Bob Kopp
LightNews LightNews
bobkopp.net
Bob Kopp
@bobkopp.net
14K followers 6.8K following 2.2K posts
#Climate & sea level science + policy. Rutgers University. All views my own. www.bobkopp.net
Posts Replies Media Videos
bobkopp.net
@mikiesherrill.bsky.social was elected on a platform aggressively critical of PJM
4) I will instruct my Attorney General to take Trump and New Jersey's grid operator, PJM, to court - in coordination with governors in our region - to force them to end their mismanagement and instead accelerate the development of cheaper and cleaner energy sources and connect new power generation to our grid immediately. a) New energy projects can't lower utility bills if they're waiting years to start producing energy, but PJM refuses to connect them quickly. b) President Trump is actively thwarting the development of new, needed power plants. I will take him on over his destructive efforts to block crucial power supplies, which will reduce our energy rates and make our air cleaner.
November 27, 2025 at 2:03 AM
bobkopp.net
Because I speak English?
alternative /ôl-tûr'na-tiv, ăl-/ 二 noun 1. One of a number of possible choices or courses of action. "There are plenty of alternatives to conventional advertising. " 2. A choice or course of action that is mutually exclusive with another: synonym: choice. "The alternative to staying in that dead-end job is to quit." Similar: choice 3. A situation presenting a choice between two mutually exclusive possibilities. "people facing the alternative of selling their homes or going bankrupt." adjective 1. Being one of a number of possible choices or courses of action. "The highway is closed. Seek alternative routes." 2. Being one of two mutually exclusive choices or courses of action. "How effective is the alternative treatment?" 3. Substitute or other. "Several members of the audience provided alternative views on the topic." The American Heritage® Dictionary of the English Language, 5th Edition • More at Wordnik
November 22, 2025 at 2:39 PM
bobkopp.net
Similar story here, though we have been working to optimize a heat pump system (4-head Fujitsu system) that was installed before we bought the house seven years ago. We didn’t disconnect, so most months are just paying $10/month service charge. Heat pump water heater and induction stove.
Bar chart of gas usage over time
November 22, 2025 at 1:52 PM
bobkopp.net
@agu.org has shown some really impressive leadership in this difficult year. Really acting like a “Union”, not just an “Association” or a “Society.” Kudos to both their elected and staff leadership.
AGU continues to serve as a nexus for action. We remain actively engaged in litigation to oppose the unjust firings of U.S. federal scientists and the capricious re-organization of federal agencies that harm the scientific enterprise. AGU also urges lawmakers in Congress to fulfill their responsibility to the nation by completing full-year appropriations for federal science agencies and ensuring those funds are protected and used as intended. Stopgap measures and short-term budgets are not enough. Science cannot operate on uncertainty, nor can the public rely on agencies hamstrung by political stalemate. The United States needs stable, long-term investment and clear guardrails so that every dollar strengthens public safety, fuels innovation, and advances our shared future. But Congress will only act if we make our voices impossible to ignore. Every scientist, every student, every advocate who believes in evidence-based decision-making has a role to play. Visit AGU's Science Policy Action Center today to contact U.S. lawmakers and demand the support that science and the people it serves deserve Together, we can keep science standing strong for the nation and the world
November 13, 2025 at 1:51 PM
bobkopp.net
Also, you can’t fit an exponential to this curve. That doesn’t mean rapid acceleration of ice sheet loss is not a concern — it absolutely is — but that you need to understand something about the relevant physics and not just curve fit like Hansen did in the 2000s and 2010s
Antarctic ice sheet mass changes, as measured by GRACE gravity satellites. Because of increased accumulation of snow in a warming climate net Antarctic mass stabilizes in the 2020s, even though substantial loss is continuing in the Amundsen Sea Embayment of West Antarctica.
November 12, 2025 at 1:59 AM
bobkopp.net
In line with projections.
