Pith. sign in

REVIEW 4 major objections 6 minor 86 references

Fusion of multi-source precipitation records via coordinate-based generative model

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a coordinate-based diffusion prior trained on reanalysis, satellite retrievals, and sparse rain gauges produces precipitation estimates that outperform any single source, and that the same prior corrects biases in…

desk verdict Solid method, shaky evaluation: the ground truth is the same merged gauge-satellite product used to fine-tune the model. read the letter →

arxiv 2506.11698 v2 pith:4TCQNG4O submitted 2025-06-13 physics.ao-ph

Sencan Sun , Congyi Nai , Baoxiang Pan , Wentao Li , Lu Li , Xin Li , Efi Foufoula-Georgiou , Yanluan Lin This is my paper · ORCID
classification physics.ao-ph
keywords precipitationdatafusiondiffusionmodelcoordinate-basedrepresentationBayesianinferencebiascorrectiondownscalingzero-shotgeneralizationgauge–satellitemerging
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

PRIMER is a generative model that treats precipitation as a continuous spatial field rather than a fixed grid, allowing dense satellite and reanalysis grids and sparse rain-gauge point observations to fuse at their native sampling structures without destructive interpolation. The paper's central claim is that a diffusion model trained in two stages—first on gridded ERA5 and IMERG data, then fine-tuned on gauge observations—yields a prior distribution that outperforms priors built from any single source. Conditioning this prior on an existing precipitation product through posterior sampling corrects biases, downscales coarse fields, and restores realistic high-frequency structure, with statistically significant error reductions at most evaluated stations and improved spatial coherence. The same prior also corrects biases in operational HRES forecasts it never saw during training, which the authors present as evidence of zero-shot generalization. A sympathetic reader should care because precipitation datasets currently disagree with each other by as much as the signal itself, and PRIMER offers a principled Bayesian route to combining imperfect sources into a single full-coverage estimate with quantified uncertainty.

What carries the argument

The load-bearing object is the coordinate-based diffusion prior, a score-based generative model defined over the Hilbert space $L^2([0,1]^n \to \mathbb{R}^d)$, so precipitation is represented as a continuous function of spatial coordinates rather than a fixed-resolution tensor. The forward process mollifies white noise with a Gaussian kernel, implemented via Fourier transforms and a Wiener filter, to keep trajectories in $L^2$; the reverse process uses a hybrid network of SparseConvResBlocks and a U-Net, modulated by a source embedding that tells the model which dataset a sample came from. Two-stage training first learns large-scale climatology from ERA5 and IMERG, then fine-tunes with gauge observations through a weighted loss that preserves the gridded priors while grounding the manifold in local measurements. Posterior inference is carried out with inpainting and SDEdit, which balance the learned prior against the conditioning observations, and a noise-level parameter controls how much the posterior may deviate from the input field.

What would settle it

Re-evaluate PRIMER's posterior outputs at raw automatic weather stations that were excluded from the merged product and from all training data, in a region with dense independent gauges, and compare mean absolute error and CRPS against raw ERA5, raw IMERG, and the original merged product; if the advantage over the raw products shrinks or vanishes under this independent comparison, the claimed bias correction is largely an artifact of the merged ground truth.

Watch

Extended reading notes

Core claim

The central discovery claimed is that an informative prior over precipitation fields can be constructed from imperfect, heterogeneous records by exploiting the spectral progression of diffusion models: as Gaussian noise gradually corrupts the target, the model learns low-frequency structure first and high-frequency details later. PRIMER therefore learns conditional source-specific priors for ERA5 and IMERG, then refines them with gauge observations to obtain a gauge-calibrated prior, using source embeddings and shared weights so each dataset contributes at its natural scale. Under posterior sampling conditioned on an existing product, the gauge-calibrated prior consistently lowers mean absolute error and continuous ranked probability score relative to the original ERA5 or IMERG fields, improves the tail of the intensity distribution, and makes the spatial lag-correlation structure closer to that of gauge observations. It also corrects biases in HRES forecasts without retraining. In the authors' framing, this turns the heterogeneity of imperfect data from a limitation into a strength.

Load-bearing premise

The load-bearing premise is that the gridded gauge–satellite merged product used as ground truth is itself trustworthy; because that product provides both the fine-tuning target and the evaluation target, any systematic bias in it would be treated as truth, and the paper's reported error reductions could reflect alignment to that bias rather than to real precipitation.

Editorial extensions

If this is right

  • Posterior sampling from the calibrated prior yields ensemble precipitation fields with a spread, so users can obtain both a bias-corrected mean and an uncertainty estimate for risk assessment.
  • Because the prior is source-agnostic after fine-tuning, the same trained model can downscale coarse reanalysis, correct satellite retrieval biases, and correct operational deterministic forecasts, as demonstrated on HRES.
  • Additional gauge observations can be injected during sampling, so the framework can act as a lightweight data-assimilation tool without retraining.
  • The two-stage training recipe—large noisy gridded data for structure, sparse accurate data for local refinement—is transferable to other Earth-system variables with similar observation trade-offs.
  • If the learned prior reproduces reference climatology and spectra, it can be used to generate physically plausible precipitation realizations for studies of extremes and for evaluating risk under rare event scenarios.

