The Cohler et al. paper fails fundamental tests of scientific credibility on nearly every dimension examined. Published in a journal founded by a Norwegian climate denial organization, authored by individuals with no ocean science expertise — including a clarinetist, a high school student, and researchers with documented fossil fuel industry funding — the paper recycles long-rebutted arguments to claim that one of modern science's most robust observational systems produces "physically meaningless" data. Every major claim in the paper is contradicted by the peer-reviewed literature, multiple independent measurement systems, and basic methodological analysis. This report evaluates each dimension in detail.
"Science of Climate Change" (SCC-Publishing, Lysaker, Norway, ISSN 2703-9072) is not indexed in Web of Science, Scopus, or the Directory of Open Access Journals — the three standard markers of a legitimate scientific journal. Founded in 2021, the journal originated from Klimarealistene ("Climate Realists of Norway"), a Norwegian climate change denial organization, and its earlier content was hosted at scc.klimarealistene.com. A separate SCC Publishing association was created in 2025 to present a more independent appearance.
The journal's content is exclusively climate-skeptic. Published papers argue CO₂ increases are natural rather than anthropogenic, challenge the greenhouse effect itself, and claim temperature records are corrupted. Most notably, Vol. 5.1 (June 2025) published a paper listing Grok 3 beta (an AI model) as lead author, with a high school student as co-author — an arrangement no mainstream journal would permit.
The editorial board consists entirely of known climate contrarians. Chief Editor Hermann Harde, a retired experimental physics professor, had his 2017 paper in Global and Planetary Change flagged by three board members of that journal for soliciting only climate-skeptic reviewers — what RealClimate described as "pal review." Co-editors include Ole Humlum (a co-author of the paper under review, creating a direct conflict of interest), Guus Berkhout (founder of CLINTEL, a climate denial organization), and Henrik Svensmark (proponent of the cosmic ray theory of climate change). The journal's "About" page explicitly states its mission is to present "alternative views," positioning it as an advocacy outlet rather than an objective scientific publication.
The author list reveals a striking absence of relevant expertise for a paper challenging the foundations of physical oceanography:
Jonathan Cohler is a professional classical clarinetist, conductor, and music educator. He holds a 1980 Harvard bachelor's degree in physics but has no graduate training, no research position, and no publications in mainstream scientific journals. He runs the website climatethetruth.com and has publicly stated that "NOBODY who understands advanced math, statistics, and physics...could possibly believe in the #ClimateScam." His listed institutional affiliation is "Cohler & Associates, Inc." — a business entity, not a research institution. The journal's correspondence email is his personal Gmail address.
David R. Legates, a geography professor at the University of Delaware, has documented ties to the fossil fuel industry. He co-authored four of the eleven papers for which Willie Soon received undisclosed fossil fuel funding. His appointment as NOAA Deputy Assistant Secretary in 2020 was widely criticized by scientists, and he was later reassigned from the White House Office of Science and Technology Policy for producing unauthorized climate-skeptic materials. He is affiliated with the Heartland Institute and Competitive Enterprise Institute.
Willie Soon (Wei-Hock Soon), who holds a PhD in aerospace engineering (not climate science), has received over $1.2 million in documented funding from ExxonMobil, the American Petroleum Institute, the Koch Foundation, and Southern Company. FOIA documents revealed he referred to his research papers as "deliverables" in progress reports to Southern Company. He failed to disclose fossil fuel funding in at least 11 published papers, violating ethical guidelines of eight journals. The Harvard-Smithsonian Center for Astrophysics, where he was an externally funded employee, expressly prohibited him from researching climate change "on the clock."
Franklin Soon, listed as affiliated with Marblehead High School in Marblehead, Massachusetts, is Willie Soon's son and a high school student. His inclusion as co-author on a paper purporting to overturn major IPCC findings underscores the journal's lack of peer review standards.
Kesten C. Green is a forecasting methods researcher at the University of South Australia Business School with no climate science training. Ole Humlum is a retired geomorphologist who sits on the editorial board of the very journal publishing this paper — a clear conflict of interest. Neither has published in mainstream climate science journals.
The paper's central theoretical claim — that global temperature metrics are "physically meaningless" because temperature is an intensive thermodynamic property — derives from Essex, McKitrick & Andresen (2007), published in the Journal of Non-Equilibrium Thermodynamics. This argument has been thoroughly examined and rejected by the scientific community.
