Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Finetuning AI Foundation Models to Develop Subgrid-Scale Parameterizations: A Case Study on Atmospheric Gravity Waves

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Fine-tuning a pretrained weather foundation model produces a gravity-wave parameterization that beats an Attention U-Net baseline across the atmosphere, including layers the model never saw in pretraining.

desk verdict Fine-tuning Prithvi WxC beats a U-Net for gravity wave flux on one month, with code and data; the 'throughout the atmosphere' claim needs more validation and a fix to an internal inconsistency. read the letter →

arxiv 2509.03816 v1 pith:ZFXZO6DW submitted 2025-09-04 physics.ao-ph cs.LG

Aman Gupta , Aditi Sheshadri , Sujit Roy , Johannes Schmude , Vishal Gaur , Wei Ji Leong , Manil Maskey , Rahul Ramachandran This is my paper
classification physics.ao-phcs.LG
keywords gravitywavessubgrid-scaleparameterizationfoundationmodelsfine-tuningmachinelearningemulationmomentumfluxERA5climatemodeling
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that a large, pre-trained AI weather model can be repurposed, cheaply and quickly, into a subgrid-scale parameterization for atmospheric gravity waves. The authors take Prithvi WxC—a 2.3-billion-parameter encoder-decoder trained on 40 years of MERRA-2—freeze its core, add shallow learnable layers, and fine-tune it on four years of gravity-wave momentum fluxes derived from ERA5. The fine-tuned model predicts instantaneous, intermittent wave-flux evolution and monthly distributions more accurately than an Attention U-Net trained on the same data, with a daily-global-distribution Hellinger distance of 0.06 versus 0.11 and a Drake Passage correlation of 0.99 versus 0.84. Improvement is largest in the upper stratosphere, a region absent from Prithvi's pre-training. If the approach holds, foundation models become reusable building blocks for observations-driven parameterizations of other unresolved climate processes, not just gravity waves.

What carries the argument

The load-bearing object is Prithvi WxC, a transformer encoder-decoder with 2.3 billion parameters pre-trained on 40 years of 3-hourly MERRA-2 with a masked-reconstruction objective. Fine-tuning freezes its encoder-decoder and adds four learnable convolutional blocks on each end, so the model is a compact reused parameterization rather than a retraining. Inputs are winds, temperature, and log pressure on 122 levels over a 64×128 grid; outputs are potential temperature and the momentum fluxes u′ω′ and v′ω′. Training labels come from ERA5 by Helmholtz decomposition, T21 filtering, and coarse-graining to ~280 km. Success is judged by Hellinger distance between predicted and ERA5 flux distributio

What would settle it

Hold out a different period—say all twelve months of a disjoint year—and recompute the Hellinger distances and hotspot correlations for both models; if the fine-tuned model's margin shrinks or reverses in some months or seasons, the 'throughout the atmosphere' claim does not generalize.

Watch

Extended reading notes

Core claim

The central claim: the latent atmospheric-evolution representation of a pretrained weather foundation model can be reused, through fine-tuning, as a subgrid-scale gravity-wave parameterization, and it beats a specialized network trained from scratch on the same limited data. Fine-tuning Prithvi WxC's frozen encoder-decoder on four years of ERA5-derived momentum fluxes yields higher Pearson correlations with ERA5 over six gravity-wave hotspots (0.99 vs 0.84 at Drake Passage), better stratospheric variability, and lower Hellinger distances—largest in the upper stratosphere, which was absent from pretraining. The paper also reports faster convergence and learned lateral propagation, and notes b

Load-bearing premise

The headline comparison rests on a single validation month, May 2015, with no error bars or demonstration that this month represents other seasons; it also treats ERA5's model-generated fluxes as the true target to learn.

Editorial extensions

If this is right

  • The fine-tuned parameterization can be coupled to a coarse-resolution climate model to supply gravity-wave tendencies that the resolved dynamics cannot produce, potentially correcting middle-atmosphere wind and temperature biases such as the cold-pole bias.
  • The same recipe—frozen pretrained encoder-decoder plus thin learnable layers—can be applied to other unresolved processes (clouds, convection, turbulence) whenever a suitable flux or tendency dataset exists, with far fewer training samples than training from scratch.
  • Because the fine-tuned model learns horizontal and lateral propagation from resolved fluxes, it captures a physical process that traditional single-column gravity-wave parameterizations omit.
  • Both ML models struggle to predict small, daily-sampled flux values near zero, so improving near-zero flux prediction is an open target for the next generation of emulators.
  • The nonlocal architecture is compatible with existing model-coupling tools, and the paper reports work underway to implement it online in a full atmospheric model.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Inference: If the one-month validation gap reflects true skill, foundation-model pretraining should transfer across vertical levels that were never in the pretraining data; a direct test is to ablate the pretrained weights (randomly initialized encoder) and compare convergence and skill.
  • Inference: The ranking between models could change if the training target changes—ERA5's fluxes are model-generated and miss waves shorter than ~150–200 km—so a kilometer-scale or observation-based flux dataset is a natural stress test of the claimed superiority.
  • Inference: Because the paper only validates May 2015, seasonal and interannual robustness is untested; extending validation to a full year or a disjoint year would clarify whether the fine-tuned model's advantage is concentrated in strong-wave months.
  • Inference: The fine-tuning recipe may let researchers build parameterizations for rare or observation-sparse processes by starting from a pretrained atmospheric representation rather than collecting large labeled training sets, at the cost of inheriting whatever biases the pretraining reanalysis carries.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes fine-tuning a pre-trained weather/climate foundation model (Prithvi WxC, 2.3B parameters, pre-trained on MERRA-2) to produce a subgrid-scale gravity wave momentum flux parameterization. The model takes coarse-grained ERA5 winds, temperature, and pressure on 122 vertical levels and outputs potential temperature and zonal/meridional momentum fluxes, using frozen Prithvi encoder/decoder blocks surrounded by new trainable convolutional blocks. The authors compare this fine-tuned model against an Attention U-Net baseline on one held-out month (May 2015), reporting lower Hellinger distances (0.062 vs 0.116 for daily-averaged global zonal flux) and higher instantaneous correlations at several hotspots, including Drake Passage (0.99 vs 0.84). The paper argues this demonstrates that foundation models can accelerate development of ML parameterizations for climate processes.

