pith. sign in

arxiv: 2412.08079 · v3 · submitted 2024-12-11 · 💻 cs.LG · cs.NA· math.NA· physics.ao-ph

Regional climate risk assessment from climate models using probabilistic machine learning

Zhong Yi Wan , Ignacio Lopez-Gomez , Robert Carver , Tapio Schneider , John Anderson , Fei Sha , Leonardo Zepeda-N\'u\~nez This is my paper

Pith reviewed 2026-05-23 07:24 UTC · model grok-4.3

classification 💻 cs.LG cs.NAmath.NAphysics.ao-ph
keywords climate downscalinggenerative modelingprobabilistic machine learningregional climate riskhazard synthesisfine-scale weather generationunpaired training
0
0 comments X

The pith

GenFocal produces fine-scale weather statistics from coarse climate projections without any paired training data.

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

The paper presents GenFocal as a probabilistic machine learning framework that downscales coarse global climate model outputs to fine-scale regional weather. It trains without needing matched pairs of simulated and observed events, yet still generates realistic complex hazards such as heat waves and tropical cyclones that may be missing or distorted in the coarse input. It also produces more accurate samples of rare high-impact events than current methods. A reader would care because climate risk decisions at the regional level require localized information that global models cannot directly supply.

Core claim

GenFocal generates statistically accurate, fine-scale weather from coarse climate projections without requiring paired simulated and observed events during training. It synthesizes complex and long-lived hazards, such as heat waves and tropical cyclones, even when they are not well represented in the coarse climate projections, and samples high-impact rare events more accurately than leading methods.

What carries the argument

GenFocal, a probabilistic generative model that learns a mapping from coarse climate inputs to fine-scale outputs without paired examples.

If this is right

  • Large-scale climate projections can be turned into localized data usable for regional adaptation planning.
  • Hazards poorly resolved in global models can still be synthesized at fine scales.
  • High-impact rare events can be sampled with higher fidelity than with existing downscaling techniques.
  • Climate risk assessment can proceed without the need to generate and store paired high-resolution simulations for every training case.

Where Pith is reading between the lines

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

  • The method could be tested on other variables or regions where observational records are sparse to check whether the unpaired training still holds.
  • If the generated statistics prove reliable, ensembles of such outputs might replace or supplement some high-resolution dynamical downscaling runs.
  • The framework might be combined with existing climate model archives to produce updated regional risk maps without retraining on new paired data each time.

Load-bearing premise

The generative model can recover accurate fine-scale statistics and rare-event distributions for hazards that the coarse input does not capture, using only unpaired training data and no explicit physical constraints.

What would settle it

Statistical comparison of generated versus observed event frequencies for a hazard such as tropical cyclone intensity or heat-wave duration in a region where the coarse model systematically under-represents the hazard.

read the original abstract

Effective climate risk assessment is hindered by the resolution gap between coarse global climate models and the fine-scale information needed for regional decisions. We introduce GenFocal, an AI framework that generates statistically accurate, fine-scale weather from coarse climate projections, without requiring paired simulated and observed events during training. GenFocal synthesizes complex and long-lived hazards, such as heat waves and tropical cyclones, even when they are not well represented in the coarse climate projections. It also samples high-impact, rare events more accurately than leading methods. By translating large-scale climate projections into actionable, localized information, GenFocal provides a powerful new paradigm to improve climate adaptation and resilience strategies.

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 paper introduces GenFocal, a probabilistic generative AI framework that produces statistically accurate fine-scale weather fields from coarse global climate model projections. Training requires no paired coarse-fine examples; instead it combines unpaired coarse projections with an unconditional fine-scale prior. The method is claimed to synthesize complex, long-lived hazards (heat waves, tropical cyclones) even when poorly resolved in the coarse driver and to sample high-impact rare events more accurately than existing approaches, thereby enabling improved regional climate risk assessment.

Significance. If the performance claims are substantiated by rigorous validation, the work would represent a meaningful advance for downscaling and risk assessment by removing the paired-data requirement that limits many current statistical and machine-learning methods. The ability to generate plausible fine-scale statistics for hazards absent from the coarse input, if demonstrated without circularity or hidden physical constraints, would be a notable technical contribution.

major comments (2)
  1. [Method / Section 3] Method / Section 3: The training procedure is described as using only unpaired coarse projections plus an unconditional fine-scale prior, with no explicit physical constraints (conservation laws, vorticity budgets, or observed tail statistics) in the loss or architecture. This is load-bearing for the central claim that the model can accurately synthesize hazards (heat waves, tropical cyclones) that are poorly represented or absent in the coarse input; without such constraints or paired conditioning, the generated conditional distributions may deviate from real statistics.
  2. [Results / validation sections] Results / validation sections: The abstract asserts superior sampling of rare high-impact events, yet the provided description contains no quantitative metrics (e.g., tail quantiles, frequency-intensity distributions, or proper scoring rules against observations) that would demonstrate the generative prior has not simply produced statistically plausible but biased samples. This directly affects the strength of the accuracy claim relative to leading methods.
minor comments (2)
  1. [Method] Notation for the generative prior and conditioning mechanism should be clarified with explicit equations to allow readers to assess whether any implicit physical regularization is present.
  2. [Figures] Figure captions and axis labels for hazard-specific diagnostics (e.g., cyclone tracks, heat-wave duration) need to include the exact observational reference datasets used for comparison.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thoughtful and constructive comments on our manuscript. We address each major comment point by point below, indicating the revisions we will make to strengthen the paper.

