Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

Leveraging GNN to Enhance MEF Method in Predicting ENSO

T0 review · 4 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Graph-based subset of 20 ensemble members improves ENSO forecasts by about 10% over the standard blend.

desk verdict The reported ~10% skill gain is an artifact of selecting the 20-member subset using the same test-period observations used for evaluation, and the paper never actually specifies or implements a GNN. read the letter →

arxiv 2508.07410 v3 pith:QSSEVZAH submitted 2025-08-10 physics.ao-ph cs.AI

Saghar Ganji , Ahmad Reza Labibzadeh , Alireza Hassani , Mohammad Naisipour This is my paper · ORCID
classification physics.ao-phcs.AI
keywords ENSOforecastinggraphneuralnetworksensembleselectionsimilaritycommunitydetectionmultimodaldeeplearninglong-leadpredictionNiño3.4index
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

The paper argues that the strongest ENSO forecasts hide inside a large ensemble, and that a graph can find them. It builds an undirected graph whose nodes are the 80 members of the MEF ensemble and whose edge weights measure how similar two forecast outputs are; a GNN and community detection then pick 20 structurally coherent members, and the final forecast is their average. The claim is that this selection raises correlation skill by roughly 10% over the baseline MEF weighting, especially at lead times beyond one year, and keeps correlation above 0.4 through month 23. If true, this would make long-lead ENSO prediction more usable and would transfer to any ensemble-based forecasting system without retraining the base models.

What carries the argument

The similarity graph G(V,E): each of the 80 ensemble outputs is a vertex, and each edge weight encodes pairwise similarity between two runs' forecast outputs, computed from metrics such as RMSE and Pearson correlation. A graph neural network embeds the nodes by aggregating neighbor information, and community detection identifies a dense cluster of coherent, accurate runs from which 20 members are selected; the final prediction is the simple average of those 20. This lets ensemble selection act on structural agreement rather than on individual scalar error alone.

What would settle it

On the same data, rebuild the similarity graph using only information available before the target season (e.g., errors on a training-period sample) and re-select 20 members, then score on the post-2000 period. If the correlation gain over MEF falls to zero or reverses, the paper's improvement is selection leakage rather than a forecasting advance. A second check is to replace the GNN with simple k-medoids on the same similarity matrix and compare.

Watch

Extended reading notes

Core claim

The authors claim that averaging a graph-selected subset of 20 out of 80 MEF ensemble members gives more skillful and more stable multi-year ENSO forecasts than the original MEF weighting. The selection is made by representing all runs as nodes of a weighted undirected graph, with edge weights derived from RMSE and correlation between run outputs, then applying a GNN and community detection to locate a dense cluster of consistent, high-performing runs. On the post-2000 test period, the graph-selected forecast consistently improves correlation skill by about 10% relative to MEF, is superior to the CNN baseline at all lead times and especially beyond 17 months, and maintains correlation skill

Load-bearing premise

The similarity graph for selecting the 20 members is built using RMSE and correlation of each forecast against the observed Niño 3.4 index for the same post-2000 period used in the evaluation, so the reported skill gain is only a true forecast improvement if this choice does not amount to picking members by looking at the answer key.

Editorial extensions

If this is right

  • Selecting and averaging 20 graph-coherent members yields roughly 10% higher correlation skill than the full MEF weighting, with the largest gains at lead times beyond one year.
  • The method keeps the forecast correlation skill above 0.4 for all lead months tested (up to 23 months) and captures roughly half of El Niño and La Niña events at the longest leads.
  • Because the selection step operates only on the outputs, it can be attached to any ensemble forecast system—statistical, dynamical, or hybrid—without changing the base model.
  • Selected members tend to be densely interconnected, low-variance, spectrally smooth runs; this fingerprint could be used as a cheap screening criterion.
  • Forecasts become more stable and consistent, particularly under compound long-lead conditions, even in scenarios where the accuracy gain over the baseline is small.

Reading between the lines

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

  • If the graph-selection gain generalizes, the practical benefit may come less from the GNN itself than from shifting from averaging everything to averaging a consensus cluster; a simpler cluster-medoid baseline might capture much of the benefit at lower cost, and the paper does not test this.
  • The paper's similarity metrics are computed against the observed Niño 3.4 index over the same post-2000 window used for scoring, so an out-of-sample re-run with training-period similarity weights would be the decisive check of whether the ~10% is real forecast skill.
  • The reported statistical fingerprint suggests that spectral smoothness and autocorrelation decay rate—cheaply computed from an output time series—could serve as a surrogate for graph centrality, letting other groups apply the idea without a GNN.
  • The same recipe could be applied to large physical ensembles such as S2S reforecast archives, where a 20-member consensus subset might also reduce the noise from poorly initialized members.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes a graph-based ensemble post-processing method for the Multimodal ENSO Forecast (MEF) model. The method constructs an undirected weighted graph whose nodes are the 40 (or later, 80) ensemble members and whose edge weights are similarity measures (RMSE and correlation) computed with respect to the observed Niño 3.4 index. A GNN/community-detection step is then used to select 20 members, whose average is the final forecast. The paper claims this selection improves MEF skill by about 10%, especially at lead times beyond 17 months, and that the approach is model-agnostic. The evaluation is performed on the post-2000 period.