How do the STAP projections for New Jersey compare with observations? In comparing the STAP projections to tide-gauge observations, it is essential to be cognizant that the STAP is projecting changes in 19-year average sea level (e.g., the reported 2040 projections are the projected average sea level over the period 2031-2049). The projections do not include interannual variability, which at Atlantic City (for example) adds + 0.2 ft (0.06 m) variability (2ơ) around the 19-year average. The closest SLR projection the STAP can compare to observations is the 2020 projection (or the measured average of sea level during the period 2011-2029). Since the 2011-2029 average sea-level has not yet been fully observed at the time of publication, the STAP must extrapolate the observed tide-gauge record to estimate the 2011-2029 average sea-level. The Federal Interagency SLR report (Sweet et al., 2022) constructed quadratic extrapolations of the 1970-2020 tide-gauge record. The Federal Interagency SLR report partially removed natural variability through regression on climate indices representing the El Niño-Southern Oscillation, Pacific Decadal Oscillation, and North Atlantic Oscillation. The Federal Interagency SLR report also accounted for serially correlated variability in estimating the associated uncertainty. The Federal Interagency SLR report's extrapolations for 2020 (adjusted to match the 2005 or 1995-2014 baseline period used in this report) are shown in Table 8. In New Jersey, the extrapolated value in 2020 relative to 1995-2014 ranges from 0.31 ÷ 0.11 ft (0.09 ÷ 0.03 m) at Atlantic City to 0.37 ‡ 0.13 ft (0.11 ‡ 0.04 m) at Cape May. The notable diference between the two relatively proximal sites likely reflects differences in subsidence, potentially associated with groundwater withdrawal.
November 11, 2025 at 8:16 PM
bobkopp.net
Here’s what we had to say about that in the US government’s 2010 social cost of carbon report. Details slightly different, basic picture not so much. And differences are all at a level of detail not provided in the RS report.
As noted above, the damage functions underlying the three IAMs used to estimate the SCC may not capture the economic effects of all possible adverse consequences of climate change and may therefore lead to underestimates of the SCC (Mastrandrea 2009). In particular, the models' functional forms may not adequately capture: (1) potentially discontinuous "tipping point" behavior in Earth systems, (2) inter-sectoral and inter-regional interactions, including global security impacts of high-end warming, and (3) limited near-term substitutability between damage to natural systems and increased consumption. It is the hope of the interagency group that over time researchers and modelers will work to fill these gaps and that the SCC estimates used for regulatory analysis by the Federal government will continue to evolve with improvements in modeling. In the meantime, we discuss some of the available evidence.
October 28, 2025 at 1:31 AM
bobkopp.net
Actually, it’s been a few years since I read John Summers’ essay, I’d forgotten this bit about robot-assisted ABA: www.linguafranca.com/p/the-mismea...
everyone. Skinner's technique of reinforcement has shaped the design of video games, dating apps, slot machines, social media, and product marketing. The digital architecture of mass behavior modification, busy with prompts, notifications, and nudges, mostly just aims to herd us into goals chosen for us. The automation of life is plain to see in the unfolding future of ABA. Seventy ABA classrooms in New England already use robots for autism instruction. I telephoned the manufacturer in Connecticut to pose a question missing from the excited newspaper stories. No, a spokesperson avowed, the company hasn't collected any data to justify its claims for robot-assisted ABA. The evidence is beside the point, but the irony is rich. Trained to disregard the inner lives of their clients, behavior analysts themselves may be replaced by robots.
October 28, 2025 at 12:26 AM
bobkopp.net
Somehow I had missed this May White House budget fact sheet, making use of a common right-wing dialectical spelling of “Jewish” www.whitehouse.gov/wp-content/u...
ENDING THE GREEN NEW SCAM President Trump is committed to eliminating funding for the globalist climate agenda while unleashing American energy production. The President's FY 2026 Budget eliminates funding for the Green New Scam.
October 20, 2025 at 9:15 PM
bobkopp.net
This is a useful table, but it seems like it's making a strawman argument about climate change governance without Earth system tipping points.

Most of these characteristics apply to climate change in general, at least if you're concerned about societal impacts.
GTP 2025, Table 1.1.1: Differences between conventional climate change governance and governance for ESTPs Characteristics include: thresholds, acceleration of change, increased magnitude of impacts, trend reversals, different distribution of impacts, novel impacts, and cascade effects.