read point-by-point responses

  1. Referee: [Method / Section 3] Method / Section 3: The training procedure is described as using only unpaired coarse projections plus an unconditional fine-scale prior, with no explicit physical constraints (conservation laws, vorticity budgets, or observed tail statistics) in the loss or architecture. This is load-bearing for the central claim that the model can accurately synthesize hazards (heat waves, tropical cyclones) that are poorly represented or absent in the coarse input; without such constraints or paired conditioning, the generated conditional distributions may deviate from real statistics.

    Authors: The GenFocal approach learns an unconditional fine-scale prior from high-resolution observations or simulations that encodes the full statistical distribution of weather phenomena, including long-lived hazards. The generative model then produces samples conditioned on the coarse input while remaining consistent with this prior, allowing synthesis of events underrepresented in the driver. We will revise Section 3 to include a clearer derivation of the training objective, additional analysis of how higher-order statistics are matched implicitly, and new experiments verifying physical consistency (e.g., integrated energy and vorticity budgets) on generated fields. We agree that making these aspects more explicit will address the concern. revision: partial

  2. Referee: [Results / validation sections] Results / validation sections: The abstract asserts superior sampling of rare high-impact events, yet the provided description contains no quantitative metrics (e.g., tail quantiles, frequency-intensity distributions, or proper scoring rules against observations) that would demonstrate the generative prior has not simply produced statistically plausible but biased samples. This directly affects the strength of the accuracy claim relative to leading methods.

    Authors: The results section contains several quantitative comparisons, but we acknowledge that tail-specific metrics and proper scoring rules for rare events are not presented with sufficient prominence or detail. We will add a dedicated subsection with tail quantile plots, frequency-intensity distributions for tropical cyclones and heat waves, and CRPS evaluations against observations. These additions will directly demonstrate that the samples are not biased relative to leading methods and will be included in the revised manuscript. revision: yes

Circularity Check

0 steps flagged

No circularity identified in derivation chain

full rationale

The provided abstract and context describe GenFocal as a probabilistic generative framework trained on unpaired coarse projections plus an unconditional fine-scale prior, with claims about synthesizing hazards and sampling rare events. No equations, loss functions, fitting procedures, or self-citations are quoted that reduce any claimed prediction or result to an input quantity by construction. No self-definitional steps, fitted inputs renamed as predictions, or load-bearing self-citations appear in the text. The derivation chain is therefore self-contained against external benchmarks, consistent with the default expectation for most papers.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract supplies no information on free parameters, background axioms, or new postulated entities.

pith-pipeline@v0.9.0 · 5662 in / 1025 out tokens · 18824 ms · 2026-05-23T07:24:34.461432+00:00 · methodology

discussion (0)

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

Forward citations

Cited by 3 Pith papers

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

  1. IPSL-AID: Generative Diffusion Models for Climate Downscaling from Global to Regional Scales

    physics.ao-ph 2026-03 unverdicted novelty 6.0

    A diffusion model trained on ERA5 reanalysis downscales global climate data to 0.25° resolution while capturing statistical distributions and extremes.

  2. EnScale: Temporally-consistent multivariate generative downscaling via proper scoring rules

    physics.ao-ph 2025-09 unverdicted novelty 6.0

    EnScale emulates high-resolution regional climate model outputs from global circulation models for multiple variables using a two-step generative process with sparse local stochastic layers and energy score optimizati...

  3. Downscaling weather forecasts from Low- to High-Resolution with Diffusion Models

    physics.ao-ph 2026-03 unverdicted novelty 5.0

    A conditional diffusion model downscales global atmospheric forecasts from 100 km to 30 km resolution while improving probabilistic skill, matching power spectra, and preserving physical relationships.