Significance. If the reported skill gains were obtained by a genuinely predictive selection rule, the method could be a useful, model-agnostic ensemble post-processing tool for long-lead ENSO forecasting. The idea of selecting a coherent subset of ensemble members based on graph structure is interesting and has potential. However, the manuscript as written does not establish the central claim: the selection rule uses test-period observations to choose the subset, and the same observations are then used to evaluate it. In addition, the GNN/community-detection methodology is not actually specified, and the reported 40/80 member inconsistency makes the pipeline ambiguous. No code, architecture details, or reproducible experimental protocol are provided. These issues are load-bearing, so the paper cannot be accepted in its current form.

major comments (4)
  1. [§2.2, §3.3] The selection rule is circular with respect to the evaluation. Section 2.2 defines edge weights using 'RMSE: absolute error of prediction with respect to the ground truth' and Pearson correlation against observed Niño 3.4. Section 3.3 evaluates the 20 selected ensemble members on the same post-2000 period used to compute these errors (Fig. S4). Thus the selected subset is chosen as the best-performing 20 on the test labels, and the reported ~10% skill gain is largely a direct consequence of test-set fitting rather than an honest forecast improvement. A proper forecast evaluation requires a temporal split in which member selection uses only data before the evaluation period, or a fully cross-validated selection protocol.
  2. [§2.2, §3.3] The manuscript does not actually specify the GNN or community-detection method despite naming it in the title and abstract. Section 2.2 states 'one may use a GNN' and lists optional node features, but no architecture, training objective, number of layers, or community detection algorithm is given. Section 3.3 refers to 'graph-based similarity clustering' without identifying the algorithm. The only concrete selection criterion that appears in the text is the test-period RMSE/correlation edge weight. Thus the central mechanistic claim—that GNN-based structural analysis improves ensemble selection—is unsupported. The paper needs to provide a reproducible description of the method: the actual graph construction, the GNN training procedure, the clustering/community detection algorithm, and the exact rule for choosing 20 members.
  3. [§2.2, §3.3, Abstract] There is a direct numerical inconsistency in the ensemble size. Section 2.2 and Section 2.3 state the 3DCNN produces 40 runs, and Section 2.3 says the MEF model comprises '40 stochastic runs.' The Abstract and Section 3.3, however, state the graph is built from 'all 80 members of the ensemble' and that 20 are selected 'from within the initial 80.' The pipeline is therefore ambiguous: does the graph contain 40 or 80 nodes? This is not a typo-level issue because the subset-selection statistics depend on the pool size. The authors must reconcile these numbers and specify exactly how the 80 members arise (e.g., 40 3DCNN plus 40 time-series runs).
  4. [Fig. 5, §3.2] The central quantitative claim—'consistently improves performance by approximately 10% over MEF, particularly in extended lead times'—is presented only in a figure caption with no supporting numeric table, confidence intervals, or significance tests. Moreover, Fig. 5 is captioned as comparing 'MEF and GNN-based ensemble selection method,' while the surrounding text and Fig. S4 compare MEF to the CNN baseline. It is unclear whether the improvement is over the original MEF weighted average or over the CNN baseline. The authors need to provide a clear comparison table with skill scores (correlation, RMSE) for MEF, the GNN-selected method, and the CNN baseline, with uncertainties, across lead months.
minor comments (4)
  1. [Global] The manuscript contains several duplicated or misordered section numbers (two '2.2' sections, '2.3' and '2.4' appear out of order relative to the description of the integrated model). The references are not formatted consistently, and several citations are not relevant to the claims (e.g., refs. 38, 53 in the context of climate forecasting). Please revise thoroughly.
  2. [§2.1] Equation (1) has formatting issues (subscripts and superscripts are not clearly delineated) and the dimensions of the input tensor are not explicitly defined. The description of the kernel sizes and pooling operations is hard to follow; a clear architecture table would improve reproducibility.
  3. [§3.3] The statement 'MEF captures nearly 50% of all El Niño and La Niña events even at 23-month horizons' is not accompanied by a definition of 'captures' (e.g., phase hit rate, amplitude threshold) or a comparison with the baseline. Please define the metric and provide quantitative evidence.
  4. [Abstract and §4] The abstract claims the method 'produces more stable and consistent outputs,' but no measure of stability or consistency is defined or reported. The conclusion similarly claims superiority 'across almost all seasons' without supporting seasonal skill scores.

Circularity Check

1 steps flagged · score 8.0 of 10

Selection uses observed Niño 3.4 in the evaluation period both to choose the 20-member subset and to score it, so the reported ~10% skill gain is circular; the GNN/community-detection step is unspecified.

  1. fitted input called prediction [Section 2.2 (similarity metrics); Section 2.3 (validation period); Section 3.3 (selection and evaluation)]
    "Root Mean Square Error (RMSE): absolute error of prediction with respect to the ground truth [65, 24] ... The 3DCNN module was trained using CMIP5 data spanning the years 1875 to 1975, with validation conducted on GODAS data from 2000 to 2017. ... It is constructed from the average of a subset of 20 of the selected ensemble members from within the initial 80 through graph-based similarity clustering as the most statistically and structurally consistent forecasts."