Significance. The approach is genuinely novel in the context of subgrid-scale parameterization: leveraging a large pre-trained atmospheric foundation model with a frozen encoder/decoder and only light trainable convolutional adaptors is a plausible route to data-efficient, nonlocal parameterizations. The authors provide open code and data pipelines, and the central comparison against a strong UNet baseline is appropriate. If the performance gains hold across seasons and years, the result would be valuable for the climate ML community and for JAMES readers. However, the current evidentiary basis is a single validation month, and the reported Hellinger metric is mislabeled as the distance rather than its square; both issues bear directly on the headline claims.

major comments (3)
  1. [Sec. 2.5 / Sec. 3] The central claim of superiority 'throughout the atmosphere' rests entirely on validation on May 2015. The manuscript states in Sec. 2.5 that only May 2015 was held out, with the other 47 months used for training. May 2015 is temporally adjacent to April and June 2015, so it is not independent of the training distribution. No error bars, no interannual/seasonal spread, and no justification that May 2015 is representative are provided. The headline numbers (Hellinger 0.062 vs 0.116; Drake Passage r=0.99 vs 0.84) could easily change on other months. Please validate on additional months/years or provide uncertainty estimates (e.g., bootstrap across days or multiple holdout months).
  2. [Eq. (6), Sec. 2.6] Equation (6) defines H(p,q) = 1 - integral sqrt(pq), which is the squared Hellinger distance. The actual Hellinger distance is the square root of this quantity. The reported values (0.062, 0.116, etc.) are therefore squared distances. This mislabeling affects the quantitative interpretation of the headline metric: the distances are roughly sqrt(0.062)=0.25 and sqrt(0.116)=0.34. While the relative ordering may persist, the manuscript's claim that these are 'Hellinger distances' in the standard sense is incorrect. Please correct the definition and all reported values, or explicitly call the quantity 'squared Hellinger affinity/distance'.
  3. [Sec. 3.2] There is an internal contradiction in the regional claims. The text first says 'in some regions, such as the lower stratosphere over Newfoundland, and the troposphere over the Southern Ocean, the Hellinger distances are slightly better for the attn unet model.' Two paragraphs later it says the fine-tuned model outperforms the baseline in 'practically all regions (the only exception being the lower stratosphere over Newfoundland).' These statements cannot both be true. Also, the abstract's 'superior performance throughout the atmosphere' is too strong given that the manuscript itself documents at least one (and possibly two) exceptions. Please reconcile the text and temper the global claim.
minor comments (5)
  1. [Throughout] The model name is inconsistently written: 'attn unet', 'attn uNet', 'Attn U-Net', and 'Attention U-Net' are all used. Please use one form consistently.
  2. [Fig. A3 caption] Typo: 'Yelllow' should be 'Yellow' and 'Southern Greeland' should be 'Southern Greenland'.
  3. [Sec. 2.2] The normalization for flux uses an exponent 1/3: u'ω' -> [(u'ω' - mean)/std]^{1/3}. The inverse transform and its effect on the loss are not discussed. Please clarify how predictions are back-transformed for evaluation and whether the cube-root transform affects the reported metrics.
  4. [Sec. 3.4 / Eq. 6] The method for estimating the probability densities used in the Hellinger distance is not specified (e.g., histogram bin width, kernel density estimator). Since the metric values depend on this choice, please provide details or a reference.
  5. [Fig. 5 caption] The caption says 'normalized flux u'ω'' for the meridional flux figure A5; the text says 'for the meridional flux is shown in Fig A5' but the figure caption and axis labels should be checked for consistency (v'ω' not u'ω').

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the fine-tuned FM is evaluated against held-out ERA5 data with a freshly trained baseline; self-citations are sources, not load-bearing assumptions.

full rationale

The central quantitative claims—Hellinger distances of 0.116 vs 0.062 for daily distributions and hotspot correlations (e.g., Drake Passage 0.84 vs 0.99)—are computed on May 2015 ERA5 data held out from the 47-month training set, with no indication that validation statistics were used to fit model parameters. The Attention U-Net baseline is retrained from scratch on the same data split, so the comparison is an independent empirical measurement rather than an imported number. The self-citations (Schmude et al. 2024 for Prithvi WxC; Gupta et al. 2024 for the baseline) provide the architecture and a benchmark, but the paper does not rely on their reported performance to establish superiority; it re-runs the baseline and measures both models against ERA5, and the code is publicly released. The admitted single-month validation (Sec. 2.5) and the internal inconsistency in Sec. 3.2 about which regions the baseline wins are correctness/generalization concerns, not circularity. Similarly, the note that ERA5 fluxes are model-generated is a physical-limitation caveat, not a circular step. No load-bearing step in the derivation chain equates a target with a fitted input, imports a uniqueness theorem, or smuggles an ansatz through self-citation. The derivation is self-contained against the held-out data.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central comparison depends on a set of training and evaluation choices: a single validation month, fixed hyperparameters, a custom normalization, and an architecture mapping that is not fully specified. No new physical entities are invoked.

