Here’s why it is a good thing climate models don’t agree
Every climate model produces errors. We can learn a lot about our climate by studying the reasons for these errors.
Earth is a laboratory, and humanity is conducting an enormous experiment to find out how our climate changes when we burn fossil fuels. The problem is that we are part of the experiment, and if it goes wrong we bear the consequences. From a scientific point of view, it’s a very bad way of conducting experiments: It’s not controlled, there is no alternative, and we can’t start from scratch and do it differently.
The only way to perform this experiment in a scientific way — that is in a controlled environment, repeatable and testable — is with simulations. A bit like virtual reality, but without the goggles (except in rare cases). Like all simulations, climate simulations involve computers and code, which in this particular case is called a “climate model”. As with Lego models, climate models aren’t perfect, but they are good enough for their purpose, which for climate models is this: Give the most accurate representation of Earth’s climate possible with today’s knowledge and technology (hence their name).
There are two main obstacles to building the perfect climate model: First, we don’t know everything ourselves, and we can’t program something we don’t know. Second, even if we knew everything, we wouldn’t be able to put all of our knowledge into the models. That’s because we are still limited by technology, which only allows so much computing power, memory, and storage.
A perfect model of our climate wouldn’t be interesting anyway, because it would be indistinguishable from the real climate. So why bother?
In order to build climate models which are good enough to serve science and fast enough to serve humanity, we let the models replace some of the hard science (and unkowns) with a good guess. And the important point here is that each model has its own strategy to do that. The consequence is that each model will give us a slightly different picture of the past, current, and possible future climates (note, however, that we are talking small differences here, and no climate model has ever predicted anything other than a warmer planet in the future).
Storms, rain and strong winds are among the most difficult things to simulate with models. Therefore, the tropical regions, where thunderstorms happen often and are strong, are places where models disagree the most. This is particularly true for the equatorial Pacific. Fortunately, this is almost entirely open ocean and none of Earth’s main population centers are located in this region.
But what recent research found is that much of the differences in midlatitudes, where most people live, can be traced back to differences in the equatorial Pacific. To say it another way, some of the regions on Earth which are the most difficult to simulate accurately have a direct impact on the climate in regions of high population (and vegetation).
Therefore, these model differences are an opportunity for scientists to study the importance of the equatorial regions for the climate around the globe: When comparing models which are tending into one direction (say too little rain in the tropics) with those tending into the other direction (say, too much rain in the tropics), we can get an idea of what periods of strong or weak rainfall in the tropics mean for the climate in, say, the US West Coast, Asia or Australia.
And so, we can learn even more about our climate by taking advantage of the quirks and limitations of our climate models. Thank goodness they aren’t perfect!
This article refers to a recently published paper available from the Journal of Climate at https://doi.org/10.1175/JCLI-D-20-0195.1