    Edge weights for graph clustering/selection are explicitly computed as RMSE between each ensemble prediction and observed Niño 3.4 ground truth. The model's validation/evaluation period is fixed as GODAS 2000–2017 (Section 2.3), and Section 3 says the paper focuses on post-2000 skill. The 20-member subset is then chosen from that graph, and 'the average of a subset of 20' is the final prediction whose correlation with the same observed Niño 3.4 record is reported (Fig. 5, Fig. S4). Because no temporal split is described, the selection is fitted to the very labels used for evaluation. Any rule that picks low-RMSE members on the evaluation period would show skill; the unspecified GNN/community-detection step cannot rescue the evaluation. The ~10% improvement over MEF is therefore a property

full rationale

The central claim of the paper is that GNN-based ensemble selection improves MEF forecast skill by ~10%, especially at long lead times. That claim is not supported as a forecast result because the selection rule uses the evaluation labels. In Section 2.2 the edge weights for the similarity graph are defined by RMSE 'with respect to the ground truth' and by correlation; in Section 3.3 the final prediction is the average of 20 members selected via graph clustering, and the same GODAS Niño 3.4 record (2000–2017) is used to evaluate it. No temporal split is stated between the period used to compute these similarity metrics and the period used to score Fig. 5/Fig. S4, so picking low-error members on the test period and then averaging them will automatically inflate skill. The GNN/community-detection algorithm is never specified (no architecture, loss, or training description), and Section 2.2 says the graph has 40 runs while Section 3.3 says 20 of an initial 80, leaving RMSE/correlation against observed Niño 3.4 as the only operative selection criterion. This is exactly the 'fitted input called prediction' pattern. The self-citations to the authors' MEF papers are not themselves the circularity; the MEF baseline is a prior model and could be an external benchmark. But the claimed improvement over it is contaminated by test-set selection. Under the given scoring rubric, the central claim reduces by construction, so the circularity score is 8.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on selecting 20 of 80 ensemble members using a similarity graph whose edge weights include RMSE against the observed Nino 3.4 index. This requires validation-period truth at selection time and assumes the selected subset's advantage persists. The base model's many hyperparameters are inherited from earlier MEF work and are not derived here, while the GNN component is never actually specified.

free parameters (3)
  • selected subset size k = 20
    The final forecast averages exactly 20 of 80 ensemble members; no sensitivity analysis over k is shown and the number appears chosen after the fact.
  • edge-weight combination of RMSE and correlation = unstated
    The graph construction uses both RMSE and correlation, but their relative weighting, normalization, or threshold is not specified. This choice controls the clusters and therefore the selected subset.
  • 3DCNN hyperparameters (filters, fully connected neurons, learning rate) = 5 or 7 filters, 50 or 70 neurons, learning rate 0.005
    These hyperparameters shape the base ensemble outputs on which selection operates; they are inherited from earlier MEF work and not derived or justified in this paper.
assumptions (4)
  • ad hoc to paper Ground-truth Nino 3.4 index is available at selection time to compute RMSE/correlation edge weights.
    Section 2.2 defines similarity using RMSE relative to ground truth; if applied in the same forecast period, this assumes the truth is known before forecasting.
  • domain assumption The skill ranking of ensemble members is stable between the selection period and the forecast period.
    The method selects members based on observed similarity to the target and assumes those members remain top performers in future, unobserved periods.
  • domain assumption CMIP5 training data from 1875 to 1975 transfers to GODAS validation data from 2000 to 2017.
    The authors intentionally use a 25-year gap, but transferability of precursor relationships under climate change is assumed without demonstration.
  • ad hoc to paper Community structure in the similarity graph corresponds to forecast quality.
    No theoretical or empirical justification is given beyond the reported results; the paper does not show that clusters of similar forecasts are more skillful than arbitrary subsets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Leveraging GNN to Enhance MEF Method in Predicting ENSO." pith.science (2026). https://pith.science/paper/QSSEVZAH

@misc{pith2026250807410,
  author       = {Pith},
  title        = {Pith review of: Leveraging GNN to Enhance MEF Method in Predicting ENSO},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QSSEVZAH}},
  note         = {Machine review of arXiv:2508.07410}
}
read the original abstract

Reliable long-lead forecasting of the El Nino Southern Oscillation (ENSO) remains a long-standing challenge in climate science. The previously developed Multimodal ENSO Forecast (MEF) model uses 80 ensemble predictions by two independent deep learning modules: a 3D Convolutional Neural Network (3D-CNN) and a time-series module. In their approach, outputs of the two modules are combined using a weighting strategy wherein one is prioritized over the other as a function of global performance. Separate weighting or testing of individual ensemble members did not occur, however, which may have limited the model to optimize the use of high-performing but spread-out forecasts. In this study, we propose a better framework that employs graph-based analysis to directly model similarity between all 80 members of the ensemble. By constructing an undirected graph whose vertices are ensemble outputs and whose weights on edges measure similarity (via RMSE and correlation), we identify and cluster structurally similar and accurate predictions. From which we obtain an optimized subset of 20 members using community detection methods. The final prediction is then obtained by averaging this optimized subset. This method improves the forecast skill through noise removal and emphasis on ensemble coherence. Interestingly, our graph-based selection shows robust statistical characteristics among top performers, offering new ensemble behavior insights. In addition, we observe that while the GNN-based approach does not always outperform the baseline MEF under every scenario, it produces more stable and consistent outputs, particularly in compound long-lead situations. The approach is model-agnostic too, suggesting that it can be applied directly to other forecasting models with gargantuan ensemble outputs, such as statistical, physical, or hybrid models.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Distillation of CNN Ensemble Results for Enhanced Long-Term Prediction of the ENSO Phenomenon

    physics.ao-ph 2025-09 reject novelty 3.0 of 10

    Selecting the five best ENSO ensemble members by their past errors and correlations, then scoring them on the same past data, shows large in-sample gains over the all-member mean but provides no out-of-sample evidence...