The most devastating critique is computational. Essex and McKitrick used Celsius rather than Kelvin for geometric mean calculations. Since Celsius includes negative values, geometric means involving 0°C yield zero regardless of other values — an elementary error that invalidates their comparison of averaging methods. Additionally, McKitrick's spreadsheet treated missing values as 0°C (an Excel default), introducing an artificial cooling bias.
On the physics, multiple rebuttals have demonstrated that the atmosphere's approximately linear relationship between energy and temperature (E = VCT for a nearly homogeneous gas) provides direct physical justification for arithmetic averaging. Climate scientists use temperature anomalies — deviations from a local baseline — which remove absolute temperature differences and make averaging highly robust. The arithmetic mean of anomalies enhances the common forced signal (greenhouse warming) while suppressing spatially incoherent noise.
A 2023 paper in Physical Review E by Allahverdyan et al. provided a rigorous thermodynamic definition of mean temperature based on equilibration processes, directly contradicting Essex et al.'s claim that no physical basis exists. As climate scientist Rasmus Benestad noted at RealClimate, the Essex argument "doesn't bring any new revelations" — it would equally invalidate regional, synoptic, or even room-scale temperature averages if taken seriously. The paper has been cited almost exclusively by other climate-skeptic publications and has not generated any productive research program.
The paper's application of the Nyquist-Shannon sampling theorem to argue that ~4,000 Argo floats cannot adequately sample the ocean rests on multiple conceptual errors.
First, the Nyquist-Shannon theorem applies to band-limited signals sampled at regular uniform intervals. Argo floats drift freely and are irregularly spaced — the classical theorem does not directly apply. Second, the theorem addresses whether a continuous signal can be perfectly reconstructed from discrete samples. Oceanographers are not attempting perfect reconstruction; they are estimating large-scale means and trends with quantified uncertainty. These are fundamentally different problems.
Third, modern signal processing theory — particularly compressed sensing (Candès, Donoho, Tao, ~2006) — demonstrates that signals can be accurately reconstructed from far fewer samples than the Nyquist rate when the signal is sparse or compressible in some domain. Ocean temperature fields concentrate most variance in large-scale, slowly-varying modes, making them highly compressible.
Oceanographers address spatial sampling through well-established methods: optimal interpolation using spatial and temporal covariance functions, Gaussian process regression with data-driven covariance parameters (Kuusela & Stein, 2018, Proceedings of the Royal Society A), and data assimilation combining observations with physics-based ocean models. These methods include formal uncertainty estimates that increase in data-sparse regions. Argo was explicitly designed for "accurate estimates of large-scale (order 100s of km), month-to-month variability" — its ~3° resolution matches its design target. The inability to resolve individual ocean eddies is a known, accepted, and irrelevant limitation for global heat content estimation.
The Argo program deploys approximately 4,000 autonomous profiling floats producing over 100,000 temperature-salinity profiles annually from the surface to 2,000 meters. Temperature sensors achieve accuracy of ±0.002°C, validated against ship-based CTD measurements. A rigorous two-tier quality control system — automated real-time checks followed by expert delayed-mode analysis — ensures data integrity. Over 6,000 scientific publications and nearly 500 PhD theses have used Argo data.
The limitations the paper raises are well-known and openly discussed in the oceanographic literature — they are not novel criticisms. Deep ocean gaps below 2,000 meters are being addressed by Deep Argo floats reaching 4,000–6,000 meters. Polar undersampling is managed through ice-avoidance firmware and specialized deployments, with OHC typically calculated 60°S–60°N. Near-coast gaps exist because floats require deep water. Each limitation has been quantified, published, and incorporated into uncertainty estimates.
Critically, multiple independent groups — IAP/CAS (China), NOAA/NCEI (USA), JAMSTEC (Japan), Hadley Centre (UK), and Scripps (USA) — produce OHC estimates using different methods, quality control procedures, and interpolation techniques. All show consistent warming trends. For 2024, IAP/CAS found OHC exceeded 2023 by 16 ± 8 ZJ, while the CIGAR-RT reanalysis found 18 ± 7 ZJ and Copernicus Marine found 40 ± 31 ZJ — independently confirming record ocean heat. This multi-product agreement is powerful evidence of robustness.
