Imagine a world where the Earth's lungs are suddenly gasping for air—our planet's land-based carbon sinks, vital guardians against climate chaos, have been slashed in half in just one year. That's the startling revelation from cutting-edge AI research, and it's a wake-up call we can't ignore. But here's where it gets controversial: Could this AI-driven prediction be overstating the crisis, or is it just the harsh truth we need to face? Let's dive in and unpack it all step by step.
A dedicated team of researchers from Peking University, spearheaded by Wang Heyuan and Wang Kai at the Institute for Carbon Neutrality (ICN), harnessed advanced artificial intelligence models to uncover a dramatic downturn in the global land carbon sink during 2024. Their groundbreaking study, titled "AI-tracked halving of global land carbon sink in 2024," was featured in the journal Science Bulletin, accessible at this link: https://linkinghub.elsevier.com/retrieve/pii/S2095927325010370.
To make this easier to grasp for newcomers, let's clarify some basics: The global carbon cycle is like Earth's breathing system, where carbon dioxide (CO₂) moves between the atmosphere, oceans, and land. Terrestrial ecosystems—think forests, grasslands, and even deserts—are the land-based heroes in this cycle. They act as carbon sinks, naturally absorbing CO₂ from the air through processes like photosynthesis, where plants take in carbon and convert it into biomass. In fact, these ecosystems soak up about one-third of the carbon emissions we humans produce every year through activities like burning fossil fuels. With climate extremes, such as scorching heatwaves and relentless droughts, becoming more frequent, keeping tabs on these changes in real time isn't just useful—it's crucial for steering us toward a sustainable future.
Traditional methods for monitoring these ecosystems often suffer from a frustrating delay, like waiting a full year or more for data to catch up. But the AI models employed by this team, known as process-based approaches, bridge that gap. They allow for near-instantaneous detection and analysis of how the carbon cycle reacts to climate shifts, empowering scientists to understand and respond to emerging changes in Earth's interconnected carbon-climate system. This innovation paves the way for smarter, evidence-backed policies that can adapt to unfolding environmental challenges. For instance, imagine policymakers using real-time data to tweak reforestation efforts before a drought wipes out entire regions— that's the kind of proactive power we're talking about.
Key Findings from the Study
At the heart of their work is an in-house AI model called Carbon Mind, which is designed to learn the intricate, rule-based mechanisms of the terrestrial carbon cycle. Unlike some black-box AI systems that are hard to interpret, Carbon Mind stays transparent and traceable, enabling quick updates as unusual climate events occur. Using this tool, the team calculated the 2024 global land carbon sink and created detailed maps showing its distribution across the planet at a fine resolution.
The results paint a stark picture: The land sink plummeted to less than half of its average over the previous decade, with the most severe drops concentrated in the tropics—the warm, biodiverse regions near the equator. For beginners, think of the tropics as Earth's tropical belt, home to lush rainforests and expansive savannas, where life thrives in abundance but is now under siege. Within these areas, grasslands and savannas—open landscapes dotted with grasses, shrubs, and scattered trees—suffered disproportionately more than tropical rainforests. This challenges the common assumption that arid or semi-arid ecosystems, adapted to dryness, might bounce back stronger from extreme weather. Instead, prolonged droughts and heatwaves have ravaged their vegetation, reducing productivity and, consequently, their ability to absorb CO₂.
Delving deeper, the research points to heat and drought as the primary culprits behind this tropical carbon sink decline. When plants can't photosynthesize efficiently due to wilting leaves or stifled growth, their carbon-storing capacity nosedives. Consider a real-world example: In places like sub-Saharan Africa or parts of Australia, where grasslands are a lifeline for grazing animals and local communities, repeated droughts have led to soil erosion and diminished plant cover, making these ecosystems even more susceptible to future shocks. And this is the part most people miss: These findings hint that tropical land systems, particularly the semi-arid grasslands and savannas, might be far more fragile than we thought, potentially accelerating global CO₂ buildup in the atmosphere and exacerbating climate change.
By weaving AI insights with atmospheric data and on-the-ground observations, such tools can revolutionize how we manage land. They enable "stress-testing" of climate models—essentially simulating worst-case scenarios to see how different strategies hold up—and support nimble policy responses, like targeted conservation programs or incentives for sustainable farming that bolster resilience.
For further details, check out the full study by Heyuan Wang et al., "AI-tracked halving of global land carbon sink in 2024," published in Science Bulletin (2025), with DOI: 10.1016/j.scib.2025.10.015, available at: https://dx.doi.org/10.1016/j.scib.2025.10.015.
Citation: Global land carbon sink halved in 2024, AI model suggests (2025, November 4), retrieved 4 November 2025 from https://phys.org/news/2025-11-global-carbon-halved-ai.html.
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.
What do you think? Does this AI model's alarm on tropical vulnerabilities align with what you've observed in climate discussions, or do you suspect it's amplifying fears for attention? Should governments prioritize protecting grasslands over rainforests in light of these findings? Share your opinions, agreements, or counterpoints in the comments—we'd love to hear diverse perspectives and spark a meaningful conversation!