free parameters (6)
  • Validation split: month of May 2015 = May 2015
    Only this month is held out for all reported metrics; the choice is not justified and could bias the comparison.
  • Training epochs = 100
    Both models trained for 100 epochs; the fine-tuned model reached baseline error at epoch 40, but the baseline may not be fully converged.
  • Baseline learning rate = 1e-4
    Fixed for the Attention U-Net; no learning rate schedule or hyperparameter tuning is reported.
  • Minibatch size = 4
    Identical for both models; affects training dynamics and convergence.
  • New convolutional block channel width C = 160
    Chosen for the four added convolutional blocks around the frozen encoder/decoder; no sensitivity analysis.
  • Flux normalization exponent = 1/3
    Applied to scale u'w' and v'w' before training; the choice affects loss weighting and, hence, the fit to small versus large fluxes.
assumptions (5)
  • domain assumption Helmholtz decomposition of ERA5 winds isolates gravity wave divergent flow (Eqs. 1-3)
    Follows Lindborg (2015); assumes the residual after T21 removal is gravity wave dominated, though it may include some convective contributions.
  • domain assumption ERA5 at 25 km resolves gravity waves with wavelengths longer than roughly 150-200 km
    Stated in Section 2.2; the paper also cautions that all ERA5 gravity waves are model-generated and may miss shorter waves.
  • ad hoc to paper The frozen Prithvi encoder/decoder can be adapted to a 122-level vertical grid via added convolutional blocks
    Section 2.4 describes the architecture but does not specify how the 122-level input is mapped to Prithvi's 14 vertical levels, yet this mapping is required for the method to work as claimed.
  • standard math Hellinger distance as defined in Eq. 6 is a valid distance measure
    The formula given is the squared Hellinger distance (1 - integral sqrt(pq)), not the Hellinger distance sqrt(1 - integral sqrt(pq)), so reported values are systematically mislabeled.
  • domain assumption MSE loss on normalized fluxes is an appropriate training objective
    Both models are trained with MSE; no weighting for rare strong events or physical conservation constraints is applied.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Finetuning AI Foundation Models to Develop Subgrid-Scale Parameterizations: A Case Study on Atmospheric Gravity Waves." pith.science (2026). https://pith.science/paper/ZFXZO6DW

@misc{pith2026250903816,
  author       = {Pith},
  title        = {Pith review of: Finetuning AI Foundation Models to Develop Subgrid-Scale Parameterizations: A Case Study on Atmospheric Gravity Waves},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZFXZO6DW}},
  note         = {Machine review of arXiv:2509.03816}
}
read the original abstract

Global climate models parameterize a range of atmospheric-oceanic processes like gravity waves, clouds, moist convection, and turbulence that cannot be sufficiently resolved. These subgrid-scale closures for unresolved processes are a leading source of model uncertainty. Here, we present a new approach to developing machine learning parameterizations of small-scale climate processes by fine-tuning a pre-trained AI foundation model (FM). FMs are largely unexplored in climate research. A pre-trained encoder-decoder from a 2.3 billion parameter FM (NASA and IBM Research's Prithvi WxC) -- which contains a latent probabilistic representation of atmospheric evolution -- is fine-tuned (or reused) to create a deep learning parameterization for atmospheric gravity waves (GWs). The parameterization captures GW effects for a coarse-resolution climate model by learning the fluxes from an atmospheric reanalysis with 10 times finer resolution. A comparison of monthly averages and instantaneous evolution with a machine learning model baseline (an Attention U-Net) reveals superior predictive performance of the FM parameterization throughout the atmosphere, even in regions excluded from pre-training. This performance boost is quantified using the Hellinger distance, which is 0.11 for the baseline and 0.06 for the fine-tuned model. Our findings emphasize the versatility and reusability of FMs, which could be used to accomplish a range of atmosphere- and climate-related applications, leading the way for the creation of observations-driven and physically accurate parameterizations for more earth-system processes.

Figures

Figures reproduced from arXiv: 2509.03816 by the authors.

Figure 1
Figure 1. Predictions from the fine-tuned gravity wave parameterization. The left plot shows the temperature structure of GWs over the Drake Passage, as seen in ERA5 reanalysis (Hersbach et al., 2020). Temperature perturbations T ′ were computed by removing the large scales, here defined as the first 21 total wavenumbers. The middle and right plots show the true and pre￾dicted momentum flux carried by the waves. 30 km is an a… view at source ↗
Figure 2
Figure 2. (a) Pre-training model architecture for Prithvi WxC. The encoder and decoder blocks from Prithvi WxC are frozen and used for fine-tuning. 10 atmospheric variables on 14 ver￾tical levels, 20 surface variables, 4 static variables, and 8 climatology variables for 366 days in a year, all on a 360 (lat) x 576 (lon) grid, for the input.(b) Model Architecture for attn unet using 3 input variables, each on 122 vertical leve… view at source ↗
Figure 3
Figure 3. Instantaneous (non-dimensional/normalized) fluxes for May 2015 resolved in ERA5 reanalysis (black), predicted fluxes from attn unet (blue), and from the fine-tuned parameter￾ization (red), over six well-known GW hotspots. The numbers show the respective Pearson correlation coefficients with respect to ERA5. The fluxes in the winter hemisphere are shown at 30 km, whereas the fluxes in the summer hemisphere are shown … view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: May 2015 averaged GW momentum flux distributions divided according to hotspots and vertical regions in the atmosphere. The figure shows non-dimensional fluxes as predicted by the models for ease of comparisons. The y-axis is the distribution density. The fluxes are ave…
Figure 5
Figure 5. Figure 5: May 2015 mean ERA5 and predicted vertical profiles of the normalized (unitless) zonal flux, Fx = u ′ω ′ over five hotspots. The exact boundaries of the hotspots are shown in [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]
Figure 6
Figure 6. Figure 6: Histogram of the (a) May 2015 averaged and (b) daily averaged GW flux u ′ω ′ . Gray shading shows the underlying ERA5 distribution, orange is the attn unet prediction, and green is the fine-tuning prediction. Numbers indicate the Hellinger distance for the correspondin…

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Interpretable Neural Networks to Predict Momentum Fluxes of Orographic Gravity Waves

    physics.ao-ph 2026-05 conditional novelty 5.0 of 10

    Neural networks predict orographic gravity wave momentum fluxes from coarse state variables with offline R² of 0.56-0.72, learn physically meaningful relationships via SHAP, and are compared to the Lott-Miller paramet...