October 14, 2025 at 2:30 PM
bobkopp.net
Sent Steve a letter:
Dear Steve, I hope you are well. In your Monday WSJ column, you criticize the National Academies for not reviewing your DOE Climate Working Group report. To my knowledge, the Academies have never offered an unrequested review of a government scientific report. I am sure they would have been happy to review your report if DOE had asked them to. Though DOE chose not to do so, under Information Quality Act guidelines, an independent peer review (like that the Academies often conducts for the federal government) is required for it to be legal to use the report in rule making. (See Section III in 70 FR 2664) In your Monday column, you do not mention that — seeing no evidence that your report had been the subject of any meaningful independent peer review — more than 85 experts volunteered our time in the month of August to review your report. The approximately 450-page compendium of comments, co-edited by Andrew Dessler and me, evaluates each portion of your report. In case you have not seen it, you can find the expert review here: https://doi.org/10.22541/essoar.175745244.41950365/v2 Your column recapitulates a number of the errors and misrepresentations in the DOE report, which are evaluated in detail in the expert review. For example, you state that US heat waves “aren’t more common in recent decades than they were in the decades around 1900”. As we write in the expert review, the DOE report “highlights the Dust Bowl period of the 1930s as a time of unusually hot temperatures in the United States.” However, “over such a small area [as the contiguous US], natural variability can play a large but temporary role. The Dust Bowl was indeed an exceptional event, although unrelated to greenhouse-gas forcing, and the decadal climate conditions that drove the Dust Bowl are not occurring today…. If you take out the 1930s, there are clear trends [in the contiguous US] towards more extreme temperatures due almost certainly to radiative forcing.” Further, “if one examines a larger region, where the Dust Bowl’s influence is diluted, then there is an obvious, strong trend towards more extreme temperatures.” You also introduce in your column some errors that are not in the DOE report. For example, the DOE report addresses global-mean sea-level rise only in passing, focusing instead on specific US tide-gauge records. In your column, you claim that a global sea-level acceleration comparable to the current one occurred in the 1930s, a claim we discussed before about eight years ago. In 2017, I told you, based on our analysis in Hay et al. (2015), that the rate of global-mean sea-level rise in the 21st century was likely faster than during its previous high in the 1930s-1940s (notably, also a period of significant global-mean warming, preceding the slower sea level rise and lull of warming in the 1950s-1970s). At the time, global-mean sea-level acceleration had already been sustained for close to five decades, compared to about two decades in the 1920s and 1930s. The evidence in the last eight years has only gotten stronger. We are now in an approximately 60-year period of nearly unbroken global-mean sea-level acceleration, and the current rate of rise revealed by both satellite data and tide-gauge analyses — exceeding 4 mm/yr — is unequivocally higher than at any previous time revealed by tide-gauge reconstructions or satellite data (see, for example, Dangendorf et al., 2024; Figure 8 therein reproduced below). (Note that this finding is supported by Frederikse et al., 2020, a key figure from which is reproduced in a simplified format in your book Unsettled.) Further, combining tide-gauge and geological data shows that there has been a sustained acceleration in the long-term rates of rise since the mid-nineteenth century. Modern rates of 60-year-average global-mean sea level rise exceeded earlier Common Era rates by the 1860s, and the current rate of global-mean sea level rise virtually certain to be higher than any century-scale period in the last three millennia (Kemp et al 2018, Walker et al 2022; Figure 1 of the latter reproduced below). Furthermore, I don’t think either of us contest the fact that global-mean surface temperature is substantially higher today than in the 1930s and 1940s. It would, indeed, then be quite physically surprising if the rate of global-mean sea level rise were also not substantially higher. It would be tedious for both of us to respond further to the errors in your column, but I point you again to the expert review of your report: https://doi.org/10.22541/essoar.175745244.41950365/v2 I encourage you not to continue to recapitulate the errors in the DOE report while ignoring the readily available evaluations in the expert review. Best, Bob P.S. I know from our past interactions that you view private correspondence sent to you as being publishable without notifying or attempting to get the consent of your correspondents. You have my permission to fairly quote from this letter, which I will also be posting publicly.
October 8, 2025 at 1:19 PM
bobkopp.net
What’s more, geological data clearly reveal that the rate of global average sea level rise over the last sixty years was faster than during any comparable period in more than 2 millennia; and, indeed, this fact has been continuously true since the late 19th century. www.nature.com/articles/s41...