The paper claims the CERES EBAF adjustment procedure creates circular reasoning because satellite measurements are "anchored" to OHC data. This reflects either a misunderstanding or a mischaracterization of the methodology.
CERES satellites measure incoming solar radiation (~340 W/m²) and outgoing radiation (~339 W/m²). The Earth Energy Imbalance (~0.7–1.0 W/m²) is the tiny difference — about 0.3% of these large fluxes. Direct satellite radiometers achieve absolute accuracy of roughly ±1% in reflected shortwave, translating to uncertainties of ~4 W/m² in net flux — one order of magnitude larger than the signal. Raw unadjusted satellite data yields an unrealistic EEI of ~4.3 W/m².
The EBAF product applies a one-time constant offset so that global mean net TOA flux over July 2005–June 2015 matches the in-situ EEI value of 0.71 W/m². This same constant is applied uniformly across the entire record. Only the absolute level is set this way — all temporal variability, trends, interannual variations, and regional patterns are determined independently by the satellite data. This is not circularity; it is applying the conservation of energy to calibrate a measurement whose relative precision (~0.2%/decade stability) far exceeds its absolute accuracy.
The proof lies in the results: after this one-time anchoring, independent satellite-derived trends match independent OHC-derived trends. Loeb et al. (2021, GRL) found both methods yield statistically indistinguishable decadal increases of 0.50 ± 0.47 W/m² per decade from mid-2005 to mid-2019. As Norman Loeb of NASA Langley stated: "The two very independent ways of looking at changes in Earth's energy imbalance are in really, really good agreement."
The paper's claim that EEI uncertainty exceeds ±1 W/m², rendering the ~0.7 W/m² estimate "statistically indistinguishable from zero," conflates different types of uncertainty in a way that is methodologically unsound.
The ±4 W/m² figure applies to raw, unadjusted satellite absolute accuracy — a number that the OHC anchoring procedure is specifically designed to address. Once properly constrained, published uncertainty estimates tell a very different story:
The paper's RSS (root-sum-of-squares) combination of uncertainties is appropriate only when error sources are truly independent and random. Treating known systematic offsets — which the OHC anchoring corrects — as additional independent random errors and adding them via RSS inflates the total uncertainty artificially. The paper also appears to ignore that for detecting trends, measurement stability matters more than absolute accuracy. CERES stability of ~0.2%/decade is far better than its ~1% absolute accuracy, making trend detection highly robust.
Most critically, the space geodetic approach — combining satellite altimetry with GRACE gravity measurements to derive thermosteric sea level change — provides an EEI estimate completely independent of Argo floats. Hakuba et al. (2021) obtained 0.94 W/m² using this method, confirming EEI is robustly positive through an entirely separate observational system.
The paper's framework implicitly assumes that if Argo data were invalidated, the case for ocean warming would collapse. This is false. At least six independent lines of evidence corroborate ocean warming and positive EEI:
Satellite altimetry measures global mean sea level rise at 3.7 ± 0.25 mm/year since 1999, accelerating to 4.62 mm/year for 2013–2022. Approximately 30% is attributable to thermal expansion. In 2024, over 70% of the 0.59 cm annual sea level rise was due to thermal expansion — directly reflecting ocean heat uptake.
GRACE/GRACE-FO gravity satellites independently measure ice mass loss: Greenland sheds ~264–270 Gt/year and Antarctica ~135–150 Gt/year. Subtracting this ocean mass change from altimetric total sea level yields the thermosteric component — an OHC estimate independent of Argo.
Marine biological indicators provide qualitative but powerful confirmation. Marine species are shifting poleward at 72 km per decade — over 10× faster than terrestrial species (Poloczanska et al., 2013, Nature). Mass coral bleaching events are increasing in frequency, with the 2014–2017 event affecting ~80% of surveyed reefs. Over 81% of 1,735 observed changes in marine ecosystems are consistent with warming.
Atmospheric temperature trends from eight independent global surface temperature datasets all show 2024–2025 among the warmest years, at 1.33–1.53°C above pre-industrial baselines. Pre-Argo ship-based measurements (XBTs), once corrected for known biases, extend ocean warming trends back to the 1960s. The consistency across all these systems — each with independent error characteristics — provides what scientists call a "closure" of the Earth's energy budget.