Reference graph

Works this paper leans on

98 extracted references · 65 canonical work pages · cited by 1 Pith paper

  1. [1]

    Wang, GG., Cheng, H., Zhang, Y, Yu, H., ENSO analysis and prediction using deep learning: A review, Neurocomputing 520, 216-229 (2023)

    2023

  2. [2]

    2010, Impact of ENSO on European Climate ECMWF Seminar on Predictability in the European and Atlantic regions, ECMWF conference, Shinfield Park, Reading

    Scaife, A., et al. 2010, Impact of ENSO on European Climate ECMWF Seminar on Predictability in the European and Atlantic regions, ECMWF conference, Shinfield Park, Reading

    2010

  3. [3]

    et al., 2019, Changing climate both increases and decreases European river floods

    Blöschl, G., Hall, J., Viglione, A. et al., 2019, Changing climate both increases and decreases European river floods. Nature 573, 108–111

    2019

  4. [4]

    J., Zebiak, S

    McPhaden, M. J., Zebiak, S. E. & Glantz, M. H., 2006, ENSO as an integrating concept in Earth science. Science 314, 1740–1745

    2006

  5. [5]

    Luo, J.-J., Masson, S., Behera, S. K. & Yamagata, T., 2008, Extended ENSO predictions using a fully coupled ocean–atmosphere model. J. Clim. 21, 84–93

    2008

  6. [6]

    & Zhang, R

    Gao, C. & Zhang, R. H., 2017, The roles of atmospheric wind and entrained water temperature (Te) in the second- year cooling of the 2010–12 La Niña event. Clim. Dyn. 48, 597–617

    2017

  7. [7]

    H., Kug, J

    Park, J. H., Kug, J. S., Li, T. & Behera, S. K., 2018, Predicting El Niño beyond 1-year lead: effect of the Western Hemisphere warm pool. Sci. Rep. 8, 14957

    2018

  8. [8]

    Kirtman, B. et al. 2013, The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) Ch. 11, Cambridge Univ. Press

    2013

Show all 98 references
  1. [9]

    Shepherd, T. G. 2014, Atmospheric circulation as a source of uncertainty in climate change projections. Nat. Geosci. 7, 703–708

    2014

  2. [10]

    https://doi.org/10.1029/2023GL108013

    Wang, R., He, J., Luo, J., J., Chen, L., Atlantic Warming Enhances the Influence of Atlantic Niño on ENSO, Geophisical Research Letters, 51, e2023GL108013. https://doi.org/10.1029/2023GL108013

    doi

  3. [11]

    & Xu, YP

    Chen, H., Teegavarapu, R.S.V. & Xu, YP. Oceanic-Atmospheric Variability Influences on Baseflows in the Continental United States. Water Resour Manage 35, 3005–3022 (2021). https://doi.org/10.1007/s11269-021- 02884-6

    2021 doi

  4. [12]

    Park, JH., Yang, YM., Ham, YG. et al. Significant winter Atlantic Niño effect on ENSO and its future projection. npj Clim Atmos Sci 7, 238 (2024). https://doi.org/10.1038/s41612-024-00790-3

    2024 doi

  5. [13]

    Park, JH., Kug, JS., Yang, YM. et al. Distinct decadal modulation of Atlantic-Niño influence on ENSO. npj Clim Atmos Sci 6, 105 (2023). https://doi.org/10.1038/s41612-023-00429-9

    2023 doi

  6. [14]

    A Non-stationary and Probabilistic Approach for Drought Characterization Using Trivariate and Pairwise Copula Construction (PCC) Model

    Dixit, S., Jayakumar, K.V. A Non-stationary and Probabilistic Approach for Drought Characterization Using Trivariate and Pairwise Copula Construction (PCC) Model. Water Resour Manage 36, 1217–1236 (2022)

    2022

  7. [15]

    et al., 2010, Influence of the state of the Indian Ocean Dipole on the following year’s El Niño

    Izumo, T. et al., 2010, Influence of the state of the Indian Ocean Dipole on the following year’s El Niño. Nature Geoscience 3, 168–172

    2010

  8. [16]

    Ling, F., Luo, JJ., Li, Y. et al. Multi-task machine learning improves multi-seasonal prediction of the Indian Ocean Dipole. Nat Commun 13, 7681 (2022)

    2022

  9. [17]

    J., Wallace, J

    Vimont, D. J., Wallace, J. M. & Battisti, D. S., 2003, The seasonal footprinting mechanism in the Pacific: implications for ENSO*. Journal ofClimate 16, 2668–2675

    2003

  10. [18]

    et al., 2007, Pacific meridional mode and El Niño—Southern Oscillation

    Chang, P. et al., 2007, Pacific meridional mode and El Niño—Southern Oscillation. Geophys. Res. Lett. 34, L16608, https://doi.org/10.1029/2007GL030302

    2007 doi

  11. [19]

    Anderson, B. T. & Perez, R. C., 2015, ENSO and non-ENSO induced charging and discharging of the equatorial Pacific. Climate Dynamics45, 2309–2327

    2015

  12. [20]

    Chen, HC., Tseng, YH., Hu, ZZ. et al. Enhancing the ENSO Predictability beyond the Spring Barrier. Sci Rep 10, 984 (2020)

    2020

  13. [21]

    Kilinc, H.C., Apak, S., Ozkan, F. et al. Multimodal Fusion of Optimized GRU–LSTM with Self-Attention Layer for Hydrological Time Series Forecasting. Water Resour Manage 38, 6045 –6062 (2024). https://doi.org/10.1007/s11269-024-03943-4

    2024 doi

  14. [22]

    Xia, B., Enhancing 3D object detection through multi-modal fusion for cooperative perception, Alexandria Engineering Journal, 104, 46-55, 2024, https://doi.org/10.1016/j.aej.2024.06.025

