mlsanityData to 7 Oct 2026

Residual Corrective Diffusion Modeling for Km-scale Atmospheric Downscaling

arXiv 2309.15214 ↗First posted 24 Sep 2023cs.LG physics.ao-phcited by 103

On the map: Diffusion models → Flow matching, time series and science

Abstract (arXiv)

The state of the art for physical hazard prediction from weather and climate requires expensive km-scale numerical simulations driven by coarser resolution global inputs. Here, a generative diffusion architecture is explored for downscaling such global inputs to km-scale, as a cost-effective machine learning alternative. The model is trained to predict 2km data from a regional weather model over Taiwan, conditioned on a 25km global reanalysis. To address the large resolution ratio, different physics involved at different scales and prediction of channels beyond those in the input data, we employ a two-step approach where a UNet predicts the mean and a corrector diffusion (CorrDiff) model predicts the residual. CorrDiff exhibits encouraging skill in bulk MAE and CRPS scores. The predicted spectra and distributions from CorrDiff faithfully recover important power law relationships in the target data. Case studies of coherent weather phenomena show that CorrDiff can help sharpen wind and temperature gradients that co-locate with intense rainfall in cold front, and can help intensify typhoons and synthesize rain band structures. Calibration of model uncertainty remains challenging. The prospect of unifying methods like CorrDiff with coarser resolution global weather models implies a potential for global-to-regional multi-scale machine learning simulation.

Key facts. Cited by 103 papers in the map (data through 7 Oct 2026). Placed in Flow matching, time series and science, one of 14 directions in Diffusion models. Builds on 34 of its 79 references that are in the map. First posted on arXiv 24 Sep 2023.

Builds on

34 of 79 references.

Closest ancestors · 3

Same direction, newest first

  1. 2023SEEDS: Emulation of Weather Forecast Ensembles with Diffusion Models24 Jun
  2. 2023Latent diffusion models for generative precipitation nowcasting with accurate uncertainty quantification25 Apr
  3. 2023Diffusion Models for High-Resolution Solar Forecasts1 Feb

Other references · 31

  1. 2017Attention Is All You Needcited by 59,932
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  4. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  5. 2020Denoising Diffusion Implicit Modelscited by 9,655
  6. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  7. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,185
  8. 2020End-to-End Object Detection with Transformerscited by 7,384
  9. 2022Classifier-Free Diffusion Guidancecited by 5,717
  10. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,067
  11. 2022Elucidating the Design Space of Diffusion-Based Generative Modelscited by 2,994
  12. 2016Improved Techniques for Training GANscited by 2,938
19 more
  1. 2020Improved Techniques for Training Score-Based Generative Modelscited by 1,036
  2. 2022Diffusion Models in Vision: A Surveycited by 1,008
  3. 2022FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operatorscited by 748
  4. 2022On Distillation of Guided Diffusion Modelscited by 721
  5. 2021Score-based Generative Modeling in Latent Spacecited by 634
  6. 2021Tackling the Generative Learning Trilemma with Denoising Diffusion GANscited by 515
  7. 2023Simple diffusion: End-to-end diffusion for high resolution imagescited by 413
  8. 2021Skillful Precipitation Nowcasting using Deep Generative Models of Radarcited by 308
  9. 2023FuXi: A cascade machine learning forecasting system for 15-day global weather forecastcited by 237
  10. 2023Spherical Fourier Neural Operators: Learning Stable Dynamics on the Spherecited by 225
  11. 2021Conditional Image Generation with Score-Based Diffusion Modelscited by 195
  12. 2022Fast Sampling of Diffusion Models via Operator Learningcited by 173
  13. 2020Stochastic Super-Resolution for Downscaling Time-Evolving Atmospheric Fields with a Generative Adversarial Networkcited by 68
  14. 2022A Generative Deep Learning Approach to Stochastic Downscaling of Precipitation Forecastscited by 64
  15. 2022Increasing the accuracy and resolution of precipitation forecasts using deep generative modelscited by 35
  16. 2018DeepDownscale: a Deep Learning Strategy for High-Resolution Weather Forecastcited by 28
  17. 2023Unpaired Downscaling of Fluid Flows with Diffusion Bridgescited by 21
  18. 2022Machine learning emulation of a local-scale UK climate modelcited by 20
  19. 2020TRU-NET: A Deep Learning Approach to High Resolution Prediction of Rainfallcited by 19

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 1; 2024: 16; 2025: 42; 2026: 44.

103 papers.

Sort:

By area:

  1. 2024DiffDA: a Diffusion Model for Weather-scale Data Assimilationcited by 59
  2. 2024Heavy-Tailed Diffusion Modelscited by 34
  3. 2023Foundation Models for Weather and Climate Data Understanding: A Comprehensive Surveycited by 32
  4. 2024Kilometer-Scale Convection Allowing Model Emulation using Generative Diffusion Modelingcited by 32
  5. 2024Probabilistic Weather Forecasting with Hierarchical Graph Neural Networkscited by 29
  6. 2024Continuous Ensemble Weather Forecasting with Diffusion modelscited by 27
  7. 2024Probabilistic Emulation of a Global Climate Model with Spherical DYffusioncited by 21
  8. 2024Dynamical-generative downscaling of climate model ensemblescited by 19
  9. 2024Machine learning emulation of precipitation from km-scale UK regional climate simulations using a diffusion modelcited by 15
  10. 2024Can AI be enabled to dynamical downscaling? A Latent Diffusion Model to mimic km-scale COSMO5.0CLM9 simulationscited by 14
  11. 2025Swift: An Autoregressive Consistency Model for Efficient Weather Forecastingcited by 13
  12. 2025Elucidated Rolling Diffusion Models for Probabilistic Forecasting of Complex Dynamicscited by 12
38 more
  1. 2025EnScale: Temporally-consistent multivariate generative downscaling via proper scoring rulescited by 10
  2. 2024DiffObs: Generative Diffusion for Global Forecasting of Satellite Observationscited by 9
  3. 2024Quantum State Generation with Structure-Preserving Diffusion Modelcited by 8
  4. 2025Dynamical Diffusion: Learning Temporal Dynamics with Diffusion Modelscited by 8
  5. 2025Diff-SPORT: Diffusion-based Sensor Placement Optimization and Reconstruction of Turbulent flows in urban environmentscited by 8
  6. 2025Multimodal Atmospheric Super-Resolution With Deep Generative Modelscited by 8
  7. 2025AERIS: Argonne Earth Systems Model for Reliable and Skillful Predictionscited by 8
  8. 2024Observation-Guided Meteorological Field Downscaling at Station Scale: A Benchmark and a New Methodcited by 7
  9. 2024CaFA: Global Weather Forecasting with Factorized Attention on Spherecited by 6
  10. 2024SRViT: Vision Transformers for Estimating Radar Reflectivity from Satellite Observations at Scalecited by 6
  11. 2024Stochastic Flow Matching for Resolving Small-Scale Physicscited by 5
  12. 2026Universal Diffusion-Based Probabilistic Downscalingcited by 5
  13. 2024Downscaling Extreme Precipitation with Wasserstein Regularized Diffusioncited by 4
  14. 2025When and how can inexact generative models still sample from the data manifold?cited by 4
  15. 2025MAUSAM: An Observations-focused assessment of Global AI Weather Prediction Models During the South Asian Monsooncited by 4
  16. 2026FloeNet: A mass-conserving global sea ice emulator that generalizes across climatescited by 4
  17. 2025DiffScale: Continuous Downscaling and Bias Correction of Subseasonal Wind Speed Forecasts using Diffusion Modelscited by 3
  18. 2025How to use score-based diffusion in earth system science: A satellite nowcasting examplecited by 3
  19. 2025MODS: Multi-source Observations Conditional Diffusion Model for Meteorological State Downscalingcited by 3
  20. 2025Efficient Kilometer-Scale Precipitation Downscaling with Conditional Wavelet Diffusioncited by 3
  21. 2025Warm Starts Accelerate Conditional Diffusioncited by 3
  22. 2025Score-based generative emulation of impact-relevant Earth system model outputscited by 3
  23. 2025Assessing the Geographic Generalization and Physical Consistency of Generative Models for Climate Downscalingcited by 3
  24. 2025Generating realistic global precipitation fields from modelled atmospheric circulationcited by 2
  25. 2025Guided Unconditional and Conditional Generative Models for Super-Resolution and Inference of Quasi-Geostrophic Turbulencecited by 2
  26. 2025Reframing Generative Models for Physical Systems using Stochastic Interpolantscited by 2
  27. 2025RainDiff: End-to-end Precipitation Nowcasting Via Token-wise Attention Diffusioncited by 2
  28. 2025A Diffusion-Based Framework for High-Resolution Precipitation Forecasting over CONUScited by 2
  29. 2026Physics-Constrained Adaptive Flow Matching for Climate Downscalingcited by 2
  30. 2025Controlling Ensemble Variance in Diffusion Models: An Application for Reanalyses Downscalingcited by 1
  31. 2025Probabilistic Super-Resolution for Urban Micrometeorology via a Schr"odinger Bridgecited by 1
  32. 2025Sparse Local Implicit Image Function for sub-km Weather Downscalingcited by 1
  33. 2025A PDE-Informed Latent Diffusion Model for 2-m Temperature Downscalingcited by 1
  34. 2025China Regional 3km Downscaling Based on Residual Corrective Diffusion Modelcited by 1
  35. 2025Learning vertical coordinates via automatic differentiation of a dynamical corecited by 1
  36. 2025Field-Space Attention for Structure-Preserving Earth System Transformerscited by 1
  37. 2025STLDM: Spatio-Temporal Latent Diffusion Model for Precipitation Nowcastingcited by 1
  38. 2026Super-Resolving Coarse-Resolution Weather Forecasts With Flow Matchingcited by 1

50 most cited of 103 All 103 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024TAUDiff: Highly efficient kilometer-scale downscaling using generative diffusion models
  2. 2024A Generative Framework for Probabilistic, Spatiotemporally Coherent Downscaling of Climate Simulation
  3. 2026Downscaling weather forecasts from Low- to High-Resolution with Diffusion Models
  4. 2024Conditional diffusion models for downscaling and bias correction of Earth system model precipitation
  5. 2026Stress-Testing Dynamical and Generative Downscaling Using Subseasonal Extreme Precipitation Forecasts
  6. 2024Enhancing Weather Predictions: Super-Resolution via Deep Diffusion Models
  7. 2024Diffusion Model-based Probabilistic Downscaling for 180-year East Asian Climate Reconstruction
  8. 2023Generating High-Resolution Regional Precipitation Using Conditional Diffusion Model