The paper's acknowledgment of extensive AI use (Grok 4.1 beta, Claude 4.5, Gemini 3 Pro, ChatGPT 5.2) for drafting, editing, and "peer review" raises fundamental concerns. No major scientific journal or body endorses AI as a substitute for expert human peer review.
Nature states that human peer review expertise is "invaluable and irreplaceable." Science (AAAS) explicitly prohibits AI use in reviewing. Elsevier bars reviewers from uploading manuscripts to AI tools. NIH and NSF prohibit AI in peer review and grant processes. A JAMA Network Open study of the top 100 medical journals found 59% explicitly prohibit AI use in peer review.
The limitations are well-documented: AI lacks domain expertise for complex scientific evaluation, cannot assess novelty or originality, introduces confidentiality risks, generates hallucinations, and — crucially — cannot be held accountable for review conclusions. Using an AI model to evaluate one's own paper against a checklist designed by one's own co-author (Green's "Compliance with Science Checklist" co-authored with Armstrong) compounds the problem. The Armstrong & Green checklist itself is not adopted by any major journal and explicitly claims "one does not need to be an expert in the field" to evaluate scientific compliance — a position antithetical to the purpose of peer review.
The paper credits Nobel laureate John Clauser's presentations claiming "Climate Change is a Myth" and proposing a "cloud thermostat" mechanism. Clauser won the 2022 Nobel Prize for quantum entanglement experiments — expertise that does not transfer to climate science. He has published zero peer-reviewed papers on climate and admitted he has "not talked to any of the modelers."
His central claim — that clouds create a thermostat so powerful it prevents CO₂-driven warming — is directly contradicted by IPCC AR6's high-confidence finding that net cloud feedback is positive (+0.42 W/m²/°C), meaning clouds slightly amplify warming. A 2022 study by Ceppi et al. found less than 2.5% probability that cloud feedback is negative. His claim that IPCC models don't include clouds is factually false — climate models have included cloud processes since at least the 1970s. NASA GISS director Gavin Schmidt published a detailed point-by-point rebuttal, and Michael Mann characterized Clauser's arguments as "pure garbage" and "pseudoscience."
If Clauser's cloud thermostat operated as described, Earth could not sustain the observed energy imbalance, could not have experienced ice age cycles, and marine species would not be migrating poleward at 72 km per decade.
The contrast between Cohler et al.'s claims and the mainstream literature could not be sharper. Recent high-impact publications paint a picture of accelerating ocean warming confirmed by multiple independent methods:
Cheng et al. (2025, Advances in Atmospheric Sciences) — a 54-author, 7-country collaboration — found OHC in 2024 exceeded 2023 by 16 ± 8 ZJ, confirmed by two independent datasets. Warming rates have increased 2–3× since the late 1980s: from ~2.9 ZJ/year (1958–1985) to ~9 ZJ/year post-1986. Pan, Cheng et al. (2026) report that 2025 OHC increased by ~23 ± 8 ZJ — the largest annual increase since 2017. Goessling et al. (2025, Nature) found the 2023–2024 sea surface temperature jump "would have been practically impossible" without anthropogenic warming. Mauritsen et al. (2025, AGU Advances) report EEI reached ~1.9 W/m² in 2023, exceeding climate model predictions.
Every year from 2019 through 2025, ocean heat content has set a new record. This unbroken streak of records, confirmed across multiple independent analysis groups using different methods, is the signature of a physically real and accelerating phenomenon — not a "computational construct."
This evaluation reveals a paper that fails on every criterion of scientific credibility. It is published in an unindexed journal founded by a climate denial organization, authored by individuals without relevant expertise and with documented financial conflicts of interest, built on arguments that contain known computational errors and have been rebutted for nearly two decades, and "peer reviewed" by AI chatbots rather than qualified scientists. Its central claims — that Argo-based OHC is meaningless, that EEI is indistinguishable from zero, and that CERES-OHC methodology is circular — are each contradicted by multiple independent lines of evidence published in Nature, Science, GRL, and other leading journals. The paper does not identify a single limitation of ocean observing systems that is not already openly discussed, quantified, and accounted for in the mainstream literature. Rather than advancing scientific understanding, it exemplifies how the apparatus of scientific publishing — formatting, citations, mathematical notation — can be used to lend an appearance of rigor to conclusions that are predetermined by ideology and unmoored from evidence.