    2024 doi

  15. [23]

    Kumshe, U.M.M., Abdulhamid, Z.M., Mala, B.A. et al. Improving Short-term Daily Streamflow Forecasting Using an Autoencoder Based CNN -LSTM Model. Water Resour Manage 38, 5973 –5989 (2024). https://doi.org/10.1007/s11269-024-03937-2

    2024 doi

  16. [24]

    Zhao, J., Luo, H., Sang, W. et al. Spatiotemporal semantic network for ENSO forecasting over long time horizon. Appl Intell 53, 6464–6480 (2023). https://doi.org/10.1007/s10489-022-03861-1

    2023 doi

  17. [25]

    Long Lead ENSO Forecast Using an Adaptive Graph Convolutional Recurrent Neural Network

    Jonnalagadda, J.; Hashemi, M. Long Lead ENSO Forecast Using an Adaptive Graph Convolutional Recurrent Neural Network. Eng. Proc. 2023, 39, 5. https://doi.org/10.3390/engproc2023039005

    2023 doi

  18. [26]

    Doi, T., S. K. Behera, and T. Yamagata, 2019, Merits of a 108-Member Ensemble System in ENSO and IOD Predictions. J. Climate, 32, 957–972

    2019

  19. [27]

    & Luo, J

    Ham, Y., Kim, J. & Luo, J. 2019, Deep learning for multi-year ENSO forecasts. Nature 573, 568–572

    2019

  20. [28]

    Transfer learning for streamflow forecasting using unguaged MOPEX basins data set

    Muhammad, A.U., Abba, S.I. Transfer learning for streamflow forecasting using unguaged MOPEX basins data set. Earth Sci Inform 16, 1241–1264 (2023). https://doi.org/10.1007/s12145-023-00952-6

    2023 doi

  21. [29]

    Kim, ES., On, KW., Unified deep learning model for El Niño/Southern Oscillation forecasts by incorporating seasonality in climate data, Science Bulletin, 66, 13, 1358-1366, (2021)

    Ham, YG., Kim, JH. Kim, ES., On, KW., Unified deep learning model for El Niño/Southern Oscillation forecasts by incorporating seasonality in climate data, Science Bulletin, 66, 13, 1358-1366, (2021)

    2021

  22. [30]

    Muhammad, A.U., Djigal, H., Muazu, T. et al. An autoencoder-based stacked LSTM transfer learning model for EC forecasting. Earth Sci Inform 16, 3369–3385 (2023). https://doi.org/10.1007/s12145-023-01096-3

    2023 doi

  23. [31]

    A Hybrid Neural Network Model for ENSO Prediction in Combination with Principal Oscillation Pattern Analyses

    Zhou, L., Zhang, RH. A Hybrid Neural Network Model for ENSO Prediction in Combination with Principal Oscillation Pattern Analyses. Adv. Atmos. Sci. 39, 889–902 (2022)

    2022

  24. [32]

    Zhou, L., Zhang, RH., A self-attention–based neural network for three-dimensional multivariate modeling and its skillful ENSO predictions. Sci. Adv.9,eadf2827(2023)

    2023

  25. [33]

    Cai, W., Borlace, S., Lengaigne, M. et al. Increasing frequency of extreme El Niño events due to greenhouse warming. Nature Clim Change 4, 111–116 (2014). https://doi.org/10.1038/nclimate2100

    2014 doi

  26. [34]

    Cai, W., McPhaden, M.J., Grimm, A.M. et al. Climate impacts of the El Niño–Southern Oscillation on South America. Nat Rev Earth Environ 1, 215–231 (2020). https://doi.org/10.1038/s43017-020-0040-3

    2020 doi

  27. [35]

    ENSO prediction,

    L'Heureux, M. L., Levine, A. F., Newman, M., Ganter, C., Luo, J. J., Tippett, M. K., et al. (2020). “ENSO prediction,” in El Niño Southern Oscillation in a Changing Climate, John Wiley & Sons, 227 –246. doi: 10.1002/9781119548164.ch10

    2020 doi

  28. [36]

    Y., et al., (2021)

    Planton, Y. Y., et al., (2021). Evaluating climate models with the CLIVAR 2020 ENSO metrics package. Bulletin of the American Meteorological Society, 102(2), E193-E217. https://doi.org/10.1175/BAMS-D-19-0337.1

    2021 doi

  29. [37]

    Natural climate variability and future climate policy

    Ricke, K., Caldeira, K. Natural climate variability and future climate policy. Nature Clim Change 4, 333–338 (2014). https://doi.org/10.1038/nclimate2186

    2014 doi

  30. [38]

    Moazezi Khah Tehran, A., Hassani, A., Mohajer, S., Darvishan, S., Shafiesabet, A., & Tashakkori, A. (2025). The Impact of Financial Literacy on Financial Behavior and Financial Resilience with the Mediating Role of Financial Self-Efficacy. International Journal of Industrial E...

