pith. sign in

arxiv: 2511.16164 · v2 · submitted 2025-11-20 · 💻 cs.LG · stat.AP

Achieving Skilled and Reliable Daily Probabilistic Forecasts of Wind Power at Subseasonal-to-Seasonal Timescales over France

Eloi Lindas , Yannig Goude , Philippe Ciais This is my paper

Pith reviewed 2026-05-17 20:45 UTC · model grok-4.3

classification 💻 cs.LG stat.AP
keywords wind power forecastingsubseasonal-to-seasonalprobabilistic forecastspost-processingECMWFrenewable energyFrancedaily resolution
0
0 comments X

The pith

A post-processing pipeline converts ECMWF subseasonal forecasts into daily wind power predictions for France that beat climatology up to 16 days ahead while staying well calibrated.

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

The paper presents a forecasting pipeline that takes subseasonal-to-seasonal weather model outputs and turns them into daily probabilistic wind power forecasts over France for lead times from one day to 46 days. A post-processing correction is then applied to the resulting power ensembles. This produces forecasts that improve on a climatological baseline, with gains ranging from 15 percent down to 5 percent on the Continuous Ranked Probability Score and from 20 percent down to 5 percent on ensemble mean squared error through day 16, before the skill returns to climatological levels. The forecasts remain near perfectly calibrated at every lead time. These longer-range predictions could let electricity market participants adjust renewable supply plans with more notice.

Core claim

A lead time and numerical weather model agnostic forecasting pipeline transforms ECMWF subseasonal-to-seasonal weather forecasts into wind power forecasts for France at daily resolution from 1 to 46 days ahead; after a post-processing step on the power ensembles, the forecasts improve the climatological baseline by 15 to 5 percent for the Continuous Ranked Probability Score and 20 to 5 percent for ensemble Mean Squared Error up to 16 days ahead, with near perfect calibration at every lead time.

What carries the argument

The lead time and numerical weather model agnostic forecasting pipeline that converts ECMWF subseasonal-to-seasonal weather forecasts into daily wind power ensembles, followed by a post-processing correction step.

If this is right

  • Electricity market players could use the extended forecast range up to two weeks to improve decision making on renewable supply.
  • Daily-resolution forecasts become feasible without requiring temporal or spatial aggregation.
  • Skill and calibration hold across the full range of lead times from one day through 46 days before the forecasts converge to climatology.

Where Pith is reading between the lines

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

  • The same pipeline could be tested on solar power or on other countries with similar wind regimes.
  • If the post-processing step generalizes, operators might reduce reliance on high-resolution local models for long-range renewable planning.
  • Integration into operational grid balancing tools becomes plausible once the calibration holds on independent recent periods.

Load-bearing premise

The post-processing correction stays effective and does not overfit on new forecast periods or modest changes in spatial aggregation, and the ECMWF subseasonal ensembles already carry usable daily surface wind information over France.

What would settle it

Apply the full pipeline to a fresh set of ECMWF forecast periods after the training window and check whether the reported skill gains and near-perfect calibration still appear at lead times up to 16 days.

Figures

Figures reproduced from arXiv: 2511.16164 by Eloi Lindas, Philippe Ciais, Yannig Goude.

