pith. sign in

arxiv: 2606.18806 · v1 · pith:WVB5ICTXnew · submitted 2026-06-17 · 📊 stat.AP

Spatial emergence of acceleration in global warming

Tanja Korsten Bugajski , Nicolai Peder Bulow Pedersen , J. Eduardo Vera-Valdes This is my paper

Pith reviewed 2026-06-26 18:59 UTC · model grok-4.3

classification 📊 stat.AP
keywords global warmingaccelerationBayesian hierarchical modelspatio-temporalhigh latitudesdetectionspatial heterogeneitytemperature trends
0
0 comments X

The pith

Detectable acceleration in global warming emerges first in selected high-latitude regions and spreads unevenly.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper applies a Bayesian hierarchical spatio-temporal model to temperature records truncated at successive end years to determine when local warming rates begin to increase. It reports that the share of grid cells with strong posterior evidence of positive acceleration grows from 13.6 percent in records ending 1990 to 39.7 percent in records ending 2026, with the earliest signals concentrated in high latitudes. This pattern implies that averaging temperatures across wide regions mixes places where acceleration has already appeared with places where it has not, postponing detection in the aggregate. The approach supplies a probabilistic way to map where and when the warming trend itself is intensifying.

Core claim

The authors estimate local warming trajectories and acceleration using a Bayesian hierarchical spatio-temporal model with structured spatial dependence, then apply the model to data ending in successive years from 1990 onward. They find the proportion of retained grid cells exceeding 90 percent posterior probability of positive acceleration rises from 13.6 percent to 39.7 percent, while the share above 50 percent rises from 46.4 percent to 70.3 percent, with early high-confidence signals concentrated in high-latitude regions. These results demonstrate that spatial aggregation delays detection by averaging cells where acceleration has emerged with cells where it remains weak or uncertain.

What carries the argument

Bayesian hierarchical spatio-temporal model with structured spatial dependence, which produces local trajectory estimates and posterior probabilities of acceleration from progressively longer temperature records.

If this is right

  • Spatial averaging of temperature data postpones statistical detection of acceleration until later dates.
  • High-latitude grid cells provide the earliest locations where acceleration exceeds high posterior thresholds.
  • The fraction of cells showing detectable acceleration continues to increase as the observational record lengthens.
  • Probabilistic thresholds applied to posterior distributions can map the spatial pattern of emerging acceleration.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The truncation method could be used on other climate fields such as precipitation or sea-level records to locate early acceleration.
  • Targeted observational networks in high-latitude zones might yield earlier confirmation of intensifying trends than global averages.
  • If applied to output from climate models, the same framework could test whether simulated acceleration emerges on the same spatial schedule as observations.

Load-bearing premise

The model isolates true acceleration from internal variability and spatial heterogeneity in the temperature data.

What would settle it

Applying the same model to synthetic temperature fields generated from a process with constant warming rate plus realistic variability and finding rising proportions of cells above the 90 percent threshold would falsify the claim that the posteriors detect acceleration.

read the original abstract

Whether global warming is accelerating remains contested because internal variability and spatial heterogeneity can obscure changes in warming rates. Here we use a Bayesian hierarchical spatio-temporal model with structured spatial dependence to estimate local warming trajectories and acceleration, and apply the model to progressively truncated observations to infer when acceleration becomes detectable. We find that detectable acceleration emerges unevenly across the climate system, with the earliest high-confidence signals concentrated in selected high-latitude regions. Across retained grid cells, the proportion exceeding a 90% posterior probability of positive acceleration increases from 13.6% for 1970-1990 to 39.7% for 1970-2026, while the proportion exceeding a 50% threshold increases from 46.4% to 70.3%. These results show that spatial aggregation can delay detection by averaging regions where acceleration has already emerged with regions where it remains weak or uncertain. The framework provides a probabilistic diagnostic for identifying where warming is intensifying and when acceleration becomes statistically detectable.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript applies a Bayesian hierarchical spatio-temporal model with structured spatial dependence to observational temperature records, using progressively truncated time series to estimate when local acceleration in warming becomes detectable. It reports that the proportion of retained grid cells with >90% posterior probability of positive acceleration rises from 13.6% (1970-1990) to 39.7% (1970-2026), and from 46.4% to 70.3% at the 50% threshold, with earliest high-confidence signals concentrated in selected high-latitude regions; spatial aggregation is argued to delay detection.