    2025 doi

  31. [39]

    & McPhaden, M

    Guan, C. & McPhaden, M. J. Ocean processes affecting the twenty- first-century shift in ENSO SST variability. J. Clim. 29, 6861–6879 (2016)

    2016

  32. [40]

    Wang, H., Hu, S., Guan, C. et al. The role of sea surface salinity in ENSO forecasting in the 21st century. npj Clim Atmos Sci 7, 206 (2024). https://doi.org/10.1038/s41612-024-00763-6

    2024 doi

  33. [41]

    Hunter, J. D. Matplotlib: a 2D graphics environment. Comput. Sci. Eng. 9, 90–95 (2007)

    2007

  34. [42]

    & Sivic, J., 2014, Learning and transferring mid-level image representations using convolutional neural networks

    Oquab, M., Bottou, L., Laptev, I. & Sivic, J., 2014, Learning and transferring mid-level image representations using convolutional neural networks. In Proc. IEEE Conference

    2014

  35. [43]

    Yoo, J. H. & Kang, I. S. 2005, Theoretical examination of a multi-model composite for seasonal prediction. Geophys. Res. Lett. 32, L18707

    2005

  36. [44]

    2022, RL based hyper-parameters optimization algorithm (ROA) for convolutional neural network

    Talaat, F.M., Gamel, S.A. 2022, RL based hyper-parameters optimization algorithm (ROA) for convolutional neural network. J Ambient Intell Human Comput

    2022

  37. [45]

    2010, Understanding the difficulty of training deep feedforward neural networks

    Glorot X, Bengio Y. 2010, Understanding the difficulty of training deep feedforward neural networks. Proceedings of the thirteenth international conference on artificial intelligence and statistics. p. 249–56

    2010

  38. [46]

    5998- 6008

    Vaswani A., Shazeer N., Parmar N., et al., 2017, Attention is all you need, Proceedings of NeurIPS, pp. 5998- 6008

    2017

  39. [47]

    Nicolas, C., Francisco, M., Gabriel, S., Nicolas, U., Alexander, K., Sergey, Z., 2020, End-to-end object detection with transformers, Proceedings of ECCV, pp. 213-229

    2020

  40. [48]

    1572-1583

    Schwaller P., Laino T., Gaudin T., Bolgar P., Hunter C., Bekas C., Lee A., 2019, Molecular transformer: A model for uncertainty-calibrated chemical reaction prediction, ACS Central Sci., 5 (9), pp. 1572-1583

    2019

  41. [49]

    Rives A., Meier J., Sercu T., Goyal S., et al., 2016, Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences, Proc. Natl. Acad. Sci., 118 (15), 10.1073/pnas.2016239118

    2016 doi

  42. [50]

    Castangia, M., Grajales, L., Aliberti, A., Rossi, C., 2023, Transformer neural networks for interpretable flood forecasting, Environmental Modelling & Software, Volume 160,105581

    2023

  43. [51]

    Ye, F., Hu, J., Huang, T., You, L., Weng, B., Gao, J., 2022, Transformer for EI Niño -Southern Oscillation Prediction, IEEE Geoscience and Remote Sensing Letters, vol. 19, pp. 1-5

    2022

  44. [52]

    Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting

    Zhou, H., Zhang, S., et al. Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting. AAAI-21 (2021)

    2021

  45. [53]

    Goodfellow, I. J et al. 2004. Elasto-plastic fracturing model for wellbore stability using non-penetrating fluids. arXiv, 10.48550/ARXIV.1406.2661

    2004 doi arXiv

  46. [54]

    Salimans T., et al. 2016. Improved Techniques for Training GANs. arXiv:1606.03498

    2016 arXiv

  47. [55]

    Computer Vision and Pattern Recognition

    Isola, P., Zhu, J., Zhou, T., Efros, A., 2017, Image-to-Image Translation with Conditional Adversarial Nets. Computer Vision and Pattern Recognition

    2017

  48. [56]

    Ho, J., Ermon, S. 2016. Generative Adversarial Imitation Learning". Advances in Neural Information Processing Systems. 29: 4565–4573. arXiv:1606.03476. Bibcode:2016arXiv160603476H

    2016 arXiv

  49. [57]

    Generative Adversarial Networks recover features in astrophysical images of galaxies beyond the deconvolution limit

    Schawinski, K., et al., 2017. Generative Adversarial Networks recover features in astrophysical images of galaxies beyond the deconvolution limit. Monthly Notices of the Royal Astronomical Society: Letters. 467 (1): L110–L114

    2017

  50. [58]

    Mustafa, M., et al. 2019. CosmoGAN: creating high-fidelity weak lensing convergence maps using Generative Adversarial Networks. Computational Astrophysics and Cosmology. 6 (1): 1. arXiv:1706.02390. ISSN 2197- 7909

    2019 arXiv

  51. [59]

    Paganini, M., de Oliveira, L., Nachman, B. 2017. Learning Particle Physics by Example: Location-Aware Generative Adversarial Networks for Physics Synthesis. Computing and Software for Big Science. 1: 4. arXiv:1701.05927. doi:10.1007/s41781-017-0004-6

    2017 arXiv

  52. [60]

    Precise Simulation of Electromagnetic Calorimeter Showers Using a Wasserstein Generative Adversarial Network

    Erdmann, M., Glombitza, J., Quast, T., 2019. Precise Simulation of Electromagnetic Calorimeter Showers Using a Wasserstein Generative Adversarial Network. Computing and Software for Big Science. 3: 4. arXiv:1807.01954

    2019 arXiv

  53. [61]

    IEEE 14th International Conference on e-Science, pp

    Salamani, D., et al., 2018, Deep generative models for fast shower simulation in ATLAS. IEEE 14th International Conference on e-Science, pp. 348-348

    2018

  54. [62]

    Semantic Segmentation using Adversarial Networks

    Luc, Pوet al., 2016. Semantic Segmentation using Adversarial Networks. NIPS Workshop on Adversarial Training, Dec, Barcelona, Spain. arXiv:1611.08408. Bibcode:2016arXiv161108408L

    2016 arXiv

  55. [63]