Walker et al 2022. Fig. 1: Time of emergence of global sea-level rise. a Sixty-year average rates over the Common Era, where pre-industrial is 0–1700 CE. b Sixty-year average rates from 1700 to 2000 CE which increase concurrently with the probability that each 60-year interval and all subsequent 60-year intervals were greater than a random 60-year interval during the pre-industrial Common Era. The time of emergence year is given for 0.66, 0.90, and 0.95 probabilities. Model predictions are the mean with 1σ uncertainty.
October 8, 2025 at 1:59 AM
bobkopp.net
But global average sea level today has been on a sustained acceleration for over 50 years, a duration without precedent in the observational record, and the rate of rise in the last decade is substantially higher than in the 1930s. essd.copernicus.org/articles/16/...
Dangendorf et al 2024. Figure 8. GMSLs and nonlinear rates in comparison to former reconstructions. (a) Shown are the GMSL time series from this study (1900 to 2022) in comparison to selected former reconstructions. All time series have been adjusted to a common mean of zero over the overlapping period from 1993 to 2010. (b) The corresponding nonlinear trends as determined by a singular spectrum analysis with an embedding dimension of 15, which extracts signals that are representative of timescales longer than 30 years. Uncertainties are shown as shaded regions and represent 95 % confidence intervals.
October 8, 2025 at 1:59 AM
bobkopp.net
I sent him an email, since he clearly hadn’t seen it…
To The.Secretary@hq.doe.gov Dear Secretary Wright, I saw that you stated in a press conference yesterday that "there is no disagreement with what's in the [Climate Working Group] report." You probably have not seen our 85+ climate expert review of the Climate Working Group report, which was submitted as a public comment on August 30, but which appears not yet to have been processed for posting on Regulations.gov. You can find our review here: https://doi.org/10.22541/essoar.175745244.41950365/v As explained at great length in the 450+ pages of our review, the Climate Working Group report currently fails to adequately represent the scientific understanding of climate change, and it would require very substantial revision before it could be relied upon by any federal agency or other entity. I hope that you find that our extensive expert review adequately addresses the question you raised in your press conference, "If the data and what we've said is wrong- as I read in countless newspaper articles, it's discredited data - what in the climate report is wrong?" Sincerely, Robert E. Kopp, Ph.D. Distinguished Professor, Department of Earth and Planetary Sciences Rutgers University
September 26, 2025 at 6:38 PM
bobkopp.net
Not a huge deal compared to everything else wrong with the DOE Climate Working Group report, but I just noticed that the report claims to be "Copyright (C) 2025 United States."

Works produced by the United States government are not protected by copyright and are in the public domain.
Title page of "A Critical Review of Impacts of Greenhouse Gas Emissions on the U.S. Cli,mate," a report by the US Department on Energy, claiming to be Copyright (C) 2025 United States.
September 26, 2025 at 1:51 PM
bobkopp.net
This is a baffling sentence — there seems to be some important premise hiding in that “therefore” that allows “predominantly genetic” to imply “largely untreatable”
From https://www.washingtonpost.com/health/2025/09/21/trump-autism-announcement-tylenol-leucovorin/ The findings on leucovorin have sent ripples through the scientific community, sparking renewed debate about the complex causes of autism - a condition some experts had recently considered predominantly genetic, and therefore largely untreatable.
September 21, 2025 at 7:58 PM
bobkopp.net
It may seem hard to focus in at a time when climate research and higher education are both under assault, but as I and my co-authors argue here: Higher education institutions can accelerate societal climate action

academic.oup.com/bioscience/a...