Reference graph

Works this paper leans on

71 extracted references · 38 canonical work pages · cited by 1 Pith paper

  1. [1]

    , Alexander, M J

    Achatz.etal2024 APACrefauthors Achatz, U. , Alexander, M J. , Becker, E. , Chun, H Y. , D \"o rnbrack, A. , Holt, L. Wright, C J. APACrefauthors \ 2024 . Atmospheric Gravity Waves : Processes and Parameterization Atmospheric Gravity Waves : Processes and Parameterization . JAS . APACrefDOI doi:10.1175/JAS-D-23-0210.1 APACrefDOI

    doi:10.1175/jas-d-23-0210.1 2024

  2. [2]

    \ Dunkerton, T J

    Alexander.Dunkerton1999 APACrefauthors Alexander, M J. \ Dunkerton, T J. APACrefauthors \ 1999 . A Spectral Parameterization of Mean-Flow Forcing Due to Breaking Gravity Waves A Spectral Parameterization of Mean-Flow Forcing due to Breaking Gravity Waves . J. Atmos. Sci. 56 24 4167--4182 . APACrefDOI doi:10.1175/1520-0469(1999)056<4167:ASPOMF>2.0.CO;2 APACrefDOI

    doi:10.1175/1520-0469(1999)056 1999

  3. [3]

    , Richter, J H

    Alexander.etal2006 APACrefauthors Alexander, M J. , Richter, J H. \ Sutherland, B R. APACrefauthors \ 2006 11 . Generation and Trapping of Gravity Waves from Convection with Comparison to Parameterization Generation and Trapping of Gravity Waves from Convection with Comparison to Parameterization . APACrefDOI doi:10.1175/JAS3792.1 APACrefDOI

    work page doi:10.1175/jas3792.1 2006

  4. [4]

    , Elafrou, A

    Atkinson.etal2025 APACrefauthors Atkinson, J. , Elafrou, A. , Kasoar, E. , Wallwork, J G. , Meltzer, T. , Clifford, S. Edsall, C. APACrefauthors \ 2025 03 . FTorch : A Library for Coupling PyTorch Models to Fortran FTorch : A library for coupling PyTorch models to Fortran . Journal of Open Source Software 10 107 7602 . APACrefDOI doi:10.21105/joss.07602 A...

    doi:10.21105/joss.07602 2025

  5. [5]

    APACrefauthors \ 2012 06

    Becker2012 APACrefauthors Becker, E. APACrefauthors \ 2012 06 . Dynamical Control of the Middle Atmosphere Dynamical Control of the Middle Atmosphere . Space Sci Rev 168 1 283--314 . APACrefDOI doi:10.1007/s11214-011-9841-5 APACrefDOI

    work page doi:10.1007/s11214-011-9841-5 2012

  6. [6]

    , Xie, L

    Bi.etal2023 APACrefauthors Bi, K. , Xie, L. , Zhang, H. , Chen, X. , Gu, X. \ Tian, Q. APACrefauthors \ 2023 . Accurate Medium-Range Global Weather Forecasting with 3D Neural Networks Accurate medium-range global weather forecasting with 3D neural networks . Nature 619 7970 533--538 . APACrefDOI doi:10.1038/s41586-023-06185-3 APACrefDOI

    doi:10.1038/s41586-023-06185-3 2023

  7. [7]

    , Bruinsma, W P

    Bodnar.etal2025 APACrefauthors Bodnar, C. , Bruinsma, W P. , Lucic, A. , Stanley, M. , Allen, A. , Brandstetter, J. Perdikaris, P. APACrefauthors \ 2025 . A Foundation Model for the Earth System A foundation model for the Earth system . Nature 641 8065 1180--1187 . APACrefDOI doi:10.1038/s41586-025-09005-y APACrefDOI

    doi:10.1038/s41586-025-09005-y 2025

  8. [8]

    , Gettelman, A

    Bogenschutz.etal2012 APACrefauthors Bogenschutz, P A. , Gettelman, A. , Morrison, H. , Larson, V E. , Schanen, D P. , Meyer, N R. \ Craig, C. APACrefauthors \ 2012 . Unified Parameterization of the Planetary Boundary Layer and Shallow Convection with a Higher-Order Turbulence Closure in the Community Atmosphere Model : Single-Column Experiments Unified pa...

    2012

Show all 71 references
  1. [9]

    , Hudson, D A

    Bommasani.etal2022 APACrefauthors Bommasani, R. , Hudson, D A. , Adeli, E. , Altman, R. , Arora, S. , von Arx , S. Liang, P. APACrefauthors \ 2022 . On the Opportunities and Risks of Foundation Models On the Opportunities and Risks of Foundation Models \ ( \ arXiv:2108.07258)....

    2022 doi arXiv

  2. [10]

    , Henn, B

    Bretherton.etal2022 APACrefauthors Bretherton, C S. , Henn, B. , Kwa, A. , Brenowitz, N D. , Watt-Meyer , O. , McGibbon, J. Harris, L. APACrefauthors \ 2022 . Correcting Coarse-Grid Weather and Climate Models by Machine Learning From Global Storm-Resolving Simulations Correcti...

    2022 doi

  3. [11]

    , Hatfield, S

    Chantry.etal2021 APACrefauthors Chantry, M. , Hatfield, S. , Dueben, P. , Polichtchouk, I. \ Palmer, T. APACrefauthors \ 2021 . Machine Learning Emulation of Gravity Wave Drag in Numerical Weather Forecasting Machine Learning Emulation of Gravity Wave Drag in Numerical Weather...

    2021 doi

  4. [12]

    \ Berner, J

    Chapman.Berner2025 APACrefauthors Chapman, W E. \ Berner, J. APACrefauthors \ 2025 . Improving Climate Bias and Variability via CNN-Based State-Dependent Model-Error Corrections Improving Climate Bias and Variability via CNN-Based State-Dependent Model-Error Corrections . Geop...

    2025 doi

  5. [13]

    , Kwon, H

    Choi.etal2024 APACrefauthors Choi, H. , Kwon, H. , Kim, S J. \ Kim, B M. APACrefauthors \ 2024 . Warmer Antarctic Summers in Recent Decades Linked to Earlier Stratospheric Final Warming Occurrences Warmer Antarctic summers in recent decades linked to earlier stratospheric fina...