Reference graph

Works this paper leans on

123 extracted references · 123 canonical work pages · cited by 3 Pith papers · 3 internal anchors

  1. [3]

    Heat waves in the United States: Mor- tality risk during heat waves and effect modification by heat wave characteristics in 43 U.S

    Anderson G Brooke and Bell Michelle L. Heat waves in the United States: Mor- tality risk during heat waves and effect modification by heat wave characteristics in 43 U.S. communities.Environmental Health Perspectives, 119:210–218, 2 2011. doi: 10.1289/ehp.1002313

    work page doi:10.1289/ehp.1002313 2011

  2. [4]

    State-of-the-art bias correction of climate models misrepresent climate science and misinform adaptation.Environmental Research Letters, 19(9):094052, 2024

    Vikram Singh Chandel, Udit Bhatia, Auroop R Ganguly, and Subimal Ghosh. State-of-the-art bias correction of climate models misrepresent climate science and misinform adaptation.Environmental Research Letters, 19(9):094052, 2024

    work page 2024

  3. [5]

    Christensen, Fredrik Boberg, Ole B

    Jens H. Christensen, Fredrik Boberg, Ole B. Christensen, and Philippe Lucas- Picher. On the need for bias correction of regional climate change projections of temperature and precipitation.Geophysical Research Letters, 35(20), 2008

    work page 2008

  4. [6]

    ERA5 hourly data on single levels from 1940 to present, 2023

    Copernicus Climate Change Service, Climate Data Store. ERA5 hourly data on single levels from 1940 to present, 2023

    work page 1940

  5. [7]

    Kristina Dahl, Rachel Licker, John T Abatzoglou, and Juan Declet-Barreto. Increased frequency of and population exposure to extreme heat index days in the United States during the 21st century.Environmental research communications, 1(7):075002, 1 August 2019

    work page 2019

  6. [8]

    Exploring super-resolution spatial downscaling of several meteorological variables and potential applications for photovoltaic power.Scientific Reports, 14(1):7254, 2024

    Alessandro Damiani, Noriko N Ishizaki, Hidetaka Sasaki, Sarah Feron, and Raul R Cordero. Exploring super-resolution spatial downscaling of several meteorological variables and potential applications for photovoltaic power.Scientific Reports, 14(1):7254, 2024

    work page 2024

  7. [9]

    C. A. Davis. Resolving tropical cyclone intensity in models.Geophysical Research Letters, 45(4):2082–2087, 2018

    work page 2082

  8. [10]

    Changes in heat index associated with CO2-induced global warming.Climatic change, 43(2):369– 386, October 1999

    Thomas L Delworth, J D Mahlman, and Thomas R Knutson. Changes in heat index associated with CO2-induced global warming.Climatic change, 43(2):369– 386, October 1999. 85

    work page 1999

  9. [11]

    Extreme weather and climate vulnerabilities of the electric grid: A summary of environmental sensitivity quan- tification methods

    Melissa Dumas, Binita Kc, and Colin I Cunliff. Extreme weather and climate vulnerabilities of the electric grid: A summary of environmental sensitivity quan- tification methods. Technical report, Oak Ridge National Lab.(ORNL), Oak Ridge, TN (United States), 2019

    work page 2019

  10. [12]

    Climate nowcasting

    Andrew Gettelman, Claudia Tebaldi, and L Ruby Leung. Climate nowcasting. Environmental Research: Climate, 4:013002, 3 2025

    work page 2025

  11. [13]

    Thirty years of regional climate modeling: Where are we and where are we going next?Journal of Geophysical Research: Atmospheres, 124:5696–5723, 6 2019

    Filippo Giorgi. Thirty years of regional climate modeling: Where are we and where are we going next?Journal of Geophysical Research: Atmospheres, 124:5696–5723, 6 2019

    work page 2019

  12. [14]

    Use-inspired, process-oriented GCM selection: Prioritizing models for regional dynamical downscaling.Bulletin of the American Meteorological Society, 104:E1619–E1629, 2023

    Naomi Goldenson, L Ruby Leung, Linda O Mearns, David W Pierce, Kevin A Reed, Isla R Simpson, Paul Ullrich, Will Krantz, Alex Hall, Andrew Jones, and Stefan Rahimi. Use-inspired, process-oriented GCM selection: Prioritizing models for regional dynamical downscaling.Bulletin of the American Meteorological Society, 104:E1619–E1629, 2023

    work page 2023

  13. [15]

    Climate change is increasing the likelihood of extreme autumn wildfire conditions across California.Environmental Research Letters, 15:094016, 9 2020

    Michael Goss, Daniel L Swain, John T Abatzoglou, Ali Sarhadi, Crystal A Kolden, A Park Williams, and Noah S Diffenbaugh. Climate change is increasing the likelihood of extreme autumn wildfire conditions across California.Environmental Research Letters, 15:094016, 9 2020

    work page 2020

  14. [16]

    Projecting regional change.Science, 346(6216):1461–1462, 2014

    Alex Hall. Projecting regional change.Science, 346(6216):1461–1462, 2014

    work page 2014

  15. [18]

    ERA5 hourly data on single levels from 1940 to present, 2023

    H Hersbach, B Bell, P Berrisford, G Biavati, A Hor´ anyi, J Mu˜ noz Sabater, J Nico- las, C Peubey, R Radu, I Rozum, D Schepers, A Simmons, C Soci, D Dee, and J-N Th´ epaut. ERA5 hourly data on single levels from 1940 to present, 2023

    work page 1940

  16. [19]

