Earth’s climate has been through a lot over the past few billion years, and at this point scientists have a pretty good handle on how our planet has changed over time.

But we talk about the future of the climate too. About how it’s changing more quickly than it has in the past, and about the dangers of rising temperatures and more severe droughts and storms.

Our methods for predicting the future of climate are a little different than those for reconstructing the past. The past leaves physical records we can study in things like tree rings and ice cores. The future requires us to be a little more abstract.

So, we mostly predict the future of Earth’s climate using computer models— mathematical reconstructions of our atmosphere that account for as much detail as possible to produce an accurate simulation. And it’s taken decades to build them up to the point where we can be confident that what they say is accurate. But it is.

And what the models say is that our climate is changing, and humans are making it happen. Global climate models, or GCMs, grew out of early computerized attempts to model the planet’s atmosphere in the 40s and 50s. Scientists weren’t even trying to predict the future— they just wanted to create a representation of the Earth’s atmospheric system as it was at the time.

That was hard enough. Climate models, then and now, divide the Earth’s surface into chunks a few hundred kilometers on a side, each with different properties like air movement and surface temperature. Then, the computer calculates how all those chunks interact with each other to see how things change over time.

Those early models were super simple. They did things like combine land and sea into a single damp surface with no geography, or represent the Earth as a cylinder instead of a sphere because the poles confused the computer. Thankfully, that didn’t lead to a whole movement of cylinder-Earthers, though.

All this simplifying was necessary because early supercomputers were kind of limited. We’re talking, like, five kilobytes of RAM. Your phone has, like, 500,000 times that much.

Still, it was enough to come up with a crude picture of Earth’s atmospheric currents and its wet and dry regions. Eventually, though, scientists wanted more than just a model of the atmosphere around a featureless cylinder— they wanted something that fit ... an actual planet. Now, they could make the model more accurate by making the chunks smaller, but that would take more computing power.

And what about the effects of mountains on air movement? What about warm and cold water circulating in the oceans? To answer those questions, climate modelers needed more juice.

By the 1970s, they were also beginning to worry about the greenhouse effect of the carbon dioxide we were adding to the atmosphere by using fossil fuels. They thought it could trap the sun’s heat and cause the planet to warm. At the same time, computers were getting powerful enough that they weren’t limited to modeling the present anymore.

So, programmers wanted to start using models to predict future changes in the climate. But before a climate model can be turned loose to predict the future, it has to be able to predict the past—what’s known as hindcasting. And that’s an important test, because if the model’s “predictions” match what we know already happened, we can be more confident about what it says will happen in the future.

Temperature records go back a century or more. If you start a model in 1850, it should be able to progress through time and match the general trends in temperature we already know happened. And by the late 1970s, climate models started to be able to do this.

In 1979, a report pulling from two different climate models suggested that the Earth’s temperature would increase as atmospheric carbon dioxide increased. Specifically, assuming twice as much CO2 in the atmosphere, they predicted an increase between 1.5 and 4.5 degrees Celsius, which trends suggested could happen before the 21st century was up. That range has proven pretty reliable ever since.

It still fits the warming we expect to happen over the next century or so, even as models have become much more powerful. So, models in the 70s were able to answer one simple “what if” question: What happens if the amount of carbon dioxide in the atmosphere doubles? But these days, you can carry more computing power in your pocket than those early programmers ever dreamed of and use it to send people pictures of your cat that disappear after 10 seconds.

Truly, it’s a golden age. And not only are cat pictures better, but so are climate models. We now have the resources to ask and answer more detailed questions.

And boy, do we have a lot of them. Governments want to know what will happen, in detail, in their corner of the world. And scientists want to know what will happen if we take measures to stop global climate change, as opposed to letting emissions run rampant.

With the models we have now, we can produce more customized predictions based on different scenarios of what humans might do in the future, like how much carbon dioxide will be emitted over a given amount of time and how much land will be used for agriculture. You can also use these models to basically run the predictions backwards, and calculate the factors that would lead to a certain amount of warming. So for example, if policymakers want to see what changes should be made to limit overall global warming to 2 degrees Celsius, they just have to use a climate model with the appropriate parameters.

