Model Intercomparison Projects (MIPs)

Read...

In the last section, we looked at results from three different climate models, one from NASA, one from the Department of Energy, and one from GFDL. And those are just three of the climate models developed in the United States! Globally, there are around thirty research groups that have their own fully developed climate models, depending on how you define "distinct" models (some can be better thought of as "families" where similar computer code gets shared between researchers).

So how do scientists compare all these models and share information to make our future predictions better? That’s where Model Intercomparison Projects (MIPs) come in—a collaborative approach to tackling these challenges.

Think of MIPs as a giant group science experiment where everyone follows the same procedure but uses slightly different tools and techniques. It’s a bit like those middle school physics contests where you build a contraption to protect an egg dropped off the roof of a building. Everyone follows the same rules—such as the height the egg is dropped from, the size limits of the contraption, or how many test drops you get—but you're free to develop your own creative strategy for keeping the egg intact. By comparing and contrasting designs, you can learn what works, what doesn’t, and how to improve your approach. In the same way, MIPs help scientists refine their models by identifying strengths and weaknesses across different methods.

An "egg drop challenge" asks participants to follow the same sets of rules, but allows them flexibility to decide how best to achieve their goal of not letting the egg break upon falling. Model Intercomparison Projects are not too dissimilar! Same rules; let's see how everyone does!

Credit. BevCanTech. "Egg Drop." Autodesk Instructables. 

These projects bring together multiple climate models and have them run under identical conditions. For example, they might tell every model exactly what the clouds should look like or specify how the ocean circulation behaves. The goal is to compare the range of outcomes these models produce and figure out where and why they might differ. By analyzing the similarities and differences, scientists can identify common errors and work collaboratively to refine their models, ultimately making climate predictions more reliable.

For example, many climate models struggle with what’s known as the "drizzle problem." This common error is called the "drizzle problem" because models tend to produce too much light precipitation, turning every day into a gray, misty one. Through MIPs, scientists realized that this issue was widespread across models from different research groups around the world. By working together, they developed strategies to address this error, improving a wide range of models. But the benefits of MIPs go beyond solving specific problems. These projects foster international collaboration and data sharing, pooling expertise and resources from scientists across countries and institutions. This collaborative spirit ensures that the global scientific community builds more comprehensive and robust climate models, essential for tackling global challenges like climate change.

MIPs also play a vital role in training the next generation of climate scientists. Participating in these large-scale projects provides early career researchers with hands-on experience using state-of-the-art climate models, while also allowing them to work alongside experienced scientists. This mentorship and collaboration are invaluable for building expertise and fostering innovation in climate science.

Sixth-generation MIP climate models (CMIP6) predict a much larger range of changes to the Atlantic Meridional Overturning Circulation (AMOC) in response to increased carbon dioxide concentrations compared to the previous generation of models (CMIP5).

Credit: © Wikimedia (Creative Commons Attribution 4.0 International) 

One of the most well-known MIPs is the Coupled Model Intercomparison Project (CMIP). Picture scientists from all over the globe, each running their climate models using the same starting conditions, like greenhouse gas concentrations, solar radiation, or volcanic activity. This standardized approach allows for a direct comparison of the outputs. I've shown one example here. Earlier in this class, we discussed the Atlantic Meridional Overturning Circulation (AMOC) and even watched a dramatic (and exaggerated!) clip from The Day After Tomorrow. The graph here shows what CMIP models, both the 5th generation (CMIP5) and 6th generation (CMIP6), predict for an AMOC "slowdown." Each line represents a different model. Despite all the models starting with the same setup, their outcomes vary, and that's actually a good thing... it ensures we're not relying on a single perspective! But when models from multiple centers consistently show a particular signal, like a future slowdown of AMOC, our confidence in that outcome grows. None of these models predict a complete shutdown of AMOC like in the movie, but the consensus points to a significant slowdown.

CMIP results are incredibly influential as a cornerstone of the Intergovernmental Panel on Climate Change (IPCC) assessment reports, which inform global climate policies and strategies. The collaborative troubleshooting process within CMIP is critical for refining models. For instance, if several models consistently overestimate the warming effect of greenhouse gases, researchers can dig into their assumptions and physics to identify the problem. These insights help make models more accurate and reliable, while also providing a range of possible future climate scenarios essential for risk assessment and planning. Speaking of scenarios, that’s exactly what we’ll dive into next!