Higher education's roles in addressing the climate challenge: Workforce Development, Research, and Community Engagement, supported on a platform of Operations
September 17, 2025 at 6:13 PM
bobkopp.net
There are definitely still people out there not paying attention, but I haven’t found it that common. After all, this is what AR6 said:
A.4.3 A substantial 'emissions gap' exists between global GHG emissions in 2030 associated with the implementation of NDCs announced prior to COP26 and those associated with modelled mitigation pathways that limit warming to 1.5°C (>50%) with no or limited overshoot or limit warming to 2°C (>67%) assuming immediate action (high confidence). This would make it likely that warming will exceed 1.5°C during the 21st century (high confidence). Global modelled mitigation pathways that limit warming to 1.5°C (>50%) with no or limited overshoot or limit warming to 2°C (>67%) assuming immediate action imply deep global GHG emissions reductions this decade (high confidence) (see SPM Box 1, Table 1, B.6)27. Modelled pathways that are consistent with NDCs announced prior to COP26 until 2030 and assume no increase in ambition thereafter have higher emissions, leading to a median global warming of 2.8 [2.1 to 3.4] °C by 2100 (medium confidence). Many countries have signalled an intention to achieve net zero GHG or net zero CO, by around mid-century but pledges differ across countries in terms of scope and specificity, and limited policies are to date in place to deliver on them. (2.3.1, Table 2.2, Figure 2.5, Table 3.1, 4.1}
September 14, 2025 at 8:27 PM
bobkopp.net
A very unusual @altmetric.com profile on this paper, by the way -- the disproportionate pick up on X probably hints at something www.altmetric.com/details/1810...
Altmetric score for Voortman & de Vos (2025): 140 total, with 10 news mentions, 3 blog mentions, 68 X mentions, and 2 Bluesky mentions
September 12, 2025 at 1:01 PM
bobkopp.net
Thanks to my colleagues for pulling this together in response to such egregiously faulty science and faulty statistics! 4/fin
Signatories: Robert Kopp Distinguished Professor of Earth & Planetary Sciences Rutgers University John Church Emeritus Professor, Climate Change Research Centre University of New South Wales Soenke Dangendorf David and Jane Flowerree Associate Professor Department of River-Coastal Science and Engineering Tulane University Baylor Fox-Kemper Professor of Earth, Environmental, and Planetary Sciences Brown University Dewi Le Bars Sea level researcher KNMI Royal Netherlands Meteorological Institute Gonéri Le Cozannet Researcher - Coastal Risks and Climate Change BRGM - French Geological Survey Robert Nicholls Professor of Climate Adaptation and Professor of Coastal Engineering University of East Anglia and University of Southampton Michael Oppenheimer Albert G. Milbank Professor of Geosciences and International Affairs Princeton University Christopher Piecuch Assistant Scientist, Physical Oceanography Department Woods Hole Oceanographic Institution. Riccardo Riva Associate Professor, Department of Geoscience and Remote Sensing Delft University of Technology, The Netherlands Aimée Slangen Research Leader NIOZ Royal Netherlands Institute for Sea Research Vivek Srikrishnan Assistant Professor of Biological & Environmental Engineering Cornell University Philip Thompson Associate Professor, Department of Oceanography University of Hawaiʻi at Mānoa Richard S.J Tol Professor of Economics University of Sussex and Vrije Universiteit Amsterdam Roderik van de Wal Professor of sea level rise and coastal impacts Utrecht University and KNMI Continuation of signatories First page of references Second page of references
September 9, 2025 at 2:13 AM
bobkopp.net
We then detail the statistical failings of the paper, the most serious of which include using a statistically invalid test as the crux on which they test for acceleration, and artificially inflating the threshold for this test from 95% to >99.9% probability of acceleration required to not reject. 3/
The authors employ a standard F-test to test whether the model with acceleration better describes a tide gauge record than the no-acceleration model. Because a model with an unknown change point does not fit the assumptions of linear least squares regression, the assumptions underlying the standard F-test do not apply (e.g., Julious, 2001). The use of the standard F-test in this context is statistically invalid, compromising the paper’s inferences and conclusions. The authors also impose non-negativity constraints on second- and third-order polynomial terms based on external scientific expectations. While scientifically motivated, this potentially introduces bias into the parameter estimation and further invalidates classical inference procedures, including the F-test. The authors correct for autocorrelation (although it is unclear how; they cite Bence (1995), who compare various methods) but do not test for non-stationarity of their residuals: A unit root would render this correction invalid. The authors do not use the standard test by Zivot and Andrews (1992) for a trend break in a non-stationary series. Thus the authors do not conduct a statistically valid test for the presence of acceleration, and so their analysis is unable to support the conclusions of their paper. 