    Hans Hersbach, Bill Bell, Paul Berrisford, Shoji Hirahara, Andr´ as Hor´ anyi, Joaqu´ ın Mu˜ noz-Sabater, Julien Nicolas, Carole Peubey, Raluca Radu, Dinand Schepers, Adrian Simmons, Cornel Soci, Saleh Abdalla, Xavier Abellan, Gian- paolo Balsamo, Peter Bechtold, Gionata Biavati, Jean Bidlot, Massimo Bonavita, Giovanna De Chiara, Per Dahlgren, Dick Dee, M...

    work page 1999

  17. [21]

    Renzhi Jing, Ning Lin, Kerry Emanuel, Gabriel Vecchi, and Thomas R. Knutson. A comparison of tropical cyclone projections in a high-resolution global climate model and from downscaling by statistical and statistical-deterministic methods. Journal of Climate, 34(23):9349 – 9364, 2021

    work page 2021

  18. [23]

    Observed 1970–2005 cool- ing of summer daytime temperatures in coastal California.Journal of Climate, 22:3558–3573, 2009

    Bereket Lebassi, Jorge Gonz´ alez, Drazen Fabris, Edwin Maurer, Norman Miller, Cristina Milesi, Paul Switzer, and Robert Bornstein. Observed 1970–2005 cool- ing of summer daytime temperatures in coastal California.Journal of Climate, 22:3558–3573, 2009

    work page 1970

  19. [24]

    Gen- erative emulation of weather forecast ensembles with diffusion models.Science Advances, 10:eadk4489, 6 2024

    Lizao Li, Robert Carver, Ignacio Lopez-Gomez, Fei Sha, and John Anderson. Gen- erative emulation of weather forecast ensembles with diffusion models.Science Advances, 10:eadk4489, 6 2024

    work page 2024

  20. [27]

    Joseph W Lockwood, Avantika Gori, and Pierre Gentine. A generative super- resolution model for enhancing tropical cyclone wind field intensity and reso- lution.Journal of Geophysical Research: Machine Learning and Computation, 1(4):e2024JH000375, 2024

    work page 2024

  21. [30]

    Meehl and Claudia Tebaldi

    Gerald A. Meehl and Claudia Tebaldi. More intense, more frequent, and longer lasting heat waves in the 21st century.Science, 305(5686):994–997, 2004

    work page 2004

  22. [31]

    Insurance in a climate of change.Science, 309:1040–1044, 8 2005

    Evan Mills. Insurance in a climate of change.Science, 309:1040–1044, 8 2005. doi: 10.1126/science.1112121. 87

    work page doi:10.1126/science.1112121 2005

  23. [32]

    Modernizing probable maximum precipitation estimation

    National Academies of Sciences, Engineering, and Medicine. Modernizing probable maximum precipitation estimation. Technical report, 2024

    work page 2024

  24. [33]

    Global prediction of extreme floods in ungauged watersheds.Nature, 627:559–563, 2024

    Grey Nearing, Deborah Cohen, Vusumuzi Dube, Martin Gauch, Oren Gilon, Shaun Harrigan, Avinatan Hassidim, Daniel Klotz, Frederik Kratzert, Asher Metzger, Sella Nevo, Florian Pappenberger, Christel Prudhomme, Guy Shalev, Shlomo Shenzis, Tadele Yednkachw Tekalign, Dana Weitzner, and Yossi Matias. Global prediction of extreme floods in ungauged watersheds.Nat...

    work page 2024

  25. [34]

    Pierce, Daniel R

    David W. Pierce, Daniel R. Cayan, Daniel R. Feldman, and Mark D. Risser. Future increases in North American extreme precipitation in CMIP6 downscaled with LOCA.Journal of Hydrometeorology, 24(5):951 – 975, 2023

    work page 2023

  26. [35]

    Pierce, Daniel R

    David W. Pierce, Daniel R. Cayan, and Bridget L. Thrasher. Statistical downscal- ing using localized constructed analogs (LOCA).Journal of Hydrometeorology, 15(6):2558 – 2585, 2014

    work page 2014

  27. [36]

    Probabilistic weather forecasting with machine learning.Nature, 637:84–90, 2025

    Ilan Price, Alvaro Sanchez-Gonzalez, Ferran Alet, Tom R Andersson, Andrew El- Kadi, Dominic Masters, Timo Ewalds, Jacklynn Stott, Shakir Mohamed, Peter Battaglia, Remi Lam, and Matthew Willson. Probabilistic weather forecasting with machine learning.Nature, 637:84–90, 2025

    work page 2025

  28. [37]

    Decarbonized energy system planning with high-resolution spatial representa- tion of renewables lowers cost.Cell Reports Sustainability, 1, 12 2024

