Local temperatures rapidly depart from historical experience beyond 1.5°C

For decades, scientists have tracked the Earth's rising average temperature, watching the global thermometer climb past the pre-industrial baseline. But for people living in specific towns, cities, and forests, the story is not just about a global average; it is about the local weather they experience every day. A crucial question has long lingered in climate science: at what point does the heat become permanent? There is a difference between a single record-breaking summer and a future where the local climate never again dips back to the range of temperatures that human societies and natural ecosystems evolved to expect. Scientists call this a "warm state," a condition where the annual average temperature stays above the highest levels seen in the historical record, never to return to the familiar cool of the past. Understanding when this shift happens is vital because our infrastructure, our crops, and our ecosystems are all built on the assumption that the climate will eventually return to its historical norms. If that return never comes, the reference points for designing our world vanish.

A new study by researchers at the University of Oxford tackles this uncertainty by asking a different question. Instead of trying to predict the exact year a specific city will break its heat record, they asked: given a certain level of global warming, how likely is it that a local temperature will stay permanently above its historical maximum for the rest of the century? To answer this, the team turned to a vast library of computer simulations. They trained a sophisticated computer program, a type of artificial intelligence known as a neural network, on 408 different climate simulations. These simulations represent various possible futures, but the researchers focused on one specific path that closely matches current global policy commitments, a scenario where greenhouse gas emissions continue to rise but eventually stabilize. The program learned to recognize patterns in the climate history of each location, such as how much the temperature fluctuates naturally and how fast the warming trend is accelerating. It then used this knowledge to calculate the probability that a place would enter a permanent "warm state" once the world reached specific warming milestones: 1.5 degrees Celsius, 2 degrees, 2.5 degrees, and 3 degrees above pre-industrial levels.

The results reveal a stark and accelerating reality. At a global warming level of 1.5 degrees Celsius, the situation remains relatively contained. In these simulations, only a tiny fraction of the world's land area, just 2.2 percent, is very likely to have crossed into this permanent warm state. However, the situation changes dramatically with just a half-degree of additional warming. When the planet reaches 2 degrees Celsius, the area of land locked into this new, unprecedented climate expands sharply to cover 40.3 percent of the global land surface. This is not just a geographic shift; it is a human one. By the time the world warms by 2 degrees, nearly half of the people currently living on Earth will be in regions where the annual temperature never drops back below the historical maximum. As warming continues to 2.5 degrees and 3 degrees, the numbers climb even higher, eventually exposing 96 percent of the global population and nearly 80 percent of the world's aboveground plant life to these sustained, record-breaking conditions.

The study also clarifies why previous estimates might have been misleading. Earlier research often treated climate models as a single, uniform voice, averaging them together to find a precise date for when the climate would change. The Oxford team found that this approach misses a critical detail: the role of natural, random weather variations. In their simulations, two identical climate models running the same scenario could produce vastly different outcomes for a specific city like Oslo. In one run, the city might enter a permanent warm state in 2058; in another, it might not happen until 2091. This difference of decades is caused by the natural ups and downs of the weather, which can either delay or hasten the moment the heat becomes permanent. By training their neural network to account for these local fluctuations and the specific history of each location, the researchers created a tool that is far more reliable than simple averages. They validated their method by testing it against climate models it had never seen before, proving that the program could accurately predict outcomes based on the unique climate history of a place, rather than just guessing based on a global average.

The implications of these findings are profound for how we view the future. The research suggests that the transition from a climate that occasionally breaks records to one that permanently exceeds them is not a slow, linear slide, but a rapid expansion that accelerates sharply between 1.5 and 2 degrees of warming. The tropics are the first to experience this shift, as their weather naturally fluctuates less than in other regions, making the signal of warming clearer. However, the study emphasizes that this is not a distant problem for the future; it is a trajectory that current policies are already setting the world on. The simulations show that under the emissions path closest to today's political commitments, a rapidly growing share of humanity and the Earth's vegetation will be committed to a future where the local climate has no historical analogue. The study does not claim to predict the exact year this will happen for every city, but it provides a highly confident map of the risk, showing that the window to avoid this permanent departure from the past is closing quickly, and the consequences of crossing that threshold are far more widespread than previously understood.