    2024 doi

  6. [14]

    \ Gerber, E P

    Connelly.Gerber2024 APACrefauthors Connelly, D S. \ Gerber, E P. APACrefauthors \ 2024 . Regression Forest Approaches to Gravity Wave Parameterization for Climate Projection Regression Forest Approaches to Gravity Wave Parameterization for Climate Projection . Journal of Advan...

    2024 doi

  7. [15]

    , Meng, F

    Cui.etal2025 APACrefauthors Cui, Z. , Meng, F. \ Luo, J. APACrefauthors \ 2025 05 . Breaking through Tropical Cyclone Intensity Prediction: A Foundation Model Prithvi-TC Breaking through tropical cyclone intensity prediction: A foundation model Prithvi-TC . Front. Comput. Sci....

    2025 doi

  8. [16]

    o rnbrack, A. , Leutbecher, M. , Kivi, R. \ Kyr \

    Doernbrack.etal1999 APACrefauthors D \"o rnbrack, A. , Leutbecher, M. , Kivi, R. \ Kyr \"o , E. APACrefauthors \ 1999 . Mountain-Wave-Induced Record Low Stratospheric Temperatures above Northern Scandinavia Mountain-wave-induced record low stratospheric temperatures above nort...

    1999 doi

  9. [17]

    , Rhode, S

    Eichinger.etal2023 APACrefauthors Eichinger, R. , Rhode, S. , Garny, H. , Preusse, P. , Pisoft, P. , Kucha r , A. Kern, B. APACrefauthors \ 2023 . Emulating Lateral Gravity Wave Propagation in a Global Chemistry--Climate Model ( EMAC v2.55.2) through Horizontal Flux Redistribu...

    2023 doi

  10. [18]

    , Sheshadri, A

    Espinosa.etal2022 APACrefauthors Espinosa, Z I. , Sheshadri, A. , Cain, G R. , Gerber, E P. \ DallaSanta, K J. APACrefauthors \ 2022 . Machine Learning Gravity Wave Parameterization Generalizes to Capture the QBO and Response to Increased CO2 Machine Learning Gravity Wave Para...

    2022 doi

  11. [19]

    \ Alexander, M J

    Fritts.Alexander2003 APACrefauthors Fritts, D C. \ Alexander, M J. APACrefauthors \ 2003 . Gravity Wave Dynamics and Effects in the Middle Atmosphere Gravity wave dynamics and effects in the middle atmosphere . Reviews of Geophysics 41 1 . APACrefDOI doi:10.1029/2001RG000106 A...

    2003 doi

  12. [20]

    , McCarty, W

    Gelaro.etal2017 APACrefauthors Gelaro, R. , McCarty, W. , Su \'a rez, M J. , Todling, R. , Molod, A. , Takacs, L. Zhao, B. APACrefauthors \ 2017 . The Modern-Era Retrospective Analysis for Research and Applications , Version 2 ( MERRA-2 ) The Modern-Era Retrospective Analysis ...

    2017 doi

  13. [21]

    , Manzini, E

    Giorgetta.etal2002 APACrefauthors Giorgetta, M A. , Manzini, E. \ Roeckner, E. APACrefauthors \ 2002 . Forcing of the Quasi-Biennial Oscillation from a Broad Spectrum of Atmospheric Waves Forcing of the quasi-biennial oscillation from a broad spectrum of atmospheric waves . Ge...

    2002 doi

  14. [22]

    , Horowitz, L W

    Golaz.etal2013 APACrefauthors Golaz, J C. , Horowitz, L W. \ Levy II, H. APACrefauthors \ 2013 . Cloud Tuning in a Coupled Climate Model: Impact on 20th Century Warming Cloud tuning in a coupled climate model: Impact on 20th century warming . Geophysical Research Letters 40 10...

    2013 doi

  15. [23]

    , Birner, T

    Gupta.etal2021a APACrefauthors Gupta, A. , Birner, T. , D \"o rnbrack, A. \ Polichtchouk, I. APACrefauthors \ 2021 . Importance of Gravity Wave Forcing for Springtime Southern Polar Vortex Breakdown as Revealed by ERA5 Importance of Gravity Wave Forcing for Springtime Southern...

    2021 doi

  16. [24]

    , Reichert, R

    Gupta.etal2024d APACrefauthors Gupta, A. , Reichert, R. , D \"o rnbrack, A. , Garny, H. , Eichinger, R. , Polichtchouk, I. Birner, T. APACrefauthors \ 2024 01 . Estimates of Southern Hemispheric Gravity Wave Momentum Fluxes Across Observations , Reanalyses , and Kilometer-scal...

    2024

  17. [25]

    , Sheshadri, A

    Gupta.etal2024c APACrefauthors Gupta, A. , Sheshadri, A. , Alexander, M J. \ Birner, T. APACrefauthors \ 2024 . Insights on Lateral Gravity Wave Propagation in the Extratropical Stratosphere from 44 Years of ERA5 Data Insights on Lateral Gravity Wave Propagation in the Extratr...

    2024 doi

  18. [26]

    , Sheshadri, A

    Gupta.etal2024 APACrefauthors Gupta, A. , Sheshadri, A. , Roy, S. , Gaur, V. , Maskey, M. \ Ramachandran, R. APACrefauthors \ 2024 06 . Machine Learning Global Simulation of Nonlocal Gravity Wave Propagation Machine Learning Global Simulation of Nonlocal Gravity Wave Propagati...

    2024 doi arXiv

  19. [27]

    , Scaife, A A

    Hardiman.etal2023 APACrefauthors Hardiman, S C. , Scaife, A A. , van Niekerk, A. , Prudden, R. , Owen, A. , Adams, S V. Madge, S. APACrefauthors \ 2023 . Machine Learning for Nonorographic Gravity Waves in a Climate Model Machine Learning for Nonorographic Gravity Waves in a C...

    2023 doi

  20. [28]

    u ndung der Theorie quadratischer Formen von unendlichvielen Ver \

    Hellinger1909 APACrefauthors Hellinger, E. APACrefauthors \ 1909 . Neue Begr \"u ndung der Theorie quadratischer Formen von unendlichvielen Ver \"a nderlichen. Neue Begr \"u ndung der Theorie quadratischer Formen von unendlichvielen Ver \"a nderlichen. Journal f \"u r die rein...