Significance. If the model posteriors correctly isolate acceleration from internal variability, the work supplies a spatially resolved probabilistic diagnostic for emergence timing that could complement global-mean analyses and highlight the masking effect of spatial averaging. The truncated-observation design is a clear methodological strength for tracking detection as data accumulate.

major comments (2)
  1. [Methods] Methods section: the claim that posterior probabilities isolate acceleration from internal variability and spatial heterogeneity is load-bearing for all reported proportions, yet no validation against synthetic data with known acceleration signals or comparison to alternative covariance structures is described, leaving the separation untested.
  2. [Results] Results, paragraph reporting the 13.6%/39.7% and 46.4%/70.3% figures: these quantities are presented without accompanying posterior uncertainty on the proportions themselves or sensitivity checks to the spatial covariance hyperparameters (the only free parameters listed), which directly affects the central claim of increasing detectability.
minor comments (2)
  1. [Abstract] Abstract does not name the temperature dataset or grid resolution, which is needed for immediate assessment of the retained-grid-cell proportions.
  2. [Methods] Notation for the acceleration parameter and the exact form of the structured spatial dependence should be stated explicitly with an equation number in the model description.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thoughtful review and constructive feedback. We address each major comment below, indicating where revisions will be made to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Methods] Methods section: the claim that posterior probabilities isolate acceleration from internal variability and spatial heterogeneity is load-bearing for all reported proportions, yet no validation against synthetic data with known acceleration signals or comparison to alternative covariance structures is described, leaving the separation untested.

    Authors: We agree that explicit validation of the model's ability to isolate acceleration signals would strengthen the claims. The current manuscript relies on the hierarchical structure and structured spatial dependence to separate signals, but does not include synthetic data experiments. In the revised version, we will add a simulation study generating data with known acceleration under the model (and under misspecification) to assess recovery of posterior probabilities of positive acceleration. We will also compare results under alternative covariance structures (e.g., Matérn with different smoothness or independent spatial effects) to demonstrate robustness of the reported proportions. revision: yes

  2. Referee: [Results] Results, paragraph reporting the 13.6%/39.7% and 46.4%/70.3% figures: these quantities are presented without accompanying posterior uncertainty on the proportions themselves or sensitivity checks to the spatial covariance hyperparameters (the only free parameters listed), which directly affects the central claim of increasing detectability.

    Authors: The proportions are computed from grid-cell posterior probabilities, but we concur that uncertainty quantification and hyperparameter sensitivity are needed for the aggregate statistics. In revision, we will derive credible intervals for the reported proportions by propagating posterior uncertainty across grid cells. We will also conduct sensitivity analyses by varying the spatial covariance hyperparameters over plausible ranges and report the resulting variation in the 13.6%/39.7% and 46.4%/70.3% figures to confirm that the increasing detectability trend is not sensitive to these choices. revision: yes

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper applies a Bayesian hierarchical spatio-temporal model to observational temperature data, then reports posterior probabilities and derived proportions of grid cells exceeding thresholds for acceleration. These outputs are statistical results from fitting the model to progressively truncated records rather than quantities defined by construction from the fitted parameters or inputs. No self-citations, ansatzes smuggled via prior work, or fitted inputs renamed as predictions appear in the derivation chain. The central claims remain independent of the inputs once the model is specified.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Review is based on the abstract only; the full model specification, priors, and data-processing steps are not available, preventing an exhaustive ledger. The approach relies on standard Bayesian assumptions for spatio-temporal processes rather than new entities.

free parameters (1)
  • spatial covariance hyperparameters
    The structured spatial dependence component of the hierarchical model requires parameters estimated from the temperature data.
axioms (1)
  • domain assumption Temperature fields follow a spatio-temporal process with structured spatial dependence that can be captured by the chosen hierarchical model
    This assumption underpins the estimation of local trajectories and posterior probabilities of acceleration.

pith-pipeline@v0.9.1-grok · 5704 in / 1323 out tokens · 37218 ms · 2026-06-26T18:59:30.408609+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

23 extracted references

  1. [1]

    Climate change 2021: The physical science basis

    IPCC. Climate change 2021: The physical science basis. Contribution of working group i to the sixth assessment report of the intergovernmental panel on climate change. (2021)

    2021

  2. [2]

    P.et al.An updated assessment of near-surface temperature change from 1850: The HadCRUT5 dataset.Journal of Geophysical Research: Atmo- spheres126, e2019JD032361 (2021)

    Morice, C. P.et al.An updated assessment of near-surface temperature change from 1850: The HadCRUT5 dataset.Journal of Geophysical Research: Atmo- spheres126, e2019JD032361 (2021). 13

    2021

  3. [3]

    H.et al.Steady global surface warming from 1973 to 2022 but increased warming rate after 1990.Communications Earth & Environment4, 400 (2023)

    Samset, B. H.et al.Steady global surface warming from 1973 to 2022 but increased warming rate after 1990.Communications Earth & Environment4, 400 (2023)