Figure 1
Figure 1. Figure 1: Forecasting pipeline of the study represented schematically. S2S weather ensembles are converted to S2S Wind Power forecasts before being post-processed to account for the bias and under dispersion. The details of the weather-to-power model and post-processing model are presented in the dashed boxes. The post-processing model follows an online procedure limiting the data available for training. 2.1 S2S For… view at source ↗
Figure 2
Figure 2. Figure 2: CRP SS evolution with lead time. The lead times between 1 and 7 days are hatched to remind that short-term weather forecasts would be more suitable to provide power forecasts at such horizons. For each lead time, the CRP S is averaged over all the available samples. In terms of skill score, both MSESEns and CRP SS (see figures 3 and 2), using S2S weather forecasts instead of a simple climatology helps to i… view at source ↗
Figure 3
Figure 3. Figure 3: MSESEns evolution with lead time. The lead times between 1 and 7 days are hatched to remind that short-term weather forecasts would be more suitable to provide power forecasts at such horizons. For each lead time, the MSE is averaged over all the available samples. It is worth mentioning that more complex post-processing methods like Quantile Regression Forest or Distributional Regression Network, describe… view at source ↗
Figure 4
Figure 4. Figure 4: Reliability plots for different lead times. A perfectly calibrated forecast would align with the first bisector represented by the black dashed line. Lead times below 7 days are not represented as short-term weather forecasts would be preferred to provide the corresponding short-term power forecasts. For each lead time, the observed frequency are computed over all the available samples. The quantile discre… view at source ↗
Figure 5
Figure 5. Figure 5: CRP SS evolution with lead time. The lead times between 1 and 7 days are hatched to remind that short-term weather forecasts would be more suitable to provide power forecasts at such horizons. For each lead time, the CRP S is averaged over all the available samples. 14 [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: MSESEns evolution with lead time. The lead times between 1 and 7 days are hatched to remind that short-term weather forecasts would be more suitable to provide power forecasts at such horizons. For each lead time, the MSE is averaged over all the available samples. 15 [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Reliability plots for different lead times. A perfectly calibrated forecast would align with the first bisector represented by the black dashed line. Lead times below 7 days are not represented as short-term weather forecasts would be preferred to provide the corresponding short-term power forecasts. For each lead time, the observed frequency are computed over all the available samples.The quantile discret… view at source ↗
Figure 8
Figure 8. Figure 8: Convergence of the post-processing methods measured via the CRP S as the training set size increases. The curves represent the average across the different lead times of the CRP S given a size of the training set. A strong annual seasonality can be observed for all the methods tested in this work. Non-parametric methods such as Quantile Regression Forest or Quantile Regression Network requires more trainin… view at source ↗
read the original abstract

In a growing renewable based energy system, accurate and reliable wind power forecasts are crucial for grid stability, balancing supply and demand and market risk management. Even though short-term weather forecasts have been thoroughly used to provide up to 3 days ahead renewable power predictions, forecasts involving prediction horizons longer than a week still need investigations. Despite the recent progress in subseasonal-to-seasonal weather probabilistic forecasting, their use for wind power prediction usually involves both temporal and spatial aggregation to achieve reasonable skill. In this study, we present a lead time and numerical weather model agnostic forecasting pipeline which enables to transform ECMWF subseasonal-to-seasonal weather forecasts into wind power forecasts for France for lead times ranging from 1 day to 46 days at daily resolution. By leveraging a post-processing step of the resulting power ensembles we show that these forecasts improve the climatological baseline by 15% to 5% for the Continuous Ranked Probability Score and 20% to 5% for ensemble Mean Squared Error up to 16 days in advance, before converging towards the climatological skill. This improvement in skill is jointly obtained with near perfect calibration of the forecasts for every lead time. The results suggest that electricity market players could benefit from the extended forecast range up to two weeks to improve their decision making on renewable supply

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces a lead-time and model-agnostic pipeline that converts ECMWF subseasonal-to-seasonal ensemble forecasts into daily-resolution probabilistic wind-power forecasts over France for lead times of 1–46 days. After applying a post-processing correction to the resulting power ensembles, the authors report skill improvements relative to climatology of 15 % to 5 % in CRPS and 20 % to 5 % in ensemble MSE up to day 16, with the skill then converging to climatological levels; these gains are accompanied by near-perfect calibration at every lead time.

Significance. If the post-processing step generalizes without temporal leakage, the work would demonstrate a practical route to calibrated daily wind-power forecasts at subseasonal horizons, which is directly relevant to grid balancing and market operations in high-renewable systems. The pipeline’s claimed agnosticism to lead time and NWP model is a useful engineering contribution.

major comments (2)
  1. [Methods / Post-processing description] The manuscript provides insufficient detail on the training and validation procedure for the post-processing correction parameters. It is unclear whether these parameters were estimated on the full verification period, via a rolling scheme, or with temporal cross-validation that prevents leakage of future information into the correction; without this information the reported 15–5 % CRPS and 20–5 % EMSE improvements cannot be confidently attributed to genuine predictive skill rather than in-sample fitting.
  2. [Results / Skill scores] The central performance claims rest on the post-processing step converting marginal raw-ensemble skill into calibrated forecasts. The paper should therefore report an ablation that isolates the contribution of the correction (raw ensembles vs. post-processed) on a strictly held-out test period, together with the exact hyper-parameter selection protocol and any spatial-aggregation sensitivity tests.
minor comments (2)
  1. [Abstract and Results] The abstract states “near perfect calibration” for every lead time; the main text should include quantitative calibration diagnostics (e.g., reliability diagrams or PIT histograms) rather than qualitative statements.
  2. [Evaluation metrics] Clarify the exact definition of the climatological baseline (e.g., whether it is a rolling climatology or a fixed historical distribution) and confirm that it is computed independently of the post-processing parameters.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review. The comments have helped us identify areas where the manuscript can be strengthened, particularly regarding methodological transparency and additional analyses. We address each major comment below and indicate the revisions made.