4. Inflation of the detection threshold through inappropriate use of the Bonferroni multiple testing correction As previously raised by Le Bars et al. (2023) in response to Voortman (2023), this paper improperly applies the Bonferroni correction across spatially correlated tide stations. In a hypothetical 50 tide-gauge dataset with uncorrelated records, the Bonferroni correction would decrease the p-value required for a detection from p=0.05 to p=0.05/50=0.001. In other words, acceleration would be declared detected at a tide gauge only if it is numerically evaluated as being 99.9% probable rather than 95% probable. That is quite a threshold inflation! (Note also the authors say “depending on the dataset consisting of 50 to 200 data series” when discussing this correction. Elsewhere their two datasets appear to have either 39 or 204 gauges, so it is unclear where the 50 and 200 numbers came from.) However, the Bonferroni method is not correctly applied. It assumes independence among tests and is overly conservative under dependence, reducing statistical power and distorting inference. Given strong spatial correlation among tide gauges (arising, for example, through common dependencies on global-mean thermal expansion, changes in the amount of ice and water stored on land, and basin-scale dynamics), this method raises the bar for a detection to an unjustifiable level, even if controlling the family-wise error rate were a suitable goal. Notably, the importance of the spatial correlation structure points to the value of spatiotemporal statistical methods (e.g., Dangendorf et al., 2024; Wang et al., 2024) as opposed to one-tide-gauge-at-a-time methods in testing hypotheses about sea-level acceleration. Also notably, the way the authors frame their paper and the way they are publicly communicating about it suggests they are arguably testing only one hypothesis: “sea level acceleration driven by global phenomena.” This further reinforces the importance of accounting for spatial correlations within the context of testing this single global hypothesis. In any case, given the context of the paper, it is unclear why minimizing the number of false positives at the cost of increasing false negatives is the right goal. The most transparent approach would be to apply no correction. If the tide gauges were uncorrelated, then one would expect 5% of them to have false positives at the 95% probability level. Deviations from this frequency could then be a point of discussion. These methodological issues are not minor technicalities; they directly undermine the paper’s central claims regarding acceleration in sea level rise. The authors’ statistical mistakes are likely to lead to a high false negative rate, and may explain a substantial part of the difference between the results of this paper and those of Wang et al. (2021, 2025). The failure of the paper to explore the reason for its differences with Wang et al. (2021, 2025) helps mask these errors. At remarkably fast speed, this paper is being heralded in certain parts of the media as disproving sea-level acceleration (e.g., New York Post headline on 4 September 2025: “Blockbuster sea level study may turn climate change orthodoxy on its head”, with subsequent 5 September editorial: “Rising-sea hysteria debunked — but the ‘climate change’ cult won’t care”). Given the numerous errors identified, this paper is no blockbuster study and can do nothing of the sort. However, its unsupported conclusions, interpreted outside of the context of its methodological errors, can create confusion about an issue of great public concern. We respectfully urge the editorial board to initiate a formal review of the paper’s statistical methodology. If the concerns outlined above are substantiated, the paper should be retracted to maintain the integrity of the journal. Thank you for your attention to this matter. Sincerely, Robert Kopp Distinguished Professor of Earth & Planetary Sciences Rutgers University John Church Emeritus Professor, Climate Change Research Centre University of New South Wales Soenke Dangendorf David and Jane Flowerree Associate Professor Department of River-Coastal Science and Engineering Tulane University
September 9, 2025 at 2:13 AM
bobkopp.net
So I & 14 colleagues (incl. @richardtol.bsky.social @vsrikrish.bsky.social @goneri76.bsky.social) wrote an expression of concern, laying out the problems with the paper and calling for its retraction. We first provide scientific context, and the paper's failure to engage with related studies. 2/