    2016, Generative Models, OpenAI

    Andrej Karpathy; et al. 2016, Generative Models, OpenAI

    2016

  56. [64]

    Mohamed, S., Lakshminarayanan, B. 2016. Learning in Implicit Generative Models. arXiv:1610.03483

    2016 arXiv

  57. [65]

    Bert: Pre -training of deep bidirectional transformers for language understanding

    Devlin, J., et al. Bert: Pre -training of deep bidirectional transformers for language understanding. arXiv:1810.04805 (2018)

    2018 arXiv

  58. [66]

    Incorporating uncertainty in the design of water distribution systems

    Spiliotis, M.; Tsakiris, G. Incorporating uncertainty in the design of water distribution systems. Ewra European Water 2017, 58, 449–456

    2017

  59. [67]

    E., Gneiting, T., & Kitsui, K

    Raftery, A. E., Gneiting, T., & Kitsui, K. (2005). Using Bayesian model averaging to forecast the likelihood of certain weather events. Monthly Weather Review, 133(5), 1155-1174

    2005

  60. [68]

    Distributed, Parallel, and Cluster Computing, arXiv:1603.04467

    Abadi, M., et al., 2016, TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems. Distributed, Parallel, and Cluster Computing, arXiv:1603.04467

    2016 arXiv

  61. [69]

    et al., 2019, Quantum supremacy using a programmable superconducting processor

    Arute, F., Arya, K., Babbush, R. et al., 2019, Quantum supremacy using a programmable superconducting processor. Nature 574, 505–510

    2019

  62. [70]

    Koldasbayeva, D., Tregubova, P., Gasanov, M. et al. Challenges in data -driven geospatial modeling for environmental research and practice. Nat Commun 15, 10700 (2024). https://doi.org/10.1038/s41467-024-55240- 8

    2024 doi

  63. [71]

    Tsakiris, G., Spiliotis, M., Uncertainty in the Analysis of Water Conveyance Systems, Procedia Engineering, Volume 162, 2016, 340-348, ISSN 1877-7058, https://doi.org/10.1016/j.proeng.2016.11.073

    2016 doi

  64. [72]

    Calibration and Uncertainty Analysis for Isovel Contours-based Stage-discharge Rating Curve by Sequential Uncertainty Fitting (SUFI -2) Method

    Vatanchi, S.M., Maghrebi, M.F. Calibration and Uncertainty Analysis for Isovel Contours-based Stage-discharge Rating Curve by Sequential Uncertainty Fitting (SUFI -2) Method. Water Resour Manage (2024). https://doi.org/10.1007/s11269-024-04039-9

    2024 doi

  65. [73]

    Uncertainty in the analysis of urban water supply and distribution systems

    Tsakiris, G.; Spiliotis, M. Uncertainty in the analysis of urban water supply and distribution systems. J. Hydroinformatics 2017, 19, 823–837

    2017

  66. [74]

    Uncertainty Analysis of River Water Quality Based on Stochastic Optimization of Waste Load Allocation Using the Generalized Likelihood Uncertainty Estimation Method

    Babamiri, O., Dinpashoh, Y. Uncertainty Analysis of River Water Quality Based on Stochastic Optimization of Waste Load Allocation Using the Generalized Likelihood Uncertainty Estimation Method. Water Resour Manage 38, 967–989 (2024). https://doi.org/10.1007/s11269-023-03704-9

    2024 doi

  67. [75]

    Khan, M., Kazmi, S.A.A., Qureshi, S.R. et al. Repeatability of instantaneous position in rotational motion of a submerged Savonius turbine driven by water surface waves using image processing: an experimental investigation. J. Ocean Eng. Mar. Energy 10, 859–877 (2024). https:/...

    2024 doi

  68. [76]

    2016 Jun;22(6):839-51

    Schurch, N., J., et al., How many biological replicates are needed in an RNA -seq experiment and which differential expression tool should you use? RNA. 2016 Jun;22(6):839-51. doi: 10.1261/rna.053959.115

    2016 doi

  69. [78]

    Seasonal Uncertainty Estimation of Surface Nuclear Magnetic Resonance Water Content using Bootstrap Statistics

    Singh, U., Sharma, P.K. Seasonal Uncertainty Estimation of Surface Nuclear Magnetic Resonance Water Content using Bootstrap Statistics. Water Resour Manage 36, 2493–2508 (2022). https://doi.org/10.1007/s11269-022- 03155-8

    2022 doi

  70. [79]

    & Chu, H

    Wang, Z., Si, Y. & Chu, H. Daily Streamflow Prediction and Uncertainty Using a Long Short-Term Memory (LSTM) Network Coupled with Bootstrap. Water Resour Manage 36, 4575 –4590 (2022). https://doi.org/10.1007/s11269-022-03264-4

    2022 doi

  71. [80]

    & Jiang, Y

    Chu, H., Wei, J. & Jiang, Y. Middle- and Long-Term Streamflow Forecasting and Uncertainty Analysis Using Lasso-DBN-Bootstrap Model. Water Resour Manage 35, 2617–2632 (2021). https://doi.org/10.1007/s11269- 021-02854-y

    2021 doi

  72. [81]

    Giudicianni, C., Di Cicco, I., Di Nardo, A. et al. Variance-based Global Sensitivity Analysis of Surface Runoff Parameters for Hydrological Modeling of a Real Peri-urban Ungauged Basin. Water Resour Manage 38, 3007– 3022 (2024). https://doi.org/10.1007/s11269-024-03802-2

    2024 doi

  73. [82]