    Liying Qiu, Rahman Khorramfar, Saurabh Amin, and Michael F Howland. Decarbonized energy system planning with high-resolution spatial representa- tion of renewables lowers cost.Cell Reports Sustainability, 1, 12 2024. doi: 10.1016/j.crsus.2024.100263

    work page doi:10.1016/j.crsus.2024.100263 2024

  29. [40]

    Pier Siebesma

    Tapio Schneider, Jo˜ ao Teixeira, Christopher S Bretherton, Florent Brient, Kyle G Pressel, Christoph Sch¨ ar, and A. Pier Siebesma. Climate goals and computing the future of clouds.Nature Climate Change, 7(1):3–5, 2017. 88

    work page 2017

  30. [42]

    Score-based generative modeling through stochastic differential equations

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Ste- fano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equations. InInternational Conference on Learning Representations, 2020

    work page 2020

  31. [43]

    NASA global daily downscaled projections, CMIP6

    Bridget Thrasher, Weile Wang, Andrew Michaelis, Forrest Melton, Tsengdar Lee, and Ramakrishna Nemani. NASA global daily downscaled projections, CMIP6. Scientific Data, 9:262, 2022

    work page 2022

  32. [44]

    Paul A. Ullrich. Validation of LOCA2 and STAR-ESDM statistically downscaled products. Technical report, Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States), 10 2023

    work page 2023

  33. [45]

    Daniel Walton, Neil Berg, David Pierce, Ed Maurer, Alex Hall, Yen-Heng Lin, Stefan Rahimi, and Dan Cayan. Understanding differences in California climate projections produced by dynamical and statistical downscaling.Journal of Geo- physical Research: Atmospheres, 125(19):e2020JD032812, 2020. e2020JD032812 2020JD032812

    work page 2020

  34. [46]

    Sources of uncertainty for wheat yield projections under future climate are site-specific.Nature Food, 1:720–728, 2020

    Bin Wang, Puyu Feng, De Li Liu, Garry J O’Leary, Ian Macadam, Cathy Waters, Senthold Asseng, Annette Cowie, Tengcong Jiang, Dengpan Xiao, Hongyan Ruan, Jianqiang He, and Qiang Yu. Sources of uncertainty for wheat yield projections under future climate are site-specific.Nature Food, 1:720–728, 2020

    work page 2020

  35. [47]

    Hydrologic implications of dynamical and statistical approaches to downscaling climate model outputs.Climatic change, 62:189–216, 2004

    Andrew W Wood, Lai R Leung, Venkataramana Sridhar, and DP Lettenmaier. Hydrologic implications of dynamical and statistical approaches to downscaling climate model outputs.Climatic change, 62:189–216, 2004

    work page 2004

  36. [48]

    Long-range experimental hydrologic forecasting for the eastern United States

    Andrew W Wood, Edwin P Maurer, Arun Kumar, and Dennis P Lettenmaier. Long-range experimental hydrologic forecasting for the eastern United States. Journal of Geophysical Research: Atmospheres, 107(D20):ACL–6, 2002

    work page 2002

  37. [49]

    Zelinka, Timothy A

    Mark D. Zelinka, Timothy A. Myers, Daniel T. McCoy, Stephen Po-Chedley, Peter M. Caldwell, Paulo Ceppi, Stephen A. Klein, and Karl E. Taylor. Causes of higher climate sensitivity in CMIP6 models.Geophysical Research Letters, 47(1):e2019GL085782, 2020. e2019GL085782 10.1029/2019GL085782

    work page doi:10.1029/2019gl085782 2020

  38. [50]

    Future climate risk from compound events.Nature Climate Change, 8:469–477, 2018

    Jakob Zscheischler, Seth Westra, Bart J J M van den Hurk, Sonia I Seneviratne, Philip J Ward, Andy Pitman, Amir AghaKouchak, David N Bresch, Michael Leonard, Thomas Wahl, and Xuebin Zhang. Future climate risk from compound events.Nature Climate Change, 8:469–477, 2018. 89

    work page 2018

  39. [51]

    Zico Kolter

    Brandon Amos, Lei Xu, and J. Zico Kolter. Input convex neural networks. In Doina Precup and Yee Whye Teh, editors,Proceedings of the 34th International Conference on Machine Learning, volume 70 ofProceedings of Machine Learning Research, pages 146–155. PMLR, 06–11 Aug 2017

    work page 2017

  40. [52]

    Reverse-time diffusion equation models.Stochastic Processes and their Applications, 12(3):313–326, 1982

    Brian DO Anderson. Reverse-time diffusion equation models.Stochastic Processes and their Applications, 12(3):313–326, 1982

    work page 1982

  41. [53]

    Tropical cyclone minimum sea level pressure/maximum sustained wind relationship for the western north pacific

    Gary D Atkinson and Charles R Holliday. Tropical cyclone minimum sea level pressure/maximum sustained wind relationship for the western north pacific. Monthly Weather Review, 105(4):421–427, 1977

    work page 1977

  42. [54]

