Long-term probabilistic temperature projections for all locations

Area of Science:

• Climate Science
• Environmental Science
• Statistical Modeling

Background:

• Current climate change projections lack full probabilistic interpretation.
• Existing global temperature forecasts do not provide regional specificity.

Purpose of the Study:

• To develop a method for probabilistic long-term spatial forecasts of local average annual temperature change.
• To provide a probability distribution for future temperatures at any location and year.

Main Methods:

• Combining a probabilistic global method with a pattern scaling approach.
• Utilizing out-of-sample predictive validation for calibration.

Main Results:

• High latitudes, continents, and the Northern Hemisphere warm more than low latitudes, oceans, and the Southern Hemisphere, respectively.
• A 5% chance of 16°C warming in the Arctic; 95% chance of at least 2°C warming in North Africa, West Asia, and most of Europe.
• Uncertainty sources shift from natural variability to model and emissions uncertainty by 2100.

Conclusions:

• The developed method provides well-calibrated, probabilistic spatial forecasts for local temperature change.
• Findings align with established patterns of differential warming across latitudes and surfaces.
• The study quantifies regional temperature change probabilities and clarifies uncertainty drivers over time.