Why climate projections are bands, not lines

The spread in 2100 temperature forecasts isn’t a failure of models; it is a deliberate mapping of human choices layered over physical uncertainty.

When the Intergovernmental Panel on Climate Change publishes a temperature projection, the headline number is often a range. Readers frequently interpret a wide band as a lack of precision, as if the models are guessing. This is incorrect. The width of the band comes from two distinct variables: what humans decide to emit, and how the Earth’s physics reacts to that carbon. A single line would imply a single future is guaranteed. A band shows the boundaries of possibility.

Climate scientists do not run one simulation to predict the future. They run ensembles. The National Oceanic and Atmospheric Administration (NOAA) and NASA Goddard Institute for Space Studies (GISS) rely on these ensembles to stress-test physical laws against different policy decisions. Each simulation changes the inputs. One simulation assumes global carbon emissions peak in 2025. Another assumes emissions continue rising through 2050. A third assumes a rapid transition to renewables by 2035. The models then apply the same physics to each scenario. The result is not a single prediction, but a probability distribution. The uncertainty in the 2100 temperature projection is not about whether the greenhouse effect works; it is about how much carbon enters the atmosphere and how sensitive the climate system is to that carbon.

The first source of uncertainty is emissions scenarios, known as Shared Socioeconomic Pathways (SSPs). These are not predictions. They are plausible storylines used by the IPCC to test physical limits. The SSP1-1.9 scenario assumes immediate, aggressive decarbonization. The SSP5-8.5 scenario assumes fossil fuel use continues to grow unchecked. These pathways diverge significantly in total cumulative emissions by 2100. The second source of uncertainty is climate sensitivity. This is defined as the amount of warming expected from a doubling of atmospheric CO2. The IPCC Sixth Assessment Report (AR6) places the likely range of equilibrium climate sensitivity between 2.5°C and 4.0°C. This is a physical property of the Earth system. It depends on feedback loops, such as melting ice reducing reflectivity or thawing permafrost releasing methane. The models agree on the direction of warming. They disagree on the exact magnitude because the feedbacks are not fully quantified.

The table below breaks down the projected warming for 2100 based on these two variables. The rows represent the emissions pathway. The columns represent the range of climate sensitivity. The values are global mean surface temperature increase above pre-industrial levels (1850-1900).

The visualization reveals that the band widens as emissions increase. In the Low Emissions column, the difference between low and high sensitivity is small. The physics is constrained because the forcing is weak. In the High Emissions column, the band stretches from 3.0°C to 5.7°C. This is not model error. This is the compounding effect of feedback loops under stress. When the atmosphere holds more heat, ice melts faster, which absorbs more heat, which melts more ice. This cycle amplifies the initial carbon input. The models capture this amplification, but the exact rate of feedback is the source of the vertical spread in the table. The uncertainty is not about whether the planet will warm; it is about how violently the feedback loops will respond to high carbon concentrations.

The tradeoff visible in the data is between control and consequence. The width of the band in the High Emissions scenario is larger than the entire width of the Low Emissions scenario. This means the cost of uncertainty is highest if emissions continue to grow. By choosing the SSP5-8.5 pathway, society accepts a range of outcomes where the worst case is 5.7°C. By choosing the SSP1-1.9 pathway, society limits the worst case to 1.8°C. The difference between a 1.8°C world and a 5.7°C world is not a matter of statistical noise. It is the difference between a manageable shift in agriculture zones and a fundamental breakdown of coastal infrastructure. The IPCC AR6 report calculates that limiting warming to 1.5°C requires global CO2 emissions to fall by 45% by 2030 relative to 2010 levels. The uncertainty in the table is a function of whether that reduction happens.

The difference between the bottom of the band and the top of the band in the High Emissions scenario is 2.7°C. That specific number represents the potential impact of unquantified feedbacks under high stress. The lower bound of the Low Emissions scenario is 1.4°C. The gap between the best-case physics and the worst-case policy is roughly 4.3°C. This gap is not something models can resolve. It is a decision space. Every 0.5°C of warming avoided by policy changes the probability of crossing irreversible thresholds, such as the collapse of the West Antarctic Ice Sheet.

The closer

The 2100 projection is not a forecast of what will happen. It is a map of what can happen. The IPCC’s 1.4°C-to-5.7°C range exists because two distinct things are unknown: how much carbon humans will emit (policy) and how strongly the Earth’s feedback loops will respond to that carbon (physics). The 2.7°C span in the high-emissions column is the physics uncertainty alone. The 4.3°C span across the whole table is the cost of treating policy as fixed. The width of the band is not noise. It is the size of the decision still in front of governments — and the size of the world that gets handed to the people living in it in 2100.