but it’s guided by something much larger and more constant, the climate.

OK back to Prof. Emanuel.

KE: [00:05:41] Good scientists are careful to quantify the uncertainty whenever they say anything about the future, whether it's a weather forecast or climate projection. It's absolutely essential to the final step that everybody really needs and wants, which is an assessment of the risks [00:06:00] associated with climate.

LHF: [00:06:02] Risk… if we aren’t sure if something really bad is going to happen, we think of it in terms of a risk. Like our house flooding or us getting an expensive medical bill. It’s why we buy insurance.

Because climate change also comes with a level of uncertainty, it’s helpful looking at it in terms of risk.

This next part is less about the science of climate change and more about how decision makers, and really all of us, can think about risk… What Prof. Emanuel says here might stick with you more than anything else in this podcast series so far.

KE: [00:06:38] When we make decisions about risk, we rarely make decisions based on the most probable outcome. Let me take a really simple example, you're walking your daughter to school, you got come to a busy intersection across which is the school bus, which has just pulled in, and you’re little bit late.

Now, you can let your little girl run for [00:07:00] the bus, and let's say in your own mind there's a 2% probability she'll be run over on the way.

LHF: [00:07:06] OK I know that’s a little dark, but we’ll continue with the example…

KE: [00:07:10] If she doesn't run you'll have to take her to school because she's going to miss the bus. Now the most probable outcome is that she could be fine, and yet that's the last thing you do. And all that illustrates is that to get the risk you have to take into account two things: the probability of the outcome and how expensive, not necessarily in monetary terms, the various outcomes are.

LHF: [00:07:37] So so how likely it's going to happen, and how bad it would be if it did happen?

KE: [00:07:40] Yeah, that's right. Both you have to take into account both.

Well, that's a metaphor for the climate system too. The most probable outcome the way we see it is if we double carbon dioxide will have a temperature increase of about 3 degrees C.

LHF: [00:07:56] An increase like this comes with some really dramatic impacts.

KE: [00:08:00] [00:08:00] You have to start moving structures that are right on the coast Inland or putting them up on pilings. You have to change your agricultural practices. You have to deal with huge immigration pressures because there are parts of the world which are already agriculturally marginal who will cease to be able to do any agriculture at all. So those people are going to want to move. So you have to deal with that. We're already dealing with it. And it's disruptive, but it's not so far catastrophe.

LHF: [00:08:30] The thing is, a catastrophe is inside the realm of possibility.

KE: [00:08:35] If it's five degrees Centigrade or 6 degrees Centigrade it may be catastrophic.

Catastrophic is going to kill you, or it's going to really harm civilization if we're talking about the whole world.

The probability of it being 6 degrees centigrade is not large, but it's not zero, and it's not tiny either. It's somewhere down there. Maybe it's low probability, but it's also a low probability that your daughter will be [00:09:00] run over if she runs for the school bus. You still have to think about it.

...

LHF: [00:09:09] Scientists have created a range of scenarios of what may happen with climate change.

Some people who look at the data think that our society should prepare for what scientists say is most likely to happen. And some people think that we should look at the best or worst that could happen, even if it’s unlikely.

When reading about climate change or listening to advocates or policymakers, you can try to understand which scenario they’re planning for here. Because what we do and how quickly we act, will differ a lot depending on which future we’re planning for.

So, what about you? What world do you think our society should prepare for? The unlikely one where climate change doesn’t really impact much at all? The likely disruptive future? Or the [00:10:00] unlikely catastrophe?

You can tweet us @tilclimate. And if you’re interested in how scientists talk about these different scenarios check out our show notes on tilclimate.mit.edu. Thanks for joining us today on TILclimate, and thanks to Prof. Emanuel for speaking with us. I’m your host Laur Hesse Fisher from the MIT Environmental Solutions Initiative, and I’ll see you next time.

Keep exploring

With more Explainers from our library:

View All