We are writing to raise serious concerns regarding Voortman, H. G., & De Vos, R. (2025). A Global Perspective on Local Sea Level Changes. Journal of Marine Science and Engineering. Voortman & de Vos (2025) attempt to test for acceleration in relative sea-level (RSL) records from 243 tide gauges. They claim to have identified statistically significant acceleration in only 12 of the 243 gauges and conclude that “This pattern is inconsistent with sea level acceleration driven by global phenomena.” This result, which stands in contrast both to other studies conducting similar tide-gauge analyses (e.g., Wang et al., 2021, 2025) and the broader body of knowledge regarding global-mean sea level change and its drivers (e.g., Fox-Kemper et al., 2021), demands scrutiny. Yet the reviewers failed to recognise the context of this manuscript and the breadth of the existing literature, resulting in at best a superficial review. Upon close examination, we believe the paper contains fatal methodological flaws that compromise the validity of its conclusions and merit retraction. Scientific Context Comparison to other studies of RSL acceleration at tide-gauges and IPCC projections: The conclusions of this study are dramatically different from those of Wang et al. (2021, 2025). Wang et al. (2025) concluded significant acceleration was present at 68% of tide gauges, while this paper identifies it at only 4%. It is surprising that the reviewers did not insist that the authors provide a discussion to explain how they came to such different results, rather than ignoring the substance of Wang et al. (2021) and the existence of Wang et al. (2025). (We describe some likely statistical reasons below.) Notably, the authors state that “a comparison [of Intergovernmental Panel on Climate Change (IPCC) projections] to locally observed sea level appears to date to have been undertaken on a very limited scale”, and have made similar claims to the media, even though they cite without discussion Wang et al. (2021), which focuses precisely on a global-scale comparison of local observations and IPCC projections from 2013 and 2019 reports. They do not cite Wang et al. (2025), which extends this analysis using the same 2021 IPCC projections that are employed in this paper. Voortman and de Vos (2025) find that the 2021 IPCC median projected rates of relative sea level rise over 2010-2030 are higher than their preferred estimates of 2020 tide-gauge long-term rates by an average of about 2 mm/yr. Because the authors’ statistical tests reject the presence of acceleration at the large majority of tide gauges, they are effectively comparing IPCC projected rates — which account for the factors driving global-mean sea-level acceleration — with average rates over the length of the tide gauge record. By contrast, Wang et al. (2025) find that tide-gauge trends and accelerations extrapolated to 2050 agree with IPCC likely range projections at 96% of locations. Similarly, the Voortman and de Vos (2025) results are also in disagreement with regional studies where model-observation comparisons have been undertaken. For instance, Dangendorf et al (2023) report tide-gauge based rates of sterodynamic sea-level rise that are in 2020 at least twice as large as those from climate models along the U.S. Southeast and Gulf coasts, while Voortman and de Vos (2025) report IPCC rates that exceed tide gauge observations at the same locations (their Figure 9). The only exception in their model is Pensacola, which appears as an outlier in their assessment. Given the coherence of observed changes along this coast (Dangendorf et al., 2023), this also raises questions why their statistical model yields so different results for neighboring sites. Voortman and de Vos (2025) do not discuss any of these discrepancies. Comparison to other constraints on sea-level acceleration driven by global phenomena: In as much as the authors’ goal is to test for “sea le The authors also engage minimally with research on the global-mean sea level budget since 1900 (c.f. Fox-Kemper et al., 2021), providing a brief dismissal of one paper: “The current state of affairs is best illustrated by the paper by Frederikse et al. [(2020)] that claims to have “closed the sea level budget”. Unfortunately, a detailed inspection revealed a number of weaknesses and even errors in Frederikse’s work.” This statement is backed only by a reference to the first author’s own work (Voortman, 2023), which does not substantively engage with Frederikse et al. (2020). Regarding Frederikse et al. (2020), Voortman (2023) simply writes about his feelings: “Albeit Frederikse’s paper is an impressive scientific achievement, I feel that this conclusion is too rash. Estimating that the number of unknowns in this problem is larger than the number of equations available, I tentatively suggest that the reconstruction by Frederikse et al. is one of many plausible reconstructions that close the sea level budget.” No more “detailed inspection” of Frederikse et al. (2020) is provided in this paper or the cited source. The authors also highlight Munk (2002)’s enigma, which was an apparent inconsistency between estimates of 20th century global-mean sea-level rise compared with constraints on changes in Earth rotation and the sea level budget, but miss subsequent work resolving it. This inconsistency was subsequently resolved by newer estimates of 20th century global-mean sea level rise, improved modeling of glacial isostatic adjustment and its effect on Earth rotation, and improved interpretation of the constraints on Earth rotation from eclipse records (e.g., Dangendorf et al., 2017; Frederikse et al., 2020; Hay et al., 2015, 2016; Mitrovica et al., 2015). While the paper’s inadequate engagement with current scientific literature raises serious concerns, its statistical errors and biases render it internally invalid. Since these errors and biases fundamentally shape the paper, they cannot be corrected without complete rewriting and therefore necessitate the paper’s retraction. An examination of the reviewer comments reveal there has been no serious statistical evaluation of the methods or their appropriateness, or of the results. 