Reading between the lines

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

  • Editorial inference: If the gauge-calibrated prior really captures shared climatology rather than the idiosyncrasies of the merged product, the same architecture could serve as a universal prior for many satellite, reanalysis, and forecast products beyond ERA5, IMERG, and HRES; this is a testable extension, not something the paper demonstrates.
  • Editorial inference: A fair stress test would be to fine-tune on one region's gauges and evaluate on another region with an independent national network, to see whether improvements come from generalizable physics or from memorizing the merged product's biases.
  • Editorial inference: The spectral view the authors adopt suggests a quantitative prediction: the largest gains from gauge fine-tuning should appear at wavelengths where gridded products are structurally deficient, so applying PRIMER to a dataset with different spectral error characteristics should produce a different improvement profile.
  • Editorial inference: If zero-shot correction of HRES is real, the framework could be deployed as an operational post-processing layer for numerical weather prediction ensembles, but the practical value would depend on the gauge network density available in the region of interest.
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

4 major / 6 minor

Summary. The paper introduces PRIMER, a coordinate-based diffusion model for fusing precipitation records from gridded reanalysis (ERA5), satellite retrievals (IMERG), and sparse gauge-type observations. The method trains in two stages: first, it learns separate priors for ERA5 and IMERG; second, it fine-tunes on a merged gauge-satellite product to obtain an updated prior P*(x). Posterior sampling with this prior is then used for bias correction, downscaling, and zero-shot correction of unseen operational forecasts (HRES). The authors report improved MAE, CRPS, intensity distributions, and spatial coherence against what they call gauge observations, based on 150 selected precipitation events from 2016.

Significance. If the central claims are substantiated, PRIMER would be a valuable contribution: the coordinate-based formulation is a principled way to handle irregular gauge locations without destructive interpolation, the two-stage training idea is plausible, and the zero-shot generalization to HRES is an interesting demonstration. The paper also provides useful methodological details, including pseudo-code, a spectral diagnostic, and a theoretical justification in the supplement. The main weakness is that the evaluation target is not independent of the fine-tuning data: Section 4.6 states that both fine-tuning and evaluation use grid cells from the same Shen et al. (2014) gauge-satellite merged product, so the quantitative improvements in Figs. 4 and 5 may partly reflect fitting the product's own biases. The stress-test concern therefore lands, and the current evidence does not support the abstract's claim of statistically significant corrections against independent gauge observations.

major comments (4)
  1. [Section 4.6 and Fig. C5] The evaluation target is not independent of the fine-tuning data. The text states that fine-tuning uses grid cells from the Shen et al. (2014) gauge-satellite merged product with at least one assimilated AWS, and that evaluation uses cells from the same product with at least four AWS. Because this product is a gridded analysis that already blends satellite retrievals with gauge information, the updated prior P*(x) is calibrated to the same analysis used for scoring. The temporal split (2015/2017 training versus 2016 testing) does not remove shared product-level biases, so the improved MAE and CRPS in Figs. 4 and 5 may reflect reproducing the merged product's gridded values rather than correcting real precipitation error. The authors should either validate against raw station observations that are not used in training or in the construction of the merged product, or explicitly reframe all headline results as bias correction relative to the Shen et al. analysis rather than to gauges.
  2. [Abstract, Section 1, and Appendix C.2] The statement in the Introduction that evaluations use 'approximately 1,000 independent rain gauges' is misleading given Section 4.6: the evaluation is performed at grid cells of a merged product, not at independent station records. In addition, the 150 test events in Appendix C.2 are selected from the same product's station intensities, specifically the 100 timestamps with the highest individual station intensities and the 50 with the highest average intensity. This concentrates the test set on extreme events and may overstate improvements that are particular to heavy-rain conditions. Please report performance on a random or complete temporal sample from 2016, and quantify how the event-selection criteria affect the headline error reductions.
  3. [Section 4.5 and Figs. 4-5] The abstract claims 'statistically significant error reductions at most stations', but no significance test is described anywhere in the Methods or figure captions. The maps in Figs. 4 and 5 show mean differences, but there is no account of spatial or temporal dependence among stations and events, no confidence intervals, and no multiplicity control. Please add a paired significance test across stations or events (e.g., a bootstrap or permutation test), state the null hypothesis, and report the fraction of stations for which the improvement is significant after accounting for spatial correlation.
  4. [Section 2.3 and SI B.1] The paper's baselines are its own posterior samples from PERA5(x) and PIMERG(x), not established bias-correction or fusion methods such as quantile mapping, CDF matching, or simple gauge interpolation. As a result, the current evaluation does not establish that PRIMER's advantage comes from the generative/Bayesian machinery rather than from the gauge information injected in Stage 2. A comparison against at least one standard non-generative fusion baseline would make the central claim more convincing. In the same vein, the theoretical justification in SI B.1 applies an ambient-diffusion bound that assumes noisy observations of a common Ptrue with known isotropic noise; ERA5 and IMERG are not independent noisy observations of the same target, and the merged product is not clean point truth, so the bound's relevance to the two-stage procedure should be stated more cautiously.
minor comments (6)
  1. [Section 4.2 and Section 4.6] The entity embedding e3 is described as 'gauge observations', but the data used for fine-tuning are grid cells from a merged gauge-satellite product; please use consistent terminology throughout.
  2. [Section 2.2] The sentence 'the ensemble-mean ΔMAE decreases from 0.46 mm/hr for P*(x|OERA5) to 0.14 mm/hr for PERA5(x|OERA5)' appears to have the sign or ordering reversed relative to the definition of ΔMAE in Section 2.3, where positive values indicate improvement of P* over the baseline; please clarify.
  3. [Table B1 and SI B.7] Reporting that batch size and learning rate 'varied between... due to intermittent training interruptions' is not reproducible; please provide the final settings or a precise schedule for all reported experiments.
  4. [Section 4.4] The SDEdit noise level τ is tuned on a single IMERG event (13 June 2016 at 23:00 UTC) and then applied to all experiments; a cross-event sensitivity check, or a rationale for why this choice transfers, would strengthen the results.
  5. [Code availability] The code availability statement says code will be released upon acceptance; an anonymized repository or model checkpoint available during review would help verify the implementation and reproducibility.
  6. [Declarations] Please fix the typo 'Acknowledgemenrs' in the Declarations section.

