How Much Carbon Dioxide Will There Be in the Atmosphere at the End of This Century?
By: Dr. David R. Legates, Guest Contributor
No one knows for sure. Yet climate models depend on assumptions about future CO₂ concentrations and the pathways that get us there. Their temperature projections hinge on these inputs. For years, modelers relied on a set of standardized scenarios. In early May, the Seventh Coupled Model Intercomparison Project (CMIP7) introduced a significant shift in how future climate projections are framed. This change will shape the next Intergovernmental Panel on Climate Change (IPCC) report and marks a welcome move toward greater realism.
Representative Concentration Pathways (RCPs), developed for the IPCC’s Fifth Assessment Report around 2014, provided standardized inputs focused on atmospheric concentrations and radiative forcing—the imbalance in Earth’s energy budget, measured in watts per square meter (W/m²) relative to pre-industrial levels by 2100. The main RCPs included:
- RCP2.6: very low emissions with strong mitigation, peaking and then declining.
- RCP4.5: intermediate reductions.
- RCP6.0: more modest reductions.
- RCP8.5: virtually no reductions, reaching about 8.5 W/m².
These were designed primarily for exploring physical climate responses across distinct forcing levels, not as precise socioeconomic forecasts. They were paired with emissions scenarios from Integrated Assessment Models (IAMs). RCP8.5 was intended as an extreme high-end case for exploratory purposes, but communication gaps led many researchers, media outlets, and policymakers to treat it as a central “business-as-usual” projection.
For CMIP6 and the IPCC’s Sixth Assessment Report, Shared Socioeconomic Pathways (SSPs) expanded the framework. SSPs combined socioeconomic narratives—covering population, economic growth, technology, and policy—with specific radiative forcing targets.
Key examples included SSP1-2.6 (sustainability-focused), SSP2-4.6 (middle-of-the-road with modest mitigation), and SSP5-8.5 (high fossil-fuel growth with no strong policies). The trailing number again indicated the forcing level in W/m² by 2100.
These scenarios served multiple roles: driving Earth system models to simulate temperature, precipitation, extremes, and sea-level rise; assessing impacts on ecosystems, agriculture, and infrastructure; evaluating mitigation costs and policy options; and supporting risk management under uncertainty. Differences stemmed from assumptions about population and GDP growth, energy mixes (coal-heavy versus renewables and nuclear), technological progress, international cooperation, and model-specific factors like climate sensitivity and carbon-cycle feedbacks.
The extreme scenarios, particularly RCP8.5 and SSP5-8.5, assumed a massive resurgence in coal use—up to fivefold increases—along with high energy demand and very slow decarbonization. Under SSP5-8.5, CO₂ concentrations were projected to exceed 936 parts per million by 2100, potentially driving warming as high as 11°F (about 6°C). These relied on Integrated Assessment Models with outdated assumptions, including “return-to-coal” hypotheses based on untested resource extraction models. Critics noted that these baselines were increasingly disconnected from reality.
By the mid-2010s, clear discrepancies emerged. Global coal use peaked around 2013 and then flattened. Costs for solar and wind fell far faster than anticipated. Emissions growth slowed dramatically from about 3% per year in the 2000s to near-flat levels post-2010. Real-world trajectories aligned far more closely with moderate scenarios like RCP4.5 or SSP2-4.5. Meta-analyses confirmed that IAMs had systematically overestimated future coal use.
As renewable energy economics improved, the extreme coal expansion assumed in high-end scenarios became highly implausible. Forcing by 2100 appeared far more likely to hover near 6 W/m² than 8.5.
Despite this, thousands of studies continued using the high-end scenarios to model impacts such as extreme heat, sea-level rise, and ecosystem disruption. This amplified perceptions of high-end risks in public and policy discussions, often without sufficient emphasis on their low probability. While defenders argued the extremes helped stress-test models and highlight vulnerabilities, emissions data increasingly showed that current policies project warming of roughly 2.5–3°C by 2100—consistent with moderate pathways like SSP2-4.5.
A notable course correction arrived in April 2026 with a paper outlining the Scenario Model Intercomparison Project for CMIP7, which will underpin the IPCC’s Seventh Assessment Report. It explicitly labels CMIP6’s highest forcing levels (such as SSP5-8.5) as “implausible.” The new framework introduces seven scenarios ranging from very low to high, without the old extreme labels but with clearly defined 2100 radiative forcing targets.
The most important change is the absence of any RCP8.5 or SSP5-8.5 equivalent. The new high-end scenario reaches only 6.7 W/m², with annual CO₂ emissions around 71 gigatons in 2100—compared to 128 gigatons in SSP5-8.5. Cumulative emissions are correspondingly lower. The previous SSP3-7.0 pathway has also been dropped, as the new high-end level features about 9% less forcing. The updated scenarios build on revised SSP narratives with more pragmatic socioeconomic assumptions. For instance, earlier high-population growth projections, particularly for Africa, had inflated some extreme outcomes; sensitivity analyses showed demographics play a meaningful role in warming projections.
This shift carries several implications. Thousands of prior studies based on the most extreme scenarios will need careful contextualization or may become less central. Future CMIP7 modeling will use more realistic ranges, treating the highest scenarios as “what-if” explorations rather than core baselines. This should reduce overemphasis on catastrophic outcomes and better align projections with observed trends in energy markets and emissions. It also encourages greater focus on near-term information, adaptation, and socioeconomic realism.
The change represents a healthy scientific self-correction. Empirical data on energy transitions, renewable cost declines, and emissions trajectories have prevailed over modeling inertia. It echoes earlier evolutions in IPCC scenario development—from the 1992 scenarios to the 2000 Special Report on Emissions Scenarios, then to RCPs and SSPs—where real-world emissions have consistently trended below the worst-case assumptions.
That said, the push for aggressive net-zero policies continues in many circles, and models will still emphasize the value of mitigation for “climate stabilization.” For the Seventh Assessment Report, however, expect moderated high-end projections and a more grounded assessment overall.
While uncertainties remain—climate sensitivity, technological breakthroughs, policy choices, and natural variability—the CMIP7 update is a positive development. It moves literature away from implausibly extreme baselines toward scenarios that better reflect plausible futures. As always in climate science, time and better data will continue to refine our understanding.
This recalibration offers hope for more balanced, evidence-driven discussions about risks, adaptation, and policy in the decades ahead. Only time will tell as even the new highest scenarios are overstated.
Dr. David R. Legates is a Retired Professor at the University of Delaware and is the Director of Research and Education at the Cornwall Alliance for the Stewardship of Creation. He also is an Advisory Board Member of A Better Delaware.