That said, just because it’s the 21st century and computers are awesome now doesn’t mean climate models have reached their perfect final form. The Earth’s atmosphere is an unbelievably complex system, and we don’t yet have models powerful enough to track absolutely everything. We also can’t account for every single thing humans do now or might do in the future.

So, we now know a lot about what will happen, but there’s still work left to do in perfecting our predictions. For example, up until 2012 or so, models weren’t great at predicting sea level rise— they undershot it pretty badly. UN reports in 2001 and 2007 made predictions for sea level rise that failed to track with what we observed via satellites by about 60%.

That might be because the models didn’t fully account for the rapid changes in the ice sheets over Antarctica and Greenland. For the UN’s most recent report in 2013, the models were re-programmed and the projections for sea level rise increased to be more in line with what we’re seeing happen. We want our models to do better, in part because we want to know what’s going to happen to our climate, and in part because accounting for all the variables is just good science.

With these constant improvements, every new prediction can be more detailed. And in the last couple of decades, we’ve started to be able to tailor predictions for specific regions. Which is helpful, because people and governments in different parts of the world want to know very different things.

A farmer in southern Europe might want to know if drought will affect their crops more severely in the future, while an islander from the Maldives might be more concerned about their country being completely swallowed by rising seas. Now we can try to answer those questions, and a lot of the answers aren’t encouraging. No matter how you look at it, the models are very clear that the planet is warming.

That’s one of the easiest things to predict— I mean, we’ve been doing it since 1979. If we pump more heat-trapping greenhouse gases like carbon dioxide into the atmosphere, the numbers come out warmer. There are factors other than greenhouse gases that the models take into account too, like particulate matter from volcanoes and coal-burning power plants, which can have a cooling effect.

And we know the oceans will slow warming for a while by absorbing CO2, until they run out of capacity. But all the predictions say the warming caused by greenhouse gases will have a stronger effect. Even in scenarios where governments around the world work quickly to limit and phase out greenhouse gas emissions, climate models predict that the Earth’s temperature increase will be at least around 2 degrees.

In other scenarios, where we take less aggressive action or no action at all, the planet’s average surface temperature goes up by more like 6 degrees Celsius. Which, if you’re used to Fahrenheit? That’s 11 degrees.

That’s hot. But higher temperatures are far from the only consequence of climate change. Using models, along with other tools, researchers predict all kinds of other effects.

A comprehensive report by the US Global Change Research Program, published in 2014, lists some of the consequences. It emphasizes effects on North America, but similar things would happen all over the world. With more and more time passing between periods of freezing temperatures, plants could experience a longer growing season.

That could actually have a net positive effect of causing plants to take up more CO2 so less stays in the atmosphere. But the good news pretty much ends there. In general, around the world and in North America, there would be changes in the distribution of precipitation, including rain and snow.

The changes tend toward increasing extremes. A warmer atmosphere holds more moisture, so wet regions will get wetter. But models show shifts in large-scale movements of air that would exacerbate dryness, too.

So, for example, the dry American Southwest will get even drier. And major events like droughts, heat waves, and powerful storms will get worse and more frequent. The effects of climate change on Atlantic hurricanes are hard to predict.

We don’t know for sure whether it was responsible for the devastating 2017 hurricane season. But we know that as the global temperature increases, the temperature of the surface of the ocean does too. And warm ocean water is hurricane fuel.

Historically, warmer years don’t produce more hurricane landfalls, so we may only see stronger storms, not more of them. But that’s not a sure thing. Climate change will have other kinds of effects, too, like rising sea levels, melting ice caps, and changes in ocean currents, but it would take a whole other episode to talk about them all.

Which is exactly what we did a few years ago. Thanks to decades of work improving climate models, we now have a very good idea of what’s happening, both globally and regionally. And even though our models have gotten much more detailed over the years, those predictions of increasing temperatures have stayed constant.

Which is a good sign for the science of climate modeling, but maybe also cause to be a little worried. We’ve always known the Earth wasn’t just a damp cylinder. But as our ability to model the climate has improved over time, we’ve learned a lot about how our planet works, and what we’re doing to it.

Thanks for watching this episode of SciShow. If you’re interested in learning more about the effects of climate change, we have another video about just that topic.