mlsanityData to 7 Oct 2026

DiffDA: a Diffusion Model for Weather-scale Data Assimilation

arXiv 2401.05932 ↗First posted 11 Jan 2024cs.CE cs.AIcited by 59

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

Abstract (arXiv)

The generation of initial conditions via accurate data assimilation is crucial for weather forecasting and climate modeling. We propose DiffDA as a denoising diffusion model capable of assimilating atmospheric variables using predicted states and sparse observations. Acknowledging the similarity between a weather forecast model and a denoising diffusion model dedicated to weather applications, we adapt the pretrained GraphCast neural network as the backbone of the diffusion model. Through experiments based on simulated observations from the ERA5 reanalysis dataset, our method can produce assimilated global atmospheric data consistent with observations at 0.25 deg (~30km) resolution globally. This marks the highest resolution achieved by ML data assimilation models. The experiments also show that the initial conditions assimilated from sparse observations (less than 0.96% of gridded data) and 48-hour forecast can be used for forecast models with a loss of lead time of at most 24 hours compared to initial conditions from state-of-the-art data assimilation in ERA5. This enables the application of the method to real-world applications, such as creating reanalysis datasets with autoregressive data assimilation.

Key facts. Cited by 59 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 16 of its 28 references that are in the map. First posted on arXiv 11 Jan 2024.

Builds on

16 of 28 references.

Closest ancestors · 2

Same direction, newest first

  1. 2023Residual Corrective Diffusion Modeling for Km-scale Atmospheric Downscaling24 Sep
  2. 2023SEEDS: Emulation of Weather Forecast Ensembles with Diffusion Models24 Jun

Other references · 14

  1. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  2. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  3. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,322
  4. 2017Decoupled Weight Decay Regularizationcited by 1,722
  5. 2022RePaint: Inpainting using Denoising Diffusion Probabilistic Modelscited by 1,419
  6. 2022Diffusion Posterior Sampling for General Noisy Inverse Problemscited by 1,375
  7. 2022FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operatorscited by 748
  8. 2020WeatherBench: A benchmark dataset for data-driven weather forecastingcited by 272
  9. 2023WeatherBench 2: A benchmark for the next generation of data-driven global weather modelscited by 227
  10. 2023GenCast: Diffusion-based ensemble forecasting for medium-range weathercited by 169
  11. 2023Score-based Data Assimilationcited by 91
  12. 2023Deep Learning for Day Forecasts from Sparse Observationscited by 70
2 more
  1. 2022ENS-10: A Dataset For Post-Processing Ensemble Weather Forecastscited by 13
  2. 2023Score-based Data Assimilation for a Two-Layer Quasi-Geostrophic Modelcited by 11

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 20; 2025: 26; 2026: 13.

59 papers.

Sort:

By area:

  1. 2024A Survey on Diffusion Models for Time Series and Spatio-Temporal Datacited by 66
  2. 2024On conditional diffusion models for PDE simulationscited by 56
  3. 2024Kilometer-Scale Convection Allowing Model Emulation using Generative Diffusion Modelingcited by 32
  4. 2024Deep Generative Data Assimilation in Multimodal Settingcited by 28
  5. 2024Generative Data Assimilation of Sparse Weather Station Observations at Kilometer Scalescited by 27
  6. 2024Generative Diffusion-based Downscaling for Climatecited by 25
  7. 2024Latent-EnSF: A Latent Ensemble Score Filter for High-Dimensional Data Assimilation with Sparse Observation Datacited by 20
  8. 2024Data driven weather forecasts trained and initialised directly from observationscited by 19
  9. 2024Fuxi-DA: A Generalized Deep Learning Data Assimilation Framework for Assimilating Satellite Observationscited by 18
  10. 2025Appa: Bending Weather Dynamics with Latent Diffusion Models for Global Data Assimilationcited by 17
  11. 2024Spatially-Aware Diffusion Models with Cross-Attention for Global Field Reconstruction with Sparse Observationscited by 13
  12. 2025FlowDAS: A Stochastic Interpolant-based Framework for Data Assimilationcited by 13
38 more
  1. 2024WeatherGFM: Learning A Weather Generalist Foundation Model via In-context Learningcited by 12
  2. 2025Elucidated Rolling Diffusion Models for Probabilistic Forecasting of Complex Dynamicscited by 12
  3. 2025Flow Matching for Efficient and Scalable Data Assimilationcited by 11
  4. 2024Data Assimilation with Machine Learning Surrogate Models: A Case Study with FourCastNetcited by 10
  5. 2024LD-EnSF: Synergizing Latent Dynamics with Ensemble Score Filters for Fast Data Assimilation with Sparse Observationscited by 10
  6. 2025PhyDA: Physics-Guided Diffusion Models for Data Assimilation in Atmospheric Systemscited by 9
  7. 2024FNP: Fourier Neural Processes for Arbitrary-Resolution Data Assimilationcited by 8
  8. 2024WeatherReal: A Benchmark Based on In-Situ Observations for Evaluating Weather Modelscited by 8
  9. 2024Exploring the design space of deep-learning-based weather forecasting systemscited by 8
  10. 2025Multimodal Atmospheric Super-Resolution With Deep Generative Modelscited by 8
  11. 2025Strictly Constrained Generative Modeling via Split Augmented Langevin Samplingcited by 6
  12. 2025LaDCast: A Latent Diffusion Model for Medium-Range Ensemble Weather Forecastingcited by 6
  13. 2024A Comprehensive Evaluation of Generative Models in Calorimeter Shower Simulationcited by 5
  14. 2025Kolmogorov Arnold Neural Interpolator for Downscaling and Correcting Meteorological Fields from In-Situ Observationscited by 5
  15. 2025Ensemble Kalman filter in latent space using a variational autoencoder paircited by 5
  16. 2025Align-DA: Align Score-based Atmospheric Data Assimilation with Multiple Preferencescited by 5
  17. 2025Training-Free Data Assimilation with GenCastcited by 5
  18. 2025Adversarial Observations in Weather Forecastingcited by 4
  19. 2024Benchmarking AI-based data assimilation to advance data-driven global weather forecastingcited by 3
  20. 2025Task-Adaptive Parameter-Efficient Fine-Tuning for Weather Foundation Modelscited by 3
  21. 2025DAISI: Data Assimilation with Inverse Sampling using Stochastic Interpolantscited by 3
  22. 2024Aardvark weather: end-to-end data-driven weather forecastingcited by 2
  23. 2024VAE-Var: Variational-Autoencoder-Enhanced Variational Assimilationcited by 2
  24. 2026Amortized Filtering and Smoothing with Conditional Normalizing Flowscited by 2
  25. 2026Training-Free Bayesian Filtering with Generative Emulatorscited by 2
  26. 2025DAWP: A framework for global observation forecasting via Data Assimilation and Weather Prediction in satellite observation spacecited by 1
  27. 2025Sparse Local Implicit Image Function for sub-km Weather Downscalingcited by 1
  28. 2025DAMBench: A Multi-Modal Benchmark for Deep Learning-based Atmospheric Data Assimilationcited by 1
  29. 2025Advancing Ocean State Estimation with efficient and scalable AIcited by 1
  30. 2026Accurate and Efficient Hybrid-Ensemble Atmospheric Data Assimilation in Latent Space with Uncertainty Quantificationcited by 1
  31. 2026ADDA: a Modular Framework for Representing, Simulating and Assimilating Dynamics with End-to-end Differentiabilitycited by 1
  32. 2025Generative diffusion model surrogates for mechanistic agent-based biological modelscited by 0
  33. 2025Probabilistic Spatial Interpolation of Sparse Data using Diffusion Modelscited by 0
  34. 2025ReconMOST: Multi-Layer Sea Temperature Reconstruction with Observations-Guided Diffusioncited by 0
  35. 2025PnP-DA: Towards Principled Plug-and-Play Integration of Variational Data Assimilation and Generative Modelscited by 0
  36. 2025On the Effect of Neural Field Reparameterization for 4DVARcited by 0
  37. 2025Diffusion-Based, Data-Assimilation-Enabled Super-Resolution of Hub-height Windscited by 0
  38. 2025Control-Augmented Autoregressive Diffusion for Data Assimilationcited by 0

50 most cited of 59 All 59 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024DiffObs: Generative Diffusion for Global Forecasting of Satellite Observations
  2. 2024Enhancing Weather Predictions: Super-Resolution via Deep Diffusion Models
  3. 2026Closed-form conditional diffusion models for data assimilation
  4. 2025Data-Efficient Ensemble Weather Forecasting with Diffusion Models
  5. 2025DEF: Diffusion-augmented Ensemble Forecasting
  6. 2026PuYun-LDM: A Latent Diffusion Model for High-Resolution Ensemble Weather Forecasts
  7. 2026DAWIS: Data Assimilation with Windowed Inverse Sampling via Multitask Interpolants
  8. 2026LD-EnFF: Latent-Dynamics Ensemble Flow Filtering for Data Assimilation with Sparse Observations