Fifty years after the start of the project to model the Earth’s future climate, we still don’t really know what awaits us. Some places are warming up with more ferocity than expected. Extreme events surprise scientists. Now, as the simple reality of climate change takes its toll on human life, scientists are seeing more clearly the limits of our ability to predict the exact future we face. The coming decades could be much worse and stranger than the best models predicted.
It’s a problem. The world has warmed enough that urban planners, public health officials, insurance companies, farmers and everyone else in the global economy want to know what’s next for their part of the planet. And telling them would require geographic precision that even the most advanced climate models don’t yet possess, as well as computing power that doesn’t yet exist. Our vision of what is happening and what will likely happen on Earth is less blurred than it has ever been. Yet the extremely local scale at which climate change is experienced and the global reach of our best tools for predicting its effects simply don’t match.
Current climate models very precisely describe the broad outlines of the Earth’s future. But warming has now progressed enough that scientists are noticing troubling discrepancies between some of their predictions and actual results. Kai Kornhuber, a climatologist at Columbia University, and his colleagues recently find that, on every continent except Antarctica, certain regions have emerged as mysterious hotspots, experiencing repeated heat waves worse than any model could predict or explain. In places where a third of humanity lives, actual daily temperature readings exceed model predictions, according to an upcoming report. research by Alexander Gottlieb and Justin Mankin of Dartmouth. And the global rise in temperatures that lasted from mid-2023 until last June remains largely unexplained, a fact that troubles Gavin Schmidt, director of NASA’s Goddard Institute for Space Studies, although this doesn’t entirely surprise him.
“Since the 1970s, people have realized that all models are wrong,” he told me. “But we’ve worked to make them more useful.” In this sense, the climate modeling project is a scientific process that is proceeding normally, if not excellently. Only now the whole world needs very specific information to make crucial decisions, and it needed it, just like yesterday. The fact that scientists don’t have these answers might sound like a failure of modeling, but in reality it’s a testament to how widespread and rapid climate change is.
Earth is a place of unfathomable complexity, a nested doll of systems within systems. The feedback loops between temperature, land, air and water are made even more complicated by the fact that every place on Earth is a little different. Natural variability and human-induced warming are further changing the rules that govern each of these fundamental interactions.
Some of these systems, such as cloud formation, are notoriously misunderstoodalthough it has a major influence on climate change. And, like clouds, many parts of the Earth system are simply too localized for climate models to detect. “We need to get closer to cloud formation because we don’t have the small scales to resolve individual water droplets coming together,” Robert Rohde, Berkeley Earth’s chief scientist, told me. an open source environmental data nonprofit. Likewise, the models approximate topography because the scale at which mountain ranges undulate is smaller than the resolution of global climate models, which tend to represent Earth with, at best, 100-kilometer pixels squares. This resolution is useful for understanding phenomena such as Arctic warming over decades. But “you can’t solve a tornado that’s worth anything,” Rohde said.
Models simply cannot work at the scale at which people live, because assessing the impact of current emissions on the future world requires hundreds of years of simulations. Modeling the Earth with one-square-kilometer pixels would require “about a hundred thousand times more calculations than we currently have,” NASA’s Schmidt told me. Nevertheless, global climate models can be useful at the local level if combined with enough regional data and the appropriate expertise, and more and more people now want to use them in this way, to understand the risks to their properties and their investments, or to develop emergency plans and construct buildings. infrastructure. “We ask a lot of models. More than in the past,” Rohde said.
For non-scientists, extracting useful information from climate models requires professional help. Climatologists have worked for years with New York City to guide choices, including where to locate infrastructure, taking into account sea level rise. But, Schmidt said, “there is no simply not enough scientists to be on the advisory board of every locality, every company, every institution, or every corporation,” helping them access the right climate data or choose which models to build on. (Some are better than others at simulating certain variables, such as day-to-night temperature variations.) Often, governments end up turning to private-sector companies that claim to be able to translate the data ; Schmidt would prefer to see his own field produce work more directly useful to the public.
At the same time, as models come up against the reality of dramatic climate change, some of their limitations become apparent. When this scientific endeavor began, the models were supposed to imagine what global temperatures might look like if greenhouse gas emissions increased, and they have done a remarkable job of doing so. But models are, even today, less capable of taking into account the side effects of these emissions that no one predicted would come, and which now seem to lead to significant changes.
Some of these variables are completely absent from climate models. Trees and land are important sinks of carbon emissions, and the fact that this may change is not taken into account in climate models. But it changes: the trees and the earth absorbed much less carbon than normal in 2023, according to research published last October. In Finland, forests have stopped absorbing the majority of carbon they once did and have recently become a net source of emissions, which, as The Guardian reportedhas wiped out all the progress the country has made in reducing emissions from every other sector since the early 1990s. The interactions of ice sheets with the oceans are also largely absent from the models, Schmidt told me, despite the fact that melting ice could change ocean temperatures, which could have significant impacts. Changing ocean temperature patterns are currently making climate modelers at NOAA rethink their El Niño and La Niña models; the agency initially predicted that La Niña’s cooling powers would manifest much sooner than it appears today.
Biases in climate models go both ways: some overestimate the risk due to various factors, and others underestimate it. Some models “run hot,” suggesting more warming than is actually occurring. But recent findings on extreme temperatures point in the other direction: models could underestimate future climate risks in several regions due to a still unclear boundary. And, according to Rohde, underestimating risk is much more dangerous than overestimating it.
Kornhuber also believes that the fact that models already appear to be significantly underestimating climate risk in several places is a bad sign for what lies ahead and for our ability to predict it. “It should be concerning that we are now entering a world where we have sort of reached the limit of our physical understanding of the Earth system,” Kornhuber said.
While models struggle to capture the world we currently live in, the planet is becoming more and more foreign to us, further from our reference ranges, as the climate continues to change. Given unlimited time, science could probably develop models that would better account for what we observe. But then it would be too late to do anything. Science has been engaged in modeling for more than five decades, and our best tools can only take us so far. “At the end of the day, we all make estimates of what’s coming,” Rohde said. “And there is no magic crystal ball to tell us the absolute truth.” Rather, we are left with a partial image, gestural in scope, pointing to a world we have never seen before.