Data-driven predictions of the time remaining until critical global warming thresholds are reached

Area of Science:

• Climate Science
• Artificial Intelligence
• Machine Learning

Background:

• Global warming thresholds (1.5°C and 2°C) pose significant risks to natural and human systems.
• Accurate prediction of these thresholds is crucial for effective climate change mitigation and adaptation strategies.
• Existing climate models have uncertainties in projecting future warming timelines.

Purpose of the Study:

• To develop a data-driven approach using artificial neural networks (ANNs) to predict the timing of critical global warming thresholds.
• To quantify the uncertainty in climate model projections by analyzing historical temperature observations.
• To identify key geographic regions influencing the prediction of warming timelines.

Main Methods:

• Training ANNs on climate model output to recognize spatial patterns of historical annual temperature.
• Using the trained ANNs to predict the time until 1.5°C and 2°C global warming thresholds are reached, without using observational data during training or validation.
• Employing explainable AI methods to understand the geographic focus of the ANNs.

Main Results:

• ANNs accurately predict historical global warming timing from temperature maps.
• The central estimate for the 1.5°C threshold is 2033-2035 (±1σ: 2028-2039) under the Intermediate (SSP2-4.5) scenario.
• A substantial probability of exceeding 2°C exists even in the Low (SSP1-2.6) scenario, potentially sooner than some previous assessments suggest.

Conclusions:

• The data-driven framework offers a novel method for analyzing climate change signals in observations and reducing projection uncertainties.
• Results indicate a heightened likelihood of reaching 2°C warming in the near future, underscoring the urgency of climate action.
• The study provides further evidence for high-impact climate change within the next three decades, necessitating immediate attention to mitigation and adaptation efforts.