    Ruby Leung, David R

    Karthik Balaguru, Wenwei Xu, Chuan-Chieh Chang, L. Ruby Leung, David R. Judi, Samson M. Hagos, Michael F. Wehner, James P. Kossin, and Mingfang Ting. Increased U.S. coastal hurricane risk under climate change.Science Advances, 9(14):eadf0259, 2023

    work page 2023

  43. [55]

    All are worth words: a vit backbone for score-based diffusion models

    Fan Bao, Chongxuan Li, Yue Cao, and Jun Zhu. All are worth words: a vit backbone for score-based diffusion models. InNeurIPS 2022 Workshop on Score- Based Methods, 2022

    work page 2022

  44. [56]

    Lumiere: A space-time diffusion model for video generation.arXiv preprint arXiv:2401.12945, 2024

    Omer Bar-Tal, Hila Chefer, Omer Tov, Charles Herrmann, Roni Paiss, Shiran Zada, Ariel Ephrat, Junhwa Hur, Yuanzhen Li, Tomer Michaeli, et al. Lumiere: A space-time diffusion model for video generation.arXiv preprint arXiv:2401.12945, 2024

    work page arXiv 2024

  45. [57]

    Unpaired downscaling of fluid flows with diffusion bridges.Artificial Intelligence for the Earth Systems, 3(2):e230039, 2024

    Tobias Bischoff and Katherine Deck. Unpaired downscaling of fluid flows with diffusion bridges.Artificial Intelligence for the Earth Systems, 3(2):e230039, 2024

    work page 2024

  46. [58]

    Dif- fusion schr¨ odinger bridge with applications to score-based generative modeling

    Valentin De Bortoli, James Thornton, Jeremy Heng, and Arnaud Doucet. Dif- fusion schr¨ odinger bridge with applications to score-based generative modeling. In A. Beygelzimer, Y. Dauphin, P. Liang, and J. Wortman Vaughan, editors, Advances in Neural Information Processing Systems, 2021

    work page 2021

  47. [59]

    Cand` es and Carlos Fernandez-Granda

    Emmanuel J. Cand` es and Carlos Fernandez-Granda. Super-resolution from noisy data.Journal of Fourier Analysis and Applications, 19(6):1229–1254, August 2013

    work page 2013

  48. [60]

    Cand` es and Carlos Fernandez-Granda

    Emmanuel J. Cand` es and Carlos Fernandez-Granda. Towards a mathematical theory of super-resolution.Communications on Pure and Applied Mathematics, 67(6):906–956, 2014

    work page 2014

  49. [61]

    Statistics of extreme events in coarse-scale cli- mate simulations via machine learning correction operators trained on nudged datasets.arXiv preprint arXiv:2304.02117, 2023

    Alexis-Tzianni Charalampopoulos, Shuai Zhang, Bryce Harrop, Lai-yung Ruby Leung, and Themistoklis Sapsis. Statistics of extreme events in coarse-scale cli- mate simulations via machine learning correction operators trained on nudged datasets.arXiv preprint arXiv:2304.02117, 2023. 90

    work page arXiv 2023

  50. [62]

    Ricky T. Q. Chen, Yulia Rubanova, Jesse Bettencourt, and David K Duvenaud. Neural ordinary differential equations. In S. Bengio, H. Wallach, H. Larochelle, K. Grauman, N. Cesa-Bianchi, and R. Garnett, editors,Advances in Neural Information Processing Systems, volume 31. Curran Associates, Inc., 2018

    work page 2018

  51. [63]

    Ilvr: Conditioning method for denoising diffusion probabilistic models,

    Jooyoung Choi, Sungwon Kim, Yonghyun Jeong, Youngjune Gwon, and Sungroh Yoon. Ilvr: Conditioning method for denoising diffusion probabilistic models. arXiv preprint arXiv:2108.02938, 2021

    work page arXiv 2021

  52. [64]

    Joint distribution optimal transportation for domain adaptation.Advances in neural information processing systems, 30, 2017

    Nicolas Courty, R´ emi Flamary, Amaury Habrard, and Alain Rakotomamonjy. Joint distribution optimal transportation for domain adaptation.Advances in neural information processing systems, 30, 2017

    work page 2017

  53. [65]

    Sinkhorn distances: Lightspeed computation of optimal transport

    Marco Cuturi. Sinkhorn distances: Lightspeed computation of optimal transport. Advances in neural information processing systems, 26, 2013

    work page 2013

  54. [66]

    The community earth system model version 2 (CESM2).Journal of Advances in Modeling Earth Systems, 12(2):e2019MS001916, 2020

    G Danabasoglu, J-F Lamarque, J Bacmeister, D A Bailey, A K DuVivier, J Edwards, L K Emmons, J Fasullo, R Garcia, A Gettelman, C Hannay, M M Holland, W G Large, P H Lauritzen, D M Lawrence, J T M Lenaerts, K Lindsay, W H Lipscomb, M J Mills, R Neale, K W Oleson, B Otto-Bliesner, A S Phillips, W Sacks, S Tilmes, L van Kampenhout, M Vertenstein, A Bertini, J...

