Earth is absorbing heat at roughly twice the pace it was two decades ago, and the real-world measurements now sit well beyond what most climate models predicted. A peer-reviewed study published in Geophysical Research Letters, led by Steven Sherwood of UNSW Sydney and Benoit Meyssignac of CNES, found that the planet’s energy imbalance doubled between 2005 and 2019. The gap between what satellites observe and what simulations forecast raises hard questions about whether current climate projections are keeping up with physical reality.
What the Satellites and Oceans Reveal
The study draws on two independent measurement systems. NASA’s Clouds and the Earth’s Radiant Energy System, known as CERES, tracks how much solar energy the planet reflects back to space and how much infrared radiation it emits. The Argo network, a global fleet of ocean-profiling floats, records how much heat the deep ocean absorbs. When both datasets tell the same story, the signal is difficult to dismiss. According to the AGU release, the two lines of evidence agree: Earth’s energy imbalance, the difference between incoming solar radiation and outgoing heat, roughly doubled over that 14-year window.
That metric matters because it functions as a running ledger of the climate system’s trajectory. A planet in energy balance radiates as much heat as it receives. A planet out of balance stores the surplus, mostly in the ocean, and that stored energy eventually surfaces as higher air temperatures, stronger storms, and accelerating ice loss. The doubling documented in this study means the rate at which Earth accumulates excess energy has increased sharply, not just the total amount stored.
Much of the technical and institutional backbone for this work comes from the broader Earth and space science community coordinated through organizations like the American Geophysical Union, which convenes researchers, sets data standards, and helps translate new findings into public-facing assessments. By combining satellite observations, in-situ measurements, and model output, these networks turn raw streams of data into a coherent picture of how quickly the climate system is changing.
Why Models Fell Short
Most climate simulations capture the broad arc of warming driven by rising greenhouse gas concentrations. But the observed acceleration in heat trapping has outpaced the range that models projected for recent decades. Real-world measurements of extra heat the Earth is trapping sit well beyond most model outputs, according to researchers at UNSW. That discrepancy is not a minor bookkeeping issue. If models consistently underestimate the pace of energy accumulation, they also underestimate the speed at which downstream impacts, from sea-level rise to marine heatwaves, will arrive.
Several physical mechanisms help explain the mismatch. The AGU press release identifies reduced cloud cover and shrinking sea ice as key contributors on the solar side: less reflective surface means more sunlight absorbed. On the infrared side, rising concentrations of greenhouse gases and water vapor trap more outgoing radiation, preventing it from escaping to space. Each of these feedbacks can amplify the others. Less ice exposes darker ocean, which warms further, which adds moisture to the atmosphere, which traps still more heat. Models represent these loops, but their timing and strength in simulations have lagged behind what the instruments record.
Scientists are now using targeted diagnostics to determine whether the problem lies mainly in how models handle clouds, sea ice, aerosols, or internal variability. That effort leans on shared archives of simulations and observations, often accessed through tools such as the AGU publications database, which allow teams to compare independent studies and test whether apparent discrepancies hold up across multiple data sets and methods.
A Modeling Protocol Built to Close the Gap
The climate science community has recognized the problem and designed a formal response. A coordinated protocol known as CERESMIP, published in Frontiers in Climate, was created specifically to diagnose why modeled energy imbalance trends diverge from CERES-era observations. The protocol asks modeling centers to isolate individual forcing agents and feedback processes so researchers can pinpoint which components of the climate system are responsible for the gap.
That kind of structured comparison is essential because the mismatch could stem from several sources. Models might underrepresent how quickly low-level clouds thin in a warming world, a problem that has dogged climate science for decades. They might also underestimate the speed of sea-ice retreat or the strength of water-vapor feedback in the tropics. CERESMIP is designed to separate these possibilities by running targeted experiments against the satellite record. Until those experiments produce results across multiple modeling centers, the exact share of the discrepancy attributable to each mechanism remains an open question.
The protocol itself reflects a broader shift toward collaborative, open modeling frameworks, supported by initiatives such as publishing partnerships that encourage standardized experiments and shared code. By aligning how models are run and evaluated, these efforts make it easier to identify which shortcomings are common across the field and which are tied to specific model designs.
Observation Networks Under Pressure
The study’s reliance on CERES and Argo also highlights a practical vulnerability. CERES instruments measure reflected shortwave and emitted longwave radiation from orbit, providing the top-of-atmosphere half of the energy budget. Argo floats supply the ocean-heat half. Both systems require sustained funding and satellite continuity. Any gap in coverage would leave scientists unable to track whether the imbalance continues to grow, stabilizes, or, in the most optimistic scenario, begins to shrink as emissions fall.
A separate review paper published in Surveys in Geophysics assessed multiple lines of observational evidence for energy imbalance changes since 2000 and found broad agreement between CERES, in-situ ocean estimates, and model simulations on the direction of the trend, even as the magnitude diverged. That review also stressed that internal climate variability, natural fluctuations in ocean circulation and cloud patterns, can temporarily amplify or mask the underlying signal, complicating efforts to attribute any single year’s reading to long-term forcing alone.
Maintaining and improving these observing systems depends not just on scientific need but on policy choices and financial support. Professional societies now provide dedicated channels for members and the public to contribute, including targeted funds such as AGU philanthropy programs that back early-career researchers, data infrastructure, and new instrumentation. Those investments help ensure that future assessments of Earth’s energy imbalance rest on continuous, high-quality records rather than fragmented snapshots.
At the same time, the people who run these networks and analyze their output rely on training and collaboration platforms, from conferences to online portals. Many of those services are integrated into membership systems like AGU’s login hub, which provides access to datasets, workshops, and working groups focused on topics such as radiative forcing, ocean heat content, and model evaluation. The institutional scaffolding may seem far removed from the physics of clouds and currents, but it underpins the collective capacity to track and understand a rapidly changing planet.
What a Faster Energy Buildup Means
For anyone trying to gauge how quickly climate risks will intensify, the doubling of Earth’s energy imbalance is a concrete, measurable warning. It means the planet is not just warming. The rate at which it stores the energy that drives warming has itself accelerated. That distinction matters for coastal planning, agricultural adaptation, and infrastructure investment, because it compresses the timeline for impacts that policymakers once assumed would arrive later in the century.
The finding also challenges a common assumption embedded in public debate: that climate models represent a worst-case framing. In this instance, the models appear to have been too conservative. Steven Sherwood, a co-author of the study, has argued that the real world is currently warming toward the upper end of what models anticipated for this period, suggesting that some impacts (such as marine heatwaves, glacier loss, and extreme rainfall) could arrive earlier or more intensely than many planning scenarios assume.
None of this means climate models are fundamentally broken; they remain indispensable tools for exploring future pathways and testing policy choices. But the emerging evidence that Earth is trapping more heat than forecast underscores the need to continually recalibrate those tools against observations. As new CERESMIP experiments, satellite missions, and ocean surveys come online, scientists will refine estimates of how quickly the imbalance might grow or stabilize under different emissions trajectories.
For now, the message is stark but actionable. The planet is taking up extra energy faster than it did at the start of the century, and physical systems, from polar ice sheets to coral reefs, are responding. Reducing greenhouse gas emissions remains the only way to ease that imbalance in the long term. In the meantime, the combination of better observations, more transparent modeling protocols, and stronger scientific institutions offers the best chance of keeping our expectations aligned with the climate system we actually inhabit, rather than the one our models once assumed.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