    Hybrid Iterative and Tree-Based Machine Learning Algorithms for Lake Water Level Forecasting

    Fijani, E., Khosravi, K. Hybrid Iterative and Tree-Based Machine Learning Algorithms for Lake Water Level Forecasting. Water Resour Manage 37, 5431–5457 (2023). https://doi.org/10.1007/s11269-023-03613-x

    2023 doi

  74. [83]

    Montgomery, Douglas C.; Runger, George C. (2014). Applied Statistics and Probability for Engineers (6th ed.). Wiley. p. 241. ISBN 9781118539712

    2014

  75. [84]

    Razmi, A., Mardani-Fard, H.A., Golian, S. et al. Time-Varying Univariate and Bivariate Frequency Analysis of Nonstationary Extreme Sea Level for New York City. Environ. Process. 9, 8 (2022). https://doi.org/10.1007/s40710-021-00553-9

    2022 doi

  76. [85]

    & Memon, F.A

    Gao, D., Chen, A.S. & Memon, F.A. A Systematic Review of Methods for Investigating Climate Change Impacts on Water-Energy-Food Nexus. Water Resour Manage 38, 1–43 (2024). https://doi.org/10.1007/s11269-023- 03659-x

    2024 doi

  77. [86]

    Bell, G. D., M. Halpert, and M. l’Heureux. (2015) ENSO and the Tropical Pacific. In State of the Climate 2015, Chapter 4: The Tropics. H. J. Diamond and C.J. Schreck, Eds

    2015

  78. [87]

    Renwick, and A

    Mekonnen, A., J.A. Renwick, and A. Sanchez-Lugo, Eds. Chapter 7: Regional Climates. In State of the Climate 2015, Blunden, J., and D. Arndt, Eds

    2015

  79. [88]

    Afshar M. H., Naisipour M., Amani J., (2011), Node moving adaptive refinement strategy for planar elasticity problems using discrete least squares meshless method, Finite Elements in Analysis and Design, 47,12, 1315- 1325

    2011

  80. [89]

    Labibzadeh, M., et al., Efficiency test of the discrete least squares meshless method in solving heat conduction problems using error estimation, Sharif Journal of Civil Engineering, 2015; 31.2(3.2): 31-40

    2015

  81. [90]

    Mooers, G., Pritchard, M., Beucler, T., et al., Srivastava, T., 2024, Comparing storm resolving models and climates via unsupervised machine learning, Scientific Reports, vol 13, 22365

    2024

  82. [92]

    al., 2024, Climate-invariant machine learning, Science Advances, 10, 6

    Beucler, T., Gentine, P., Yuval, J., et. al., 2024, Climate-invariant machine learning, Science Advances, 10, 6

    2024

  83. [93]

    Kochkov, D., Yuval, J., Langmore, I. et al. Neural general circulation models for weather and climate. Nature 632, 1060–1066 (2024). https://doi.org/10.1038/s41586-024-07744-y

    2024 doi

  84. [94]

    Multimodal Deep Learning for Two-Year ENSO Forecast

    Naisipour, M., Saeedpanah, I., & Adib, A. Multimodal Deep Learning for Two-Year ENSO Forecast. Water Resources Management, 39, 3745–3775 (2025). https://doi.org/10.1007/s11269-025-04128-3

    2025 doi

  85. [95]

    Toward long-range ENSO prediction with an explainable deep learning model

    Chen, Q., Cui, Y., Hong, G., et al. Toward long-range ENSO prediction with an explainable deep learning model. npj Climate and Atmospheric Science, 8, 259 (2025). https://doi.org/10.1038/s41612-025-01159-w

    2025 doi

  86. [96]

    Novel Deep Learning Method for Forecasting ENSO

    Naisipour, M., Saeedpanah, I., & Adib, A. Novel Deep Learning Method for Forecasting ENSO. Journal of Hydraulic Structures, 11(3): 14–25 (2025). https://doi.org/10.22055/jhs.2024.47965.1322

    2025

  87. [97]

    Forecasting El Niño Six Months in Advance Utilizing Augmented Convolutional Neural Network

    Naisipour, M., Saeedpanah, I., Adib, A., & Pour, M.H.N. Forecasting El Niño Six Months in Advance Utilizing Augmented Convolutional Neural Network. 2024 14th International Conference on Computer and Knowledge Engineering (ICCKE), Mashhad, Iran. pp. 359–364. https://doi.org/10....

    2024

  88. [98]

    An error indicator for two-dimensional elasticity problems in the discrete least squares meshless method

    Naisipour, M., Afshar, M.H., Hassani, B., & Zeinali, M. An error indicator for two-dimensional elasticity problems in the discrete least squares meshless method. 8th International Congress on Civil Engineering, Shiraz, Iran, 2009

    2009

  89. [99]

    Efficiency Test of the Discrete Least Squares Meshless Method in Solving Heat Conduction Problems Using Error Estimation

    Labibzadeh, M., Modaresi Rajab, R., & Naisipour, M. Efficiency Test of the Discrete Least Squares Meshless Method in Solving Heat Conduction Problems Using Error Estimation. Sharif Journal of Civil Engineering, 31- 2(3.2): 31–40 (2015). Available from: https://sid.ir/paper/127886/en

    2015

  90. [100]

    Groom, M., Bassetti, D., Horenko, I., & O’Kane, T. J. Entropic learning enables skilful forecasts of ENSO phase at up to two years lead time. arXiv preprint, arXiv:2503.01412 (2025). Available at: https://arxiv.org/abs/2503.01412

    2025 arXiv

Pith tools

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