    work page 2020

  55. [67]

    Diffusion models beat gans on image synthesis.Advances in Neural Information Processing Systems, 34:8780–8794, 2021

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat gans on image synthesis.Advances in Neural Information Processing Systems, 34:8780–8794, 2021

    work page 2021

  56. [68]

    Evaluating the stationarity assumption in statistically downscaled climate projections: is past performance an indicator of future results?Climatic Change, 135(3-4):395–408, 2016

    Keith W Dixon, John R Lanzante, Mary J Nath, Katharine Hayhoe, Anne Stoner, Aparna Radhakrishnan, Venkat Balaji, and Carlos F Gaitan. Evaluating the stationarity assumption in statistically downscaled climate projections: is past performance an indicator of future results?Climatic Change, 135(3-4):395–408, 2016

    work page 2016

  57. [69]

    Tweedie’s formula and selection bias.Journal of the American Statistical Association, 106(496):1602–1614, 2011

    Bradley Efron. Tweedie’s formula and selection bias.Journal of the American Statistical Association, 106(496):1602–1614, 2011

    work page 2011

  58. [70]

    Response of global tropical cyclone activity to increasing CO2: Results from downscaling CMIP6 models.Journal of Climate, 34(1):57 – 70, 2021

    Kerry Emanuel. Response of global tropical cyclone activity to increasing CO2: Results from downscaling CMIP6 models.Journal of Climate, 34(1):57 – 70, 2021

    work page 2021

  59. [71]

    Overview of the coupled model intercomparison project phase 6 (CMIP6) experimental design and organization

    Veronika Eyring, Sandrine Bony, Gerald A Meehl, Catherine A Senior, Bjorn Stevens, Ronald J Stouffer, and Karl E Taylor. Overview of the coupled model intercomparison project phase 6 (CMIP6) experimental design and organization. Geoscientific model development, 9(5):1937–1958, 26 May 2016. 91

    work page 1937

  60. [72]

    An overview of the E3SM version 2 large ensemble and comparison to other E3SM and CESM large ensembles.Earth system dynamics, 15(2):367–386, 8 April 2024

    John T Fasullo, Jean-Christophe Golaz, Julie M Caron, Nan Rosenbloom, Ger- ald A Meehl, Warren Strand, Sasha Glanville, Samantha Stevenson, Maria Molina, Christine A Shields, Chengzhu Zhang, James Benedict, Hailong Wang, and Tony Bartoletti. An overview of the E3SM version 2 large ensemble and comparison to other E3SM and CESM large ensembles.Earth system...

    work page 2024

  61. [73]

    User-defined event sampling and uncertainty quantifi- cation in diffusion models for physical dynamical systems

    Marc Anton Finzi, Anudhyan Boral, Andrew Gordon Wilson, Fei Sha, and Leonardo Zepeda-N´ u˜ nez. User-defined event sampling and uncertainty quantifi- cation in diffusion models for physical dynamical systems. In Andreas Krause, Emma Brunskill, Kyunghyun Cho, Barbara Engelhardt, Sivan Sabato, and Jonathan Scarlett, editors,Proceedings of the 40th Internati...

    work page 2023

  62. [74]

    Optimal transport for domain adaptation.IEEE Trans

    R Flamary, N Courty, D Tuia, and A Rakotomamonjy. Optimal transport for domain adaptation.IEEE Trans. Pattern Anal. Mach. Intell, 1, 2016

    work page 2016

  63. [75]

    Garner, Robert E

    Andra J. Garner, Robert E. Kopp, and Benjamin P. Horton. Evolving tropi- cal cyclone tracks in the north atlantic in a warming climate.Earth’s Future, 9(12):e2021EF002326, 2021. e2021EF002326 2021EF002326

    work page 2021

  64. [76]

    Climalign: Unsupervised statistical downscaling of climate variables via normalizing flows

    Brian Groenke, Luke Madaus, and Claire Monteleoni. Climalign: Unsupervised statistical downscaling of climate variables via normalizing flows. InProceedings of the 10th International Conference on Climate Informatics, CI2020, page 60–66, New York, NY, USA, 2021. Association for Computing Machinery

    work page 2021

  65. [77]

    Alignflow: Cycle consistent learning from multiple domains via normalizing flows, 2019

    Aditya Grover, Christopher Chute, Rui Shu, Zhangjie Cao, and Stefano Ermon. Alignflow: Cycle consistent learning from multiple domains via normalizing flows, 2019

    work page 2019

  66. [78]

    Wuebbles

    Katharine Hayhoe, Ian Scott-Fleming, Anne Stoner, and Donald J. Wuebbles. STAR-ESDM: A generalizable approach to generating high-resolution climate projections through signal decomposition.Earth’s Future, 12(7):e2023EF004107, 2024

    work page 2024

  67. [79]