    1909 doi

  21. [29]

    , Bell, B

    Hersbach.etal2023 APACrefauthors Hersbach, H. , Bell, B. , Berrisford, P. , Biavati, G. , Hor \'a nyi, A. , Munoz-Sabater , J. Th \'e paut, J N. APACrefauthors \ 2023 . ERA5 Hourly Data on Pressure Levels from 1940 to Present. ERA5 hourly data on pressure levels from 1940 to p...

    2023 doi

  22. [30]

    , Bell, B

    Hersbach.etal2020 APACrefauthors Hersbach, H. , Bell, B. , Berrisford, P. , Hirahara, S. , Hor \'a nyi, A. , Mu \ n oz-Sabater, J. Th \'e paut, J N. APACrefauthors \ 2020 . The ERA5 Global Reanalysis The ERA5 global reanalysis . Quarterly Journal of the Royal Meteorological So...

    2020 doi

  23. [31]

    , Wright, C J

    Hindley.etal2020 APACrefauthors Hindley, N P. , Wright, C J. , Hoffmann, L. , Moffat-Griffin , T. \ Mitchell, N J. APACrefauthors \ 2020 . An 18- Year Climatology of Directional Stratospheric Gravity Wave Momentum Flux From 3- D Satellite Observations An 18- Year Climatology o...

    2020 doi

  24. [32]

    APACrefauthors \ 2024

    Hirschfeld2024 APACrefauthors Hirschfeld, A. APACrefauthors \ 2024 . Singapore Airlines Death: Is Climate Change Making Air Turbulence Worse? Singapore Airlines death: Is climate change making air turbulence worse? https://www.aljazeera.com/economy/2024/5/23/singapore-airlines...

    2024

  25. [33]

    , Spang, R

    Hoffmann.etal2017 APACrefauthors Hoffmann, L. , Spang, R. , Orr, A. , Alexander, M J. , Holt, L A. \ Stein, O. APACrefauthors \ 2017 . A Decadal Satellite Record of Gravity Wave Activity in the Lower Stratosphere to Study Polar Stratospheric Cloud Formation A decadal satellite...

    2017 doi

  26. [34]

    , Larsen, N

    Hoepfner.etal2006 APACrefauthors H \"o pfner, M. , Larsen, N. , Spang, R. , Luo, B P. , Ma, J. , Svendsen, S H. Fischer, H. APACrefauthors \ 2006 . MIPAS Detects Antarctic Stratospheric Belt of NAT PSCs Caused by Mountain Waves MIPAS detects Antarctic stratospheric belt of NAT...

    2006 doi

  27. [35]

    , Mlawer, E J

    Iacono.etal2000 APACrefauthors Iacono, M J. , Mlawer, E J. , Clough, S A. \ Morcrette, J J. APACrefauthors \ 2000 . Impact of an Improved Longwave Radiation Model, RRTM , on the Energy Budget and Thermodynamic Properties of the NCAR Community Climate Model, CCM3 Impact of an i...

    2000

  28. [36]

    , Roy, S

    Jakubik.etal2023 APACrefauthors Jakubik, J. , Roy, S. , Phillips, C E. , Fraccaro, P. , Godwin, D. , Zadrozny, B. Ramachandran, R. APACrefauthors \ 2023 . Foundation Models for Generalist Geospatial Artificial Intelligence Foundation Models for Generalist Geospatial Artificial...

    2023 doi arXiv

  29. [37]

    \ Chun, H Y

    Kim.Chun2015 APACrefauthors Kim, Y H. \ Chun, H Y. APACrefauthors \ 2015 06 . Momentum Forcing of the Quasi-Biennial Oscillation by Equatorial Waves in Recent Reanalyses Momentum forcing of the quasi-biennial oscillation by equatorial waves in recent reanalyses . Atmospheric C...

    2015 doi

  30. [38]

    , Green, B

    Kohler.etal2023 APACrefauthors K \"o hler, L. , Green, B. \ Stephan, C C. APACrefauthors \ 2023 . Comparing Loon Superpressure Balloon Observations of Gravity Waves in the Tropics With Global Storm-Resolving Models Comparing Loon Superpressure Balloon Observations of Gravity W...

    2023 doi

  31. [39]

    , Alexander, M J

    Kruse.etal2022 APACrefauthors Kruse, C G. , Alexander, M J. , Hoffmann, L. , van Niekerk, A. , Polichtchouk, I. , Bacmeister, J T. Stein, O. APACrefauthors \ 2022 04 . Observed and Modeled Mountain Waves from the Surface to the Mesosphere near the Drake Passage Observed and Mo...

    2022 doi

  32. [40]

    , Sanchez-Gonzalez , A

    Lam.etal2023 APACrefauthors Lam, R. , Sanchez-Gonzalez , A. , Willson, M. , Wirnsberger, P. , Fortunato, M. , Alet, F. Battaglia, P. APACrefauthors \ 2023 . GraphCast : Learning Skillful Medium-Range Global Weather Forecasting GraphCast : Learning skillful medium-range global ...

    2023 doi arXiv

  33. [41]

    , Wright, C J

    Lear.etal2024 APACrefauthors Lear, E J. , Wright, C J. , Hindley, N P. , Polichtchouk, I. \ Hoffmann, L. APACrefauthors \ 2024 . Comparing Gravity Waves in a Kilometer-Scale Run of the IFS to AIRS Satellite Observations and ERA5 Comparing Gravity Waves in a Kilometer-Scale Run...

    2024 doi

  34. [42]

    , Calvin, K

    Lee.etal2023 APACrefauthors Lee, H. , Calvin, K. , Dasgupta, D. , Krinner, G. , Mukherji, A. , Thorne, P. Barrett, K. APACrefauthors \ 2023 . Climate Change 2023: Synthesis Report. Contribution of Working Groups I , II and III to the Sixth Assessment Report of the Intergovernm...