read point-by-point responses

  1. Referee: [Methods / Post-processing description] The manuscript provides insufficient detail on the training and validation procedure for the post-processing correction parameters. It is unclear whether these parameters were estimated on the full verification period, via a rolling scheme, or with temporal cross-validation that prevents leakage of future information into the correction; without this information the reported 15–5 % CRPS and 20–5 % EMSE improvements cannot be confidently attributed to genuine predictive skill rather than in-sample fitting.

    Authors: We agree that the original description of the post-processing training procedure was insufficiently detailed. The parameters were in fact estimated using a rolling temporal cross-validation scheme in which, for each forecast issuance date, only data from prior years were used for training and hyper-parameter tuning; no information from the verification period or future dates entered the correction. We have revised Section 3.2 to provide a complete description of this procedure, including the exact length of the training window, the validation split, and explicit confirmation that temporal leakage was prevented. These clarifications should allow readers to attribute the reported skill improvements to genuine out-of-sample performance. revision: yes

  2. Referee: [Results / Skill scores] The central performance claims rest on the post-processing step converting marginal raw-ensemble skill into calibrated forecasts. The paper should therefore report an ablation that isolates the contribution of the correction (raw ensembles vs. post-processed) on a strictly held-out test period, together with the exact hyper-parameter selection protocol and any spatial-aggregation sensitivity tests.

    Authors: We acknowledge the value of an explicit ablation. In the revised manuscript we have added a new subsection (4.3) and accompanying figure that directly compares raw ECMWF ensemble power forecasts against the post-processed versions on the same strictly held-out test period used for the main results. We have also documented the hyper-parameter selection protocol (minimization of CRPS on a temporally separated validation fold) and included a sensitivity analysis showing that national-level skill scores remain stable under different spatial aggregation choices (e.g., regional vs. country-wide). These additions isolate the contribution of the post-processing step and address the referee’s request for transparency. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected in the forecasting pipeline

full rationale

The paper presents an empirical forecasting pipeline that applies post-processing to ECMWF subseasonal ensembles to generate daily wind power forecasts over France. Skill improvements are reported relative to an independent climatological baseline using standard metrics (CRPS, EMSE) and calibration checks. No self-definitional steps, fitted inputs renamed as predictions, or load-bearing self-citations appear in the abstract or described method. The post-processing is a standard correction step whose parameters, if trained with proper temporal separation or cross-validation, yield measured out-of-sample performance rather than a result forced by construction. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 0 axioms · 0 invented entities

The claim rests on the assumption that a post-processing step can extract usable daily wind-power signal from ECMWF subseasonal ensembles without introducing new fitted entities beyond standard ensemble calibration parameters.

free parameters (1)
  • post-processing correction parameters
    Parameters of the ensemble post-processing step are fitted to historical forecast-observation pairs to achieve the reported CRPS and MSE improvements.

pith-pipeline@v0.9.0 · 5542 in / 1252 out tokens · 32699 ms · 2026-05-17T20:45:10.561400+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

69 extracted references · 69 canonical work pages · 1 internal anchor

  1. [1]

    Weather and climate data for energy applications.Renewable and Sustainable Energy Reviews, 192:114247, 2024

    Amin Amin and Monjur Mourshed. Weather and climate data for energy applications.Renewable and Sustainable Energy Reviews, 192:114247, 2024. ISSN 1364-0321. doi:https://doi.org/10.1016/j.rser.2023.114247. URL https: //www.sciencedirect.com/science/article/pii/S136403212301105X

    work page doi:10.1016/j.rser.2023.114247 2024

  2. [2]

    Susana Bayo-Besteiro, Michael García-Rodríguez, Xavier Labandeira, and Juan A. Añel. Seasonal and subseasonal wind power characterization and forecasting for the iberian peninsula and the canary islands: A systematic review. International Journal of Climatology, 42(5):2601–2613, 2022. doi:https://doi.org/10.1002/joc.7359. URL https: //rmets.onlinelibrary....