    Hans Hersbach, Bill Bell, Paul Berrisford, Shoji Hirahara, Andr´ as Hor´ anyi, Joaqu´ ın Mu˜ noz-Sabater, Julien Nicolas, Carole Peubey, Raluca Radu, Dinand Schepers, Adrian Simmons, Cornel Soci, Saleh Abdalla, Xavier Abellan, Gian- paolo Balsamo, Peter Bechtold, Gionata Biavati, Jean Bidlot, Massimo Bonavita, Giovanna De Chiara, Per Dahlgren, Dick Dee, M...

    work page 1999

  68. [80]

    Ouarda, and Andr´ e St-Hilaire

    Masoud Hessami, Philippe Gachon, Taha B.M.J. Ouarda, and Andr´ e St-Hilaire. Automated regression-based statistical downscaling tool.Environmental Mod- elling and Software, 23(6):813–834, 2008

    work page 2008

  69. [81]

    Classifier-free diffusion guidance, 2022

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance, 2022

    work page 2022

  70. [82]

    Hua and T

    Y. Hua and T. K. Sarkar. On SVD for estimating generalized eigenvalues of singular matrix pencil in noise.IEEE Trans. Sig. Proc., 39(4):892–900, April 1991

    work page 1991

  71. [83]

    Diffenbaugh, and Eran Bendavid

    Renzhi Jing, Jianxiong Gao, Yunuo Cai, Dazhi Xi, Yinda Zhang, Yanwei Fu, Kerry Emanuel, Noah S. Diffenbaugh, and Eran Bendavid. TC-GEN: Data-driven tropical cyclone downscaling using machine learning-based high- resolution weather model.Journal of Advances in Modeling Earth Systems, 16(10):e2023MS004203, 2024. e2023MS004203 2023MS004203

    work page 2024

  72. [84]

    Elucidating the design space of diffusion-based generative models.Advances in neural information processing systems, 35:26565–26577, 2022

    Tero Karras, Miika Aittala, Timo Aila, and Samuli Laine. Elucidating the design space of diffusion-based generative models.Advances in neural information processing systems, 35:26565–26577, 2022

    work page 2022

  73. [85]

    Camargo, Johnny C

    Thomas Knutson, Suzana J. Camargo, Johnny C. L. Chan, Kerry Emanuel, Chang-Hoi Ho, James Kossin, Mrutyunjay Mohapatra, Masaki Satoh, Masato Sugi, Kevin Walsh, and Liguang Wu. Tropical cyclones and climate change assess- ment: Part ii: Projected response to anthropogenic warming.Bulletin of the American Meteorological Society, 101(3):E303 – E322, 2020

    work page 2020

  74. [86]

    The poleward migra- tion of the location of tropical cyclone maximum intensity.Nature, 509:349–352, 2014

    James P Kossin, Kerry A Emanuel, and Gabriel A Vecchi. The poleward migra- tion of the location of tropical cyclone maximum intensity.Nature, 509:349–352, 2014

    work page 2014

  75. [87]

    Srdiff: Single image super-resolution with diffusion probabilistic models.Neurocomputing, 479:47–59, 2022

    Haoying Li, Yifan Yang, Meng Chang, Shiqi Chen, Huajun Feng, Zhihai Xu, Qi Li, and Yueting Chen. Srdiff: Single image super-resolution with diffusion probabilistic models.Neurocomputing, 479:47–59, 2022

    work page 2022

  76. [88]

    Rectified Flow: A Marginal Preserving Approach to Optimal Transport

    Qiang Liu. Rectified flow: A marginal preserving approach to optimal transport. arXiv preprint arXiv:2209.14577, 2022

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  77. [89]

    Flow straight and fast: Learning to generate and transfer data with rectified flow

    Xingchao Liu, Chengyue Gong, and Qiang Liu. Flow straight and fast: Learning to generate and transfer data with rectified flow. InThe Eleventh International Conference on Learning Representations, 2023

    work page 2023

  78. [90]

    Dynamical-generative downscaling of climate model ensembles.Proceedings of the National Academy of Sciences, 122:e2420288122, 4

    Ignacio Lopez-Gomez, Zhong Yi Wan, Leonardo Zepeda-N´ u˜ nez, Tapio Schneider, John Anderson, and Fei Sha. Dynamical-generative downscaling of climate model ensembles.Proceedings of the National Academy of Sciences, 122:e2420288122, 4

    work page

  79. [91]

    doi: 10.1073/pnas.2420288122. 93

    work page doi:10.1073/pnas.2420288122

  80. [92]

    SRFlow: Learning the super-resolution space with normalizing flow

    Andreas Lugmayr, Martin Danelljan, Luc Van Gool, and Radu Timofte. SRFlow: Learning the super-resolution space with normalizing flow. InECCV, 2020

    work page 2020

Showing first 80 references.