Circularity Check

2 steps flagged · score 6.0 of 10

Evaluation 'ground truth' is the same gauge-satellite merged product used for Stage 2 fine-tuning, so PRIMER's claimed bias corrections are partly a fit to that product.

  1. fitted input called prediction [Section 4.6 (Data), used for evaluations in Figs. 3-5 and the abstract's ground-truth claim]
    "For fine-tuning, we use dataset from Shen et al. [29], constructed using over 30,000 Automatic Weather Stations (AWS) across China. The gridded dataset has a spatial resolution of 0.1° and a temporal resolution of 1 hour, covering 2015 and 2017 for training, and 2016 for testing. We select grid cells with at least one assimilated AWS observation for training, and use a subset with no fewer than four AWS observations as ground truth for evaluation."

    Stage 2 fine-tuning minimizes L_fine-tuning = α1 L_ERA5 + α2 L_IMERG + α3 L_gauge with α3 = 0.5 against grid cells of the Shen et al. gauge-satellite merged product. The headline evaluation then computes MAE/CRPS at 'gauge locations' drawn from the same product (grid cells with ≥4 AWS). Thus the reported error reductions largely measure how well PRIMER reproduces the merged-product values it was calibrated to reproduce, rather than independent raw gauge truth. The 2016 temporal holdout gives some independence, but because the product's systematic bias field is shared across years, the improvement is partly forced by construction.

  2. other [Abstract and Section 4.6; Appendix C.2]
    "Using gauge observations as ground truth, PRIMER effectively corrects biases in existing datasets... For fine-tuning, we use dataset from Shen et al. [29], constructed using over 30,000 Automatic Weather Stations (AWS) across China... We select grid cells with at least one assimilated AWS observation for training, and use a subset with no fewer than four AWS observations as ground truth for evaluation."

    The 'gauge observations' invoked as ground truth are not independent station measurements but grid cells from a gauge-satellite merged analysis. The same product supplies both the Stage-2 fine-tuning targets and the evaluation reference. The abstract's phrasing therefore renames the calibration product as external truth, which is what makes the reported bias corrections appear to validate the method against data it was not fitted to. Any systematic bias in the Shen et al. product is inherited by both training and evaluation, so the improvements are not evidence against independent gauge data.

full rationale

The derivation chain is not fully self-contained: Stage 2 fine-tunes the prior P*(x) against the Shen et al. gauge-satellite merged product through the gauge loss term (α3 = 0.5), and the headline evaluation uses grid cells from that same product (cells with at least four assimilated AWS) as 'ground truth'. The 2016 temporal holdout and the zero-shot HRES experiment provide some independent content, so this is not a complete reduction of the method to a fit. However, the central claim—that PRIMER corrects biases using gauge observations as ground truth—is partly circular because the truth product is the calibration target, and any systematic bias in that product is shared by both training and evaluation. Appendix C.2's event selection, based on the same product's intensity values, further aligns the test set with the calibration product. No load-bearing self-citation issue was found; the SI B.1 theoretical bound is imported from external ambient-diffusion results rather than from the authors' prior work. Overall, the headline error-reduction numbers are partially forced by construction, meriting a 6 on the circularity scale.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