    2023

  35. [43]

    , Luise, I

    Lessig.etal2023 APACrefauthors Lessig, C. , Luise, I. , Gong, B. , Langguth, M. , Stadtler, S. \ Schultz, M. APACrefauthors \ 2023 . AtmoRep : A Stochastic Model of Atmosphere Dynamics Using Large Scale Representation Learning AtmoRep : A stochastic model of atmosphere dynamic...

    2023 doi arXiv

  36. [44]

    APACrefauthors \ 2015

    Lindborg2015 APACrefauthors Lindborg, E. APACrefauthors \ 2015 . A Helmholtz Decomposition of Structure Functions and Spectra Calculated from Aircraft Data A Helmholtz decomposition of structure functions and spectra calculated from aircraft data . Journal of Fluid Mechanics 7...

    2015 doi

  37. [45]

    \ Miller, M J

    Lott.Miller1997 APACrefauthors Lott, F. \ Miller, M J. APACrefauthors \ 1997 . A New Subgrid-Scale Orographic Drag Parametrization: Its Formulation and Testing A new subgrid-scale orographic drag parametrization: Its formulation and testing . Quarterly Journal of the Royal Met...

    1997 doi

  38. [46]

    Lu.etal2024 APACrefauthors Lu, Y. , Xu, X. , Wang, L. , Liu, Y. , Wu, T. , Jie, W. \ Sun, J. APACrefauthors \ 2024 . Machine Learning Emulation of Subgrid-Scale Orographic Gravity Wave Drag in a General Circulation Model With Middle Atmosphere Extension Machine Learning Emulat...

    2024 doi

  39. [47]

    , Stevens, B

    Mauritsen.etal2012 APACrefauthors Mauritsen, T. , Stevens, B. , Roeckner, E. , Crueger, T. , Esch, M. , Giorgetta, M. Tomassini, L. APACrefauthors \ 2012 . Tuning the Climate of a Global Model Tuning the climate of a global model . Journal of Advances in Modeling Earth Systems...

    2012 doi

  40. [48]

    , Shepherd, T G

    McLandress.etal2012a APACrefauthors McLandress, C. , Shepherd, T G. , Polavarapu, S. \ Beagley, S R. APACrefauthors \ 2012 . Is Missing Orographic Gravity Wave Drag near 60 ^ S the Cause of the Stratospheric Zonal Wind Biases in Chemistry -- Climate Models ? Is Missing Orograp...

    2012 doi

  41. [49]

    \ Lawrence, P

    Morrison.Lawrence2020 APACrefauthors Morrison, M A. \ Lawrence, P. APACrefauthors \ 2020 . Understanding Model-Based Uncertainty in Climate Science Understanding Model-Based Uncertainty in Climate Science . G. Pellegrino\ M. Di Paola\ ( ), Handbook of Philosophy of Climate Cha...

    2020 doi

  42. [50]

    , Schlemper, J

    Oktay.etal2018 APACrefauthors Oktay, O. , Schlemper, J. , Folgoc, L L. , Lee, M. , Heinrich, M. , Misawa, K. Rueckert, D. APACrefauthors \ 2018 . Attention U-Net : Learning Where to Look for the Pancreas Attention U-Net : Learning Where to Look for the Pancreas \ ( \ arXiv:180...

    2018 doi arXiv

  43. [51]

    , Wallace, J M

    Pahlavan.etal2023 APACrefauthors Pahlavan, H A. , Wallace, J M. \ Fu, Q. APACrefauthors \ 2023 02 . Characteristics of Tropical Convective Gravity Waves Resolved by ERA5 Reanalysis Characteristics of Tropical Convective Gravity Waves Resolved by ERA5 Reanalysis . Journal of th...

    2023 doi

  44. [52]

    , Shutts, G J

    Palmer.etal1986 APACrefauthors Palmer, T N. , Shutts, G J. \ Swinbank, R. APACrefauthors \ 1986 . Alleviation of a Systematic Westerly Bias in General Circulation and Numerical Weather Prediction Models through an Orographic Gravity Wave Drag Parametrization Alleviation of a s...

    1986

  45. [53]

    , de la C \'a mara , A

    Plougonven.etal2020 APACrefauthors Plougonven, R. , de la C \'a mara , A. , Hertzog, A. \ Lott, F. APACrefauthors \ 2020 . How Does Knowledge of Atmospheric Gravity Waves Guide Their Parameterizations? How does knowledge of atmospheric gravity waves guide their parameterizatio...

    2020 doi

  46. [54]

    , van Niekerk, A

    Polichtchouk.etal2023 APACrefauthors Polichtchouk, I. , van Niekerk, A. \ Wedi, N. APACrefauthors \ 2023 01 . Resolved Gravity Waves in the Extratropical Stratosphere : Effect of Horizontal Resolution Increase from O (10) to O (1) Km Resolved Gravity Waves in the Extratropical...

    2023 doi

  47. [55]

    , Wedi, N

    Polichtchouk.etal2022 APACrefauthors Polichtchouk, I. , Wedi, N. \ Kim, Y H. APACrefauthors \ 2022 . Resolved Gravity Waves in the Tropical Stratosphere: Impact of Horizontal Resolution and Deep Convection Parametrization Resolved gravity waves in the tropical stratosphere: Im...

    2022 doi

  48. [56]

    , Sanchez-Gonzalez , A

    Price.etal2025 APACrefauthors Price, I. , Sanchez-Gonzalez , A. , Alet, F. , Andersson, T R. , El-Kadi , A. , Masters, D. Willson, M. APACrefauthors \ 2025 . Probabilistic Weather Forecasting with Machine Learning Probabilistic weather forecasting with machine learning . Natur...

    2025 doi

  49. [57]

    , Kruse, C G

    Prochazkova.etal2023 APACrefauthors Proch \'a zkov \'a , Z. , Kruse, C G. , Alexander, M J. , Hoffmann, L. , Bacmeister, J T. , Holt, L. S \'a cha, P. APACrefauthors \ 2023 06 . Sensitivity of Mountain Wave Drag Estimates on Separation Methods and Proposed Improvements Sensiti...