    work page doi:10.1002/joc.7359 2022

  3. [3]

    Remo Beerli, Heini Wernli, and Christian M. Grams. Does the lower stratosphere provide predictability for month-ahead wind electricity generation in Europe?Quarterly Journal of the Royal Meteorological Society, 143(709):3025–3036, October

    work page

  4. [4]

    doi:10.1002/qj.3158

    ISSN 0035-9009, 1477-870X. doi:10.1002/qj.3158. URL https://rmets.onlinelibrary.wiley.com/doi/10. 1002/qj.3158

    work page doi:10.1002/qj.3158

  5. [5]

    Bett, Hazel E

    Philip E. Bett, Hazel E. Thornton, Alberto Troccoli, Matteo De Felice, Emma Suckling, Laurent Dubus, Yves-Marie Saint-Drenan, and David J. Brayshaw. A simplified seasonal forecasting strategy, applied to wind and solar power in Europe.Climate Services, 27:100318, August 2022. ISSN 24058807. doi:10.1016/j.cliser.2022.100318. URL https: //linkinghub.elsevie...

    work page doi:10.1016/j.cliser.2022.100318 2022

  6. [6]

    H. C. Bloomfield, C. C. Suitters, and D. R. Drew. Meteorological Drivers of European Power System Stress.Journal of Renewable Energy, 2020:1–12, August 2020. ISSN 2314-4386, 2314-4394. doi:10.1155/2020/5481010. URL https: //www.hindawi.com/journals/jre/2020/5481010/. Publisher: Wiley

    work page doi:10.1155/2020/5481010 2020

  7. [7]

    Bloomfield, David J

    Hannah C. Bloomfield, David J. Brayshaw, Paula L. M. Gonzalez, and Andrew Charlton-Perez. Sub-seasonal forecasts of demand, wind power and solar powergeneration for 28 European Countries, January 2021. URL https://essd. copernicus.org/articles/13/2259/2021/essd-13-2259-2021-discussion.html

    work page 2021

  8. [8]

    How to use expert advice

    Nicolò Cesa-Bianchi, David Helmbold, David Haussler, Robert Schapire, and Manfred Warmuth. How to use expert advice. volume 44, pages 382–391, 01 1993. doi:10.1145/167088.167198

    work page doi:10.1145/167088.167198 1993

  9. [9]

    Validation of the medium-range and sub-seasonal fore- cast of solar irradiance and wind speed using ecmwf.Energy Reports, 10:3908–3913, 2023

    Veeranjaneyulu Chinta, Guiting Song, and Wei Zhang. Validation of the medium-range and sub-seasonal fore- cast of solar irradiance and wind speed using ecmwf.Energy Reports, 10:3908–3913, 2023. ISSN 2352-4847. doi:https://doi.org/10.1016/j.egyr.2023.10.058. URL https://www.sciencedirect.com/science/article/pii/ S2352484723014907

    work page doi:10.1016/j.egyr.2023.10.058 2023

  10. [10]

    Chiffres clés du parc de production d’electricité français en

    Réseau de Transport de l’Electricité (RTE). Chiffres clés du parc de production d’electricité français en

    work page

  11. [11]

    Accessed: September 15, 2025

    https://assets.rte-france.com/analyse-et-donnees/2025-03/BE2024%20-%20Fiche%20Parc%20de% 20Production.pdf, 2025. Accessed: September 15, 2025

    work page 2025

  12. [12]

    An efficient multi quantile regression network with ad hoc prevention of quantile crossing, 2024

    Jens Decke, Arne Jenß, Bernhard Sick, and Christian Gruhl. An efficient multi quantile regression network with ad hoc prevention of quantile crossing, 2024. URLhttps://arxiv.org/abs/2406.00080

    work page arXiv 2024

  13. [13]

    Joshua Dorrington, Isla Finney, Tim Palmer, and Antje Weisheimer. Beyond skill scores: exploring sub-seasonal forecast value through a case-study of french month-ahead energy prediction.Quarterly Journal of the Royal Meteorological Society, 146(733):3623–3637, 2020. doi:https://doi.org/10.1002/qj.3863. URL https://rmets.onlinelibrary.wiley.com/ doi/abs/10...

    work page doi:10.1002/qj.3863 2020

  14. [14]

    Space-Time Variability of Climate Variables and Intermittent Renewable