No new physical entities are introduced. The main assumptions are the reliability of the merged gauge-satellite product, the functional-space diffusion formulation, and the validity of approximate posterior sampling. The free parameters are empirically chosen weights and hyperparameters.

free parameters (4)
  • Fine-tuning loss weights alpha1, alpha2, alpha3 = alpha1=0.1, alpha2=0.4, alpha3=0.5
    Manually chosen in Stage 2 fine-tuning (Eq. 4); no hyperparameter search reported.
  • SDEdit noise level tau = tau=0.6
    Selected via sensitivity analysis on a single IMERG event (13 June 2016) and used for all posterior sampling.
  • Diffusion steps = 1000
    Empirically chosen, no search.
  • Learning rate and batch size = Varied between 1e-4 and 1e-6; batch size 2-6 per GPU
    Adjusted during intermittent training interruptions; not fixed.
assumptions (4)
  • domain assumption The Shen et al. (2014) gauge-satellite merged product is a reliable ground truth.
    Section 4.6 uses grid cells from this product as training targets and as evaluation truth, assuming cells with >=4 AWS are accurate.
  • standard math Precipitation fields lie in L2([0,1]^n) Hilbert space.
    Inherited from Bond-Taylor and Willcocks (infinity-diffusion), needed for the coordinate-based diffusion formulation.
  • standard math The Wasserstein bound from Daras et al. (2024) applies to the two-stage training setup.
    Used in Appendix B.1 to justify two-stage training; the bound is cited from the literature and not derived here.
  • domain assumption SDEdit and inpainting approximate posterior sampling in diffusion models.
    Adopted from prior work; the paper uses these methods for posterior inference without proving convergence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fusion of multi-source precipitation records via coordinate-based generative model." pith.science (2026). https://pith.science/paper/4TCQNG4O

@misc{pith2026250611698,
  author       = {Pith},
  title        = {Pith review of: Fusion of multi-source precipitation records via coordinate-based generative model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4TCQNG4O}},
  note         = {Machine review of arXiv:2506.11698}
}
read the original abstract

Precipitation remains one of the most challenging climate variables to observe and predict accurately. Existing datasets face intricate trade-offs: gauge observations are relatively trustworthy but sparse, satellites provide global coverage with retrieval uncertainties, and numerical models offer physical consistency but are biased and computationally intensive. Here we introduce PRIMER (Precipitation Record Infinite MERging), a deep generative framework that fuses these complementary sources to produce accurate, high-resolution, full-coverage precipitation estimates. PRIMER employs a coordinate-based diffusion model that learns from arbitrary spatial locations and associated precipitation values, enabling seamless integration of gridded data and irregular gauge observations. Through two-stage training--first learning large-scale patterns, then refining with accurate gauge measurements--PRIMER captures both large-scale climatology and local precision. Once trained, it can downscale forecasts, interpolate sparse observations, and correct systematic biases within a principled Bayesian framework. Using gauge observations as ground truth, PRIMER effectively corrects biases in existing datasets, yielding statistically significant error reductions at most stations and furthermore enhancing the spatial coherence of precipitation fields. Crucially, it generalizes without retraining, correcting biases in operational forecasts it has never seen. This demonstrates how generative AI can transform Earth system science by combining imperfect data, providing a scalable solution for global precipitation monitoring and prediction.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

86 extracted references · 59 canonical work pages

  1. [1]

    & Wenz, L

    Kotz, M., Levermann, A. & Wenz, L. The effect of rainfall changes on economic production.Nature601, 223–227 (2022)

    2022

  2. [2]

    & Portmann, R

    Sun, Y., Solomon, S., Dai, A. & Portmann, R. W. How often does it rain?Journal of climate19, 916–934 (2006)

    2006

  3. [3]

    Pendergrass, A. G. & Knutti, R. The uneven nature of daily precipitation and its change.Geophysical Research Letters45, 11–980 (2018)

    2018

  4. [4]

    & Feingold, G

    Stevens, B. & Feingold, G. Untangling aerosol effects on clouds and precipitation in a buffered system.Nature461, 607–613 (2009)

    2009

  5. [5]

    Birch, C.et al.Impact of soil moisture and convectively generated waves on the initiation of a west african mesoscale convective system.Quarterly Journal of the Royal Meteorological Society139, 1712–1730 (2013)

    2013

  6. [6]

    F., Mooney, P

    Prein, A. F., Mooney, P. A. & Done, J. M. The multi-scale interactions of atmospheric phenomenon in mean and extreme precipitation.Earth’s Future11, e2023EF003534 (2023)

    2023

  7. [7]

    Teixeira, J.et al.Parameterization of the atmospheric boundary layer: a view from just above the inversion.Bulletin of the American Meteorological Society89, 453–458 (2008)

    work page 2008

  8. [8]

    & Molini, A

    Lepore, C., Veneziano, D. & Molini, A. Temperature and cape dependence of rainfall extremes in the eastern united states.Geophysical Research Letters42, 74–83 (2015)

    work page 2015

Show all 86 references
  1. [9]

    The cumulus parameterization problem: Past, present, and future

    Arakawa, A. The cumulus parameterization problem: Past, present, and future. Journal of climate17, 2493–2525 (2004)

    2004

  2. [10]

    Houze Jr, R. A. Mesoscale convective systems.Reviews of Geophysics42(2004)

    2004

  3. [11]

    Sun, Q.et al.A review of global precipitation data sets: Data sources, estimation, and intercomparisons.Reviews of Geophysics56, 79–107 (2018)

    2018

  4. [12]

    & Huffman, G

    Kidd, C. & Huffman, G. Global precipitation measurement.Meteorological Applica- tions18, 334–353 (2011). URL https://rmets.onlinelibrary.wiley.com/doi/abs/10. 1002/met.284

    2011

  5. [13]

    Y.et al.The global precipitation measurement mission.Bulletin of the American meteorological Society95, 701–722 (2014)

    Hou, A. Y.et al.The global precipitation measurement mission.Bulletin of the American meteorological Society95, 701–722 (2014)

    2014

  6. [14]

    & Meneguzzo, F

    Levizzani, V., Amorati, R. & Meneguzzo, F. A review of satellite-based rainfall estimation methods.European Commission Project MUSIC Report (EVK1-CT-2000- 00058)66(2002)

    2002

  7. [15]

    & Brunet, G

    Bauer, P., Thorpe, A. & Brunet, G. The quiet revolution of numerical weather prediction.Nature525, 47–55 (2015)

    2015

  8. [16]

    J.et al.Is precipitation a good metric for model performance?Bulletin of the American Meteorological Society100, 223–233 (2019)

    Tapiador, F. J.et al.Is precipitation a good metric for model performance?Bulletin of the American Meteorological Society100, 223–233 (2019). 44

    2019

  9. [17]

    He, J.et al.The first high-resolution meteorological forcing dataset for land process studies over China.Scientific data7, 25 (2020)

    2020

  10. [18]

    Journal of Geophysical Research: Atmospheres123, 814–834 (2018)

    Ma, Y.et al.Performance of optimally merged multisatellite precipitation products using the dynamic Bayesian model averaging scheme over the Tibetan Plateau. Journal of Geophysical Research: Atmospheres123, 814–834 (2018)

    2018

  11. [19]

    M.et al.Rf-mep: A novel random forest method for merging gridded precipitation products and ground-based measurements.Remote Sensing of Environment239, 111606 (2020)

    Baez-Villanueva, O. M.et al.Rf-mep: A novel random forest method for merging gridded precipitation products and ground-based measurements.Remote Sensing of Environment239, 111606 (2020)

    2020

  12. [20]

    & Wen, Y

    Ur Rahman, K., Shang, S., Shahid, M. & Wen, Y. An appraisal of dynamic Bayesian model averaging-based merged multi-satellite precipitation datasets over complex topography and the diverse climate of Pakistan.Remote Sensing12, 10 (2019)

    2019

  13. [21]

    K., Rathinasamy, M

    Yumnam, K., Guntu, R. K., Rathinasamy, M. & Agarwal, A. Quantile-based Bayesian model averaging approach towards merging of precipitation products.Journal of Hydrology604, 127206 (2022)

    2022

  14. [22]

    & Xiong, A.-Y

    Xie, P. & Xiong, A.-Y. A conceptual model for constructing high-resolution gauge- satellite merged precipitation analyses.Journal of Geophysical Research: Atmospheres 116(2011)

    2011

  15. [23]

    M., Sivakumar, B

    Woldemeskel, F. M., Sivakumar, B. & Sharma, A. Merging gauge and satellite rainfall with specification of associated uncertainty across Australia.Journal of Hydrology499, 167–176 (2013)

    2013

  16. [24]

    Fan, Z.et al.A comparative study of four merging approaches for regional precipitation estimation.IEEE Access9, 33625–33637 (2021)

    2021

  17. [25]

    Zhang, L.et al.Merging multiple satellite-based precipitation products and gauge observations using a novel double machine learning approach.Journal of Hydrology 594, 125969 (2021)

    2021

  18. [26]

    Bhuiyan, M. A. E., Yang, F., Biswas, N. K., Rahat, S. H. & Neelam, T. J. Machine learning-based error modeling to improve GPM IMERG precipitation product over the brahmaputra river basin.Forecasting2, 248–266 (2020)

    2020

  19. [27]

    Bhuiyan, M. A. E., Nikolopoulos, E. I., Anagnostou, E. N., Quintana-Segu ´ ı, P. & Barella-Ortiz, A. A nonparametric statistical technique for combining global precipitation datasets: Development and hydrological evaluation over the Iberian Peninsula.Hydrology and Earth System...

    2018

  20. [28]

    & Wang, G

    Wu, H., Yang, Q., Liu, J. & Wang, G. A spatiotemporal deep fusion model for merging satellite and gauge precipitation in China.Journal of Hydrology584, 124664 (2020)

    2020

  21. [29]

    Shen, Y., Zhao, P., Pan, Y. & Yu, J. A high spatiotemporal gauge-satellite merged precipitation analysis over china.Journal of Geophysical Research: Atmospheres 119, 3063–3075 (2014)

    2014

  22. [30]

    Box, G. E. & Tiao, G. C.Bayesian inference in statistical analysis(John Wiley & Sons, 2011). 45

    2011

  23. [31]

    & Closas, P

    Wu, P., Imbiriba, T., Elvira, V. & Closas, P. Bayesian data fusion with shared priors. IEEE Transactions on Signal Processing72, 275–288 (2023)

    2023

  24. [32]

    Price, I.et al.Gencast: Diffusion-based ensemble forecasting for medium-range weather.arXiv preprint arXiv:2312.15796(2023)

    2023 arXiv

  25. [33]

    Goodfellow, I.et al.Generative adversarial networks.Communications of the ACM 63, 139–144 (2020)

    2020

  26. [34]

    Kingma, D. P. & Welling, M. Auto-encoding variational bayes (2022). URL https: //arxiv.org/abs/1312.6114. arXiv:1312.6114

    2022 arXiv

  27. [35]

    J., Mohamed, S

    Papamakarios, G., Nalisnick, E., Rezende, D. J., Mohamed, S. & Lakshminarayanan, B. Normalizing flows for probabilistic modeling and inference.Journal of Machine Learning Research22, 1–64 (2021)

    2021

  28. [36]

    & Abbeel, P

    Ho, J., Jain, A. & Abbeel, P. Denoising diffusion probabilistic models.Advances in neural information processing systems33, 6840–6851 (2020)

    2020

  29. [37]

    & Ermon, S

    Song, J., Meng, C. & Ermon, S. Denoising diffusion implicit models (2022). URL https://arxiv.org/abs/2010.02502. arXiv:2010.02502

    2022 arXiv

  30. [38]

    & Nichol, A

    Dhariwal, P. & Nichol, A. Diffusion models beat gans on image synthesis (2021). URL https://arxiv.org/abs/2105.05233. arXiv:2105.05233

    2021 arXiv

  31. [39]

    Yim, J.et al.Diffusion models in protein structure and docking.Wiley Interdisciplinary Reviews: Computational Molecular Science14, e1711 (2024)

    2024

  32. [40]

    URL https://arxiv.org/abs/2410.00083

    Daras, G.et al.A survey on diffusion models for inverse problems (2024). URL https://arxiv.org/abs/2410.00083. arXiv:2410.00083

    2024 arXiv

  33. [41]

    URL https://arxiv.org/abs/2503.11043

    Zheng, H.et al.Inversebench: Benchmarking plug-and-play diffusion priors for inverse problems in physical sciences (2025). URL https://arxiv.org/abs/2503.11043. arXiv:2503.11043

    2025

  34. [42]

    & Boers, N

    Hess, P., Aich, M., Pan, B. & Boers, N. Fast, scale-adaptive and uncertainty-aware downscaling of earth system model fields with generative machine learning.Nature Machine Intelligence1–11 (2025)

    2025

  35. [43]

    Diffusion is spectral autoregression (2024)

    Dieleman, S. Diffusion is spectral autoregression (2024). URL https://sander.ai/ 2024/09/02/spectral-autoregression.html

    2024

  36. [44]

    & Brox, T

    Ronneberger, O., Fischer, P. & Brox, T. U-net: Convolutional networks for biomedical image segmentation (2015). URL https://arxiv.org/abs/1505.04597. arXiv:1505.04597

    2015 arXiv

  37. [45]

    A.et al.Are climate model simulations of clouds improving? an evaluation using the ISCCP simulator.Journal of Geophysical Research: Atmospheres118, 1329–1342 (2013)

    Klein, S. A.et al.Are climate model simulations of clouds improving? an evaluation using the ISCCP simulator.Journal of Geophysical Research: Atmospheres118, 1329–1342 (2013)

    2013

  38. [46]

    6): A Python- based diagnostics package for Earth system models evaluation.Geoscientific model development discussions2022, 1–35 (2022)

    Zhang, C.et al.The E3SM diagnostics package (E3SM diags v2. 6): A Python- based diagnostics package for Earth system models evaluation.Geoscientific model development discussions2022, 1–35 (2022). 46

    2022

  39. [47]

    Lee, J.et al.Systematic and objective evaluation of Earth system models: PCMDI Metrics Package (PMP) version 3.Geoscientific Model Development17, 3919–3948 (2024)

    2024

  40. [48]

    & Huffman, G

    Guilloteau, C., Foufoula-Georgiou, E., Kirstetter, P., Tan, J. & Huffman, G. J. How well do multisatellite products capture the space–time dynamics of precipitation? part i: Five products assessed via a wavenumber–frequency decomposition.Journal of Hydrometeorology22, 2805–2823 (2021)

    2021

  41. [49]

    & Huffman, G

    Guilloteau, C., Foufoula-Georgiou, E., Kirstetter, P., Tan, J. & Huffman, G. J. How well do multisatellite products capture the space–time dynamics of precipitation? part ii: Building an error model through spectral system identification.Journal of Hydrometeorology23, 1383–1399 (2022)

    2022

  42. [50]

    URL https://www.ecmwf

    Buizza, R.et al.The development and evaluation process followed at ECMWF to upgrade the Integrated Forecasting System (IFS) (2018). URL https://www.ecmwf. int/node/18658

    2018

  43. [51]

    Zhang, J.et al.Multi-radar multi-sensor (mrms) quantitative precipitation estimation: Initial operating capabilities.Bulletin of the American Meteorological Society97, 621–638 (2016)

    2016

  44. [52]

    E.et al.Mswep: 3-hourly 0.25 global gridded precipitation (1979–2015) by merging gauge, satellite, and reanalysis data.Hydrology and Earth System Sciences 21, 589–615 (2017)

    Beck, H. E.et al.Mswep: 3-hourly 0.25 global gridded precipitation (1979–2015) by merging gauge, satellite, and reanalysis data.Hydrology and Earth System Sciences 21, 589–615 (2017)

    2017

  45. [53]

    A perspective on the future of CMIP.AGU Advances5, e2023A V001086 (2024)

    Stevens, B. A perspective on the future of CMIP.AGU Advances5, e2023A V001086 (2024)

    2024

  46. [54]

    Eyring, V.et al.Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization.Geoscientific Model Development9, 1937–1958 (2016)

    2016

  47. [55]

    & Berkhahn, F

    Guo, C. & Berkhahn, F. Entity embeddings of categorical variables (2016). URL https://arxiv.org/abs/1604.06737. arXiv:1604.06737

    2016 arXiv

  48. [56]

    Song, Y.et al.Score-based generative modeling through stochastic differential equations.arXiv preprint arXiv:2011.13456(2020)

    2020 arXiv

  49. [57]

    & Ermon, S

    Song, Y. & Ermon, S. Generative modeling by estimating gradients of the data distribution.Advances in neural information processing systems32(2019)

    2019

  50. [58]

    Understanding diffusion models: A unified perspective (2022)

    Luo, C. Understanding diffusion models: A unified perspective (2022). URL https: //arxiv.org/abs/2208.11970. arXiv:2208.11970

    2022 arXiv

  51. [59]

    & Willcocks, C

    Bond-Taylor, S. & Willcocks, C. G. ∞-diff: Infinite resolution diffusion with subsampled mollified states (2024). URL https://arxiv.org/abs/2303.18242. arXiv:2303.18242

    2024 arXiv

  52. [60]

    & Wang, S

    Pidstrigach, J., Marzouk, Y., Reich, S. & Wang, S. Infinite-dimensional diffusion models (2023). URL https://arxiv.org/abs/2302.10130. arXiv:2302.10130

    2023 arXiv

  53. [61]

    & Wonka, P

    Zhang, B. & Wonka, P. Functional diffusion (2023). URL https://arxiv.org/abs/ 2311.15435. arXiv:2311.15435. 47

    2023 arXiv

  54. [62]

    Azizzadenesheli, K.et al.Neural operators for accelerating scientific simulations and design.Nature Reviews Physics6, 320–328 (2024)

    2024

  55. [63]

    Biemond, J., Lagendijk, R. L. & Mersereau, R. M. Iterative methods for image deblurring.Proceedings of the IEEE78, 856–883 (2002)

    2002

  56. [64]

    URL https://arxiv.org/abs/2003.03485

    Li, Z.et al.Neural operator: Graph kernel network for partial differential equations (2020). URL https://arxiv.org/abs/2003.03485. arXiv:2003.03485

    2020 arXiv

  57. [65]

    Kovachki, N.et al.Neural operator: Learning maps between function spaces with applications to pdes.Journal of Machine Learning Research24, 1–97 (2023)

    2023

  58. [66]

    arXiv preprint arXiv:2010.08895(2020)

    Li, Z.et al.Fourier neural operator for parametric partial differential equations. arXiv preprint arXiv:2010.08895(2020)

    2020 arXiv

  59. [67]

    & Han, S

    Tang, H., Liu, Z., Li, X., Lin, Y. & Han, S. Torchsparse: Efficient point cloud inference engine (2022). URL https://arxiv.org/abs/2204.10319. arXiv:2204.10319

    2022 arXiv

  60. [68]

    Ruiz, N.et al.Dreambooth: Fine tuning text-to-image diffusion models for subject- driven generation (2022)

    2022

  61. [69]

    T., Klasky, M

    Chung, H., Kim, J., Mccann, M. T., Klasky, M. L. & Ye, J. C. Diffusion posterior sampling for general noisy inverse problems (2024). URL https://arxiv.org/abs/2209. 14687. arXiv:2209.14687

    2024 arXiv

  62. [70]

    Chao, J.et al.Learning to infer weather states using partial observations.Journal of Geophysical Research: Machine Learning and Computation2, e2024JH000260 (2025)

    2025

  63. [71]

    URL https://arxiv.org/abs/2201.09865

    Lugmayr, A.et al.Repaint: Inpainting using denoising diffusion probabilistic models (2022). URL https://arxiv.org/abs/2201.09865. arXiv:2201.09865

    2022 arXiv

  64. [72]

    URL https://arxiv.org/abs/2304.03322

    Zhang, G.et al.Towards coherent image inpainting using denoising diffusion implicit models (2023). URL https://arxiv.org/abs/2304.03322. arXiv:2304.03322

    2023 arXiv

  65. [73]

    URL https://arxiv.org/abs/2108.01073

    Meng, C.et al.SDEdit: Guided image synthesis and editing with stochastic differential equations (2022). URL https://arxiv.org/abs/2108.01073. arXiv:2108.01073

    2022 arXiv

  66. [74]

    & Raftery, A

    Gneiting, T. & Raftery, A. E. Strictly proper scoring rules, prediction, and estimation. Journal of the American statistical Association102, 359–378 (2007)

    2007

  67. [75]

    A primer for eof analysis of climate data.Department of Meteorology, University of Reading1, 3 (2004)

    Hannachi, A. A primer for eof analysis of climate data.Department of Meteorology, University of Reading1, 3 (2004)

    2004

  68. [76]

    0).Geoscientific Model Development12, 4185–4219 (2019)

    Pulkkinen, S.et al.Pysteps: An open-source Python library for probabilistic pre- cipitation nowcasting (v1. 0).Geoscientific Model Development12, 4185–4219 (2019)

    2019

  69. [77]

    J.et al.Nasa global precipitation measurement (GPM) integrated multi-satellite retrievals for GPM (IMERG).Algorithm theoretical basis document (ATBD) version4, 30 (2015)

    Huffman, G. J.et al.Nasa global precipitation measurement (GPM) integrated multi-satellite retrievals for GPM (IMERG).Algorithm theoretical basis document (ATBD) version4, 30 (2015)

    2015

  70. [78]

    Hersbach, H.et al.The ERA5 global reanalysis.Quarterly journal of the royal meteorological society146, 1999–2049 (2020). 48

    2020

  71. [79]

    Rasp, S.et al.Weatherbench: a benchmark data set for data-driven weather forecasting.Journal of Advances in Modeling Earth Systems12, e2020MS002203 (2020)

    2020

  72. [80]

    & Messori, G

    Olivetti, L. & Messori, G. Do data-driven models beat numerical models in fore- casting weather extremes? a comparison of ifs hres, pangu-weather, and graphcast. Geoscientific Model Development17, 7915–7962 (2024)

    2024

  73. [81]

    & Daskalakis, C

    Daras, G., Cherapanamjeri, Y. & Daskalakis, C. How much is a noisy image worth? data scaling laws for ambient diffusion (2024). URL https://arxiv.org/abs/2411.02780. arXiv:2411.02780

    2024 arXiv

  74. [82]

    Advances in Neural Information Processing Systems36, 288–313 (2023)

    Daras, G.et al.Ambient diffusion: Learning clean distributions from corrupted data. Advances in Neural Information Processing Systems36, 288–313 (2023)

    2023

  75. [83]

    & Daskalakis, C

    Daras, G., Dagan, Y., Dimakis, A. & Daskalakis, C. Consistent diffusion models: Mit- igating sampling drift by learning to be consistent.Advances in Neural Information Processing Systems36, 42038–42063 (2023)

    2023

  76. [84]

    Daras, G., Dimakis, A. G. & Daskalakis, C. Consistent diffusion meets tweedie: Train- ing exact ambient diffusion models with noisy data.arXiv preprint arXiv:2404.10177 (2024)

    2024 arXiv

  77. [85]

    Panaretos, V. M. & Zemel, Y. Statistical aspects of wasserstein distances.Annual review of statistics and its application6, 405–431 (2019)

    2019

  78. [86]

    Paszke, A.et al.Automatic differentiation in pytorch (2017). 49

    2017

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.