1. Issues with model specification The paper hinges on the comparison of two models: one linear in time (no acceleration), and one with a changepoint after which quadratic and cubic terms (and thus acceleration) are present. The authors’ model with acceleration has 3 more fitted parameters than their no-acceleration model (a quadratic coefficient, a cubic coefficient, and a change point). We are not aware of any previous assessment of the third derivative of sea level (the cubic term), and a minimally larger model with acceleration requires only one fitted parameter (a quadratic coefficient; the change point could be fixed at, say, 1970 for consistency with other literature). Because hypothesis testing metrics penalize the loss of degrees of freedom, using three additional parameters raises the threshold of detection for acceleration more than is necessary and increases the risk of false negatives. It is also well known that interannual/decadal variability is one of the largest signals that makes detecting an acceleration in single time series challenging. However, there is no attempt to limit the impact of this variability on the detection of any acceleration (contrast Wang et al., 2021, 2025), and this critical issue is not even discussed. 2. Issues with model estimation 3. Invalid use of the F-test for hypothesis testing
September 9, 2025 at 2:13 AM
bobkopp.net
I see no documented evidence of peer review, and am not impressed with the so-called peer review if there was any, given the state of the report. There are requirement for peer review that include a peer review report, of which there is no indication. www.federalregister.gov/documents/20...
Excerpt from OMB Final Information Quality Bulletin for Peer Review: Section III(6) requires that agencies instruct reviewers to prepare a peer review report that describes the nature and scope of their review and their findings and conclusions. The report shall disclose the name of each peer reviewer and a brief description of his or her organizational affiliation, credentials and relevant experiences. The peer review report should either summarize the views of the group as a whole (including any dissenting views) or include a verbatim copy of the comments of the individual reviewers (with or without attribution of specific views to specific names). The agency shall also prepare a written response to the peer review report, indicating whether the agency agrees with the reviewers and what actions the agency has taken or plans to take to address the points made by reviewers. The agency is required to disseminate the peer review report and the agency's response to the report on the agency's Web site, including all the materials related to the peer review such as the charge statement, peer review report, and agency response to the review. If the scientific information is used to support a final rule then, where practicable, the peer review report shall be made available to the public with enough time for the public to consider the implications of the peer review report for the rule being considered.
September 8, 2025 at 1:09 AM
bobkopp.net
It's not how I would do the analysis, but I think it accurately corresponds to what Voortman and de Vos do?
A test that explicitly judges the difference in variance of the residuals and penalizes for model complexity is the F-test for reduced models [86]. The F-test compares the full model to the reduced model by calculating and judging the F-statistic, given by: 𝐹=(𝑆𝑆𝑟𝑒𝑑𝑢𝑐𝑒𝑑−𝑆𝑆𝑓𝑢𝑙𝑙𝑘−𝑚)÷(𝑆𝑆𝑓𝑢𝑙𝑙𝑁−𝑘) where SSreduced and SSfull are the sum of squared residuals of the reduced and the full model, respectively, k is the number of parameters in the full regression model, and m is the number of parameters in the reduced regression model [86]. The null hypothesis in the F-test is the assumption that the reduced model is an adequate description of the data. The F-test is executed by comparing the calculated F-statistic to a value obtained from an F-distribution at a specified confidence level.
September 5, 2025 at 7:45 PM
bobkopp.net
Funny... published in Feb. 2021: www.nature.com/articles/s41...
“It is crazy that it had not been done. I started doing this research in 2021 by doing the literature review. ‘Who has done the comparison of the projections with the observations?’ And there were none,” he told Shellenberger.
September 5, 2025 at 7:36 PM