    2023 doi

  50. [58]

    \ Gupta, A

    RoyGupta2025 APACrefauthors Roy, S. \ Gupta, A. APACrefauthors \ 2025 08 . Amangupta2/Gravity-Wave-Finetuning-James: Release v1.0.0 -- Gravity Wave Flux Fine-Tuning with Prithvi WxC [ Software ]. Amangupta2/gravity-wave-finetuning-james: Release v1.0.0 -- Gravity Wave Flux Fin...

    2025 doi

  51. [59]

    , KUMAR, ANKUR

    Roy.etal2025 APACrefauthors Roy, S. , KUMAR, ANKUR . , Shinde, R. , Gupta, A. \ simonpf . APACrefauthors \ 2025 08 . Amangupta2/ WxC-Bench : Release v1.0.0 -- WxC-Bench : A Benchmark Suite for AI Foundation Models in Weather and Climate [ Software ]. Amangupta2/ WxC-Bench : Re...

    2025 doi

  52. [60]

    , Tateno, S

    Sato.etal2012 APACrefauthors Sato, K. , Tateno, S. , Watanabe, S. \ Kawatani, Y. APACrefauthors \ 2012 . Gravity Wave Characteristics in the Southern Hemisphere Revealed by a High-Resolution Middle-Atmosphere General Circulation Model Gravity Wave Characteristics in the Southe...

    2012 doi

  53. [61]

    , Roy, S

    Schmude.etal2024 APACrefauthors Schmude, J. , Roy, S. , Trojak, W. , Jakubik, J. , Civitarese, D S. , Singh, S. Ramachandran, R. APACrefauthors \ 2024 . Prithvi WxC : Foundation Model for Weather and Climate Prithvi WxC : Foundation Model for Weather and Climate . APACrefDOI d...

    2024 doi arXiv

  54. [62]

    , Strube, C

    Stephan.etal2019 APACrefauthors Stephan, C C. , Strube, C. , Klocke, D. , Ern, M. , Hoffmann, L. , Preusse, P. \ Schmidt, H. APACrefauthors \ 2019 09 . Intercomparison of Gravity Waves in Global Convection-Permitting Models Intercomparison of Gravity Waves in Global Convection...

    2019 doi

  55. [63]

    , Pahlavan, H A

    Sun.etal2024 APACrefauthors Sun, Y Q. , Pahlavan, H A. , Chattopadhyay, A. , Hassanzadeh, P. , Lubis, S W. , Alexander, M J. Guan, Y. APACrefauthors \ 2024 . Data Imbalance , Uncertainty Quantification , and Transfer Learning in Data-Driven Parameterizations : Lessons From the...

    2024

  56. [64]

    \ Chantry, M

    Ukkonen.Chantry2024 APACrefauthors Ukkonen, P. \ Chantry, M. APACrefauthors \ 2024 . Representing Sub-Grid Processes in Weather and Climate Models via Sequence Learning. Representing sub-grid processes in weather and climate models via sequence learning. APACrefDOI doi:10.2254...

    2024

  57. [65]

    o l \"o ni, G. , Kim, Y H. , Z \

    Voelker.etal2023 APACrefauthors Voelker, G S. , B \"o l \"o ni, G. , Kim, Y H. , Z \"a ngl, G. \ Achatz, U. APACrefauthors \ 2023 . MS-GWaM : A 3-Dimensional Transient Gravity Wave Parametrization for Atmospheric Models MS-GWaM : A 3-dimensional transient gravity wave parametr...

    2023 arXiv

  58. [66]

    , Zhang, F

    Wei.etal2022 APACrefauthors Wei, J. , Zhang, F. , Richter, J H. , Alexander, M J. \ Sun, Y Q. APACrefauthors \ 2022 . Global Distributions of Tropospheric and Stratospheric Gravity Wave Momentum Fluxes Resolved by the 9-Km ECMWF Experiments Global Distributions of Tropospheric...

    2022 doi

  59. [67]

    , Miao, C

    Wu.etal2022 APACrefauthors Wu, Y. , Miao, C. , Fan, X. , Gou, J. , Zhang, Q. \ Zheng, H. APACrefauthors \ 2022 . Quantifying the Uncertainty Sources of Future Climate Projections and Narrowing Uncertainties With Bias Correction Techniques Quantifying the Uncertainty Sources of...

    2022 doi

  60. [68]

    , Tomikawa, Y

    Yoshida.etal2024 APACrefauthors Yoshida, L. , Tomikawa, Y. , Ejiri, M K. , Tsutsumi, M. , Kohma, M. \ Sato, K. APACrefauthors \ 2024 . Large- Amplitude Inertia Gravity Waves Over Syowa Station : Comparison of PANSY Radar and ERA5 Reanalysis Data Large- Amplitude Inertia Gravit...

    2024 doi

  61. [69]

    , Golaz, J C

    Zhao.etal2018 APACrefauthors Zhao, M. , Golaz, J C. , Held, I M. , Guo, H. , Balaji, V. , Benson, R. Xiang, B. APACrefauthors \ 2018 . The GFDL Global Atmosphere and Land Model AM4 .0/ LM4 .0: 1. Simulation Characteristics With Prescribed SSTs The GFDL Global Atmosphere and La...

    2018 doi

  62. [70]

    write newline

    " write newline " cite write " FUNCTION editor.postfix editor num.names #1 > "( )" "( )" if FUNCTION editor.trans.postfix editor num.names #1 > "( )" "( )" if FUNCTION trans.postfix translator num.names #1 > "( )" "( )" if FUNCTION authors.editors.reflist.apa5 'field := 'dot :...

  63. [71]

    write newline

    " write newline " cite write " FUNCTION editor.postfix editor num.names #1 > "( )" "( )" if FUNCTION editor.trans.postfix editor num.names #1 > "( )" "( )" if FUNCTION trans.postfix translator num.names #1 > "( )" "( )" if FUNCTION authors.editors.reflist.apa5 'field := 'dot :...

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.