    1973

  4. [4]

    von, Sallée, J

    Miniere, A., Schuckmann, K. von, Sallée, J. B. & Vogt, L. Robust acceleration of earth system heating observed over the past six decades.Scientific Reports13, 22975 (2023)

    2023

  5. [5]

    & Shi, X

    Beaulieu, C., Gallagher, C., Killick, R., Lund, R. & Shi, X. A recent surge in global warming is not detectable yet.Communications Earth & Environment5, 576 (2024)

    2024

  6. [6]

    Richardson, M. T. Prospects for detecting accelerated global warming.Geophysical Research Letters49, e2021GL095782 (2022)

    2022

  7. [7]

    S.et al.A fluctuation in surface temperature in historical context

    Risbey, J. S.et al.A fluctuation in surface temperature in historical context. Environmental Research Letters13, 123008 (2018)

    2018

  8. [8]

    & Rahmstorf, S

    Foster, G. & Rahmstorf, S. Global Warming Has Accelerated Significantly.Geo- physical Research Letters53, e2025GL118804 (2026)

    2026

  9. [9]

    & Sutton, R

    Hawkins, E. & Sutton, R. Time of emergence of climate signals.Geophysical Research Letters39, L01702 (2012)

    2012

  10. [10]

    & Parnell, A

    Cahill, N., Rahmstorf, S. & Parnell, A. C. Change points of global temperature. Environmental Research Letters10, 084002 (2015)

    2015

  11. [11]

    T., Dong, B

    Sutton, R. T., Dong, B. & Gregory, J. M. Land/sea warming ratio in response to climate change: IPCC AR4 model results and comparison with observations. Geophysical Research Letters34, L02701 (2007)

    2007

  12. [12]

    vonet al.Heat stored in the earth system 1960–2020: Where does the energy go?Earth System Science Data15, 1675–1709 (2023)

    Schuckmann, K. vonet al.Heat stored in the earth system 1960–2020: Where does the energy go?Earth System Science Data15, 1675–1709 (2023)

    1960

  13. [14]

    Vaccaro, A.et al.Climate field completion via markov random fields: Application to the HadCRUT4.6 temperature dataset.Journal of Climate34, 4169–4188 (2021)

    2021

  14. [16]

    Screen, J. A. & Simmonds, I. The central role of diminishing sea ice in recent arctic temperature amplification.Nature464, 1334–1337 (2010)

    2010

  15. [17]

    Eyring, V.et al.Taking climate model evaluation to the next level.Nature Climate Change9, 102–110 (2019)

    2019

  16. [18]

    & Lindström, J

    Lindgren, F., Rue, H. & Lindström, J. An explicit link between gaussian fields and gaussian markov random fields: The stochastic partial differential equation ap- proach.Journal of the Royal Statistical Society: Series B (Statistical Methodology) 73, 423–498 (2011). 14

    2011

  17. [19]

    HadCRUT5 observational dataset

    Met Office Hadley Centre. HadCRUT5 observational dataset. (2022)

    2022

  18. [20]

    Bayesian Data Analysis

    Gelman, A.et al. Bayesian Data Analysis. (CRC Press, 2013)

    2013

  19. [22]

    & Rubin, D

    Gelman, A. & Rubin, D. B. Inference from iterative simulation using multiple sequences.Statistical Science7, 457–472 (1992). 15 Supplementary Information for: Spatial emergence of detectable acceleration in global warming Tanja Korsten Bugajski1, Nicolai Peder Bülow Pedersen 1, and J. Eduardo Vera-Valdés1,2 1Department of Mathematical Sciences, Aalborg Un...

    1992

  20. [23]

    & Smith, R

    Katzfuss, M., Hammerling, D. & Smith, R. L. A bayesian hierarchical model for climate change detection and attribution.Geophysical Research Letters44, 5720–5728 (2017)

    2017

  21. [24]

    T.Random Fields for Spatial Data Modeling: A Primer for Scientists and Engineers

    Hristopulos, D. T.Random Fields for Spatial Data Modeling: A Primer for Scientists and Engineers. (Springer, 2020)

    2020

  22. [25]

    & Held, L.Gaussian Markov Random Fields: Theory and Applications

    Rue, H. & Held, L.Gaussian Markov Random Fields: Theory and Applications. (Chapman; Hall/CRC, 2005)

    2005

  23. [26]

    & Oates, C

    Porcu, E., Bevilacqua, M., Schaback, R. & Oates, C. J. The matérn model: A journey through statistics, numerical analysis and machine learning. https: //arxiv.org/abs/2303.02759 (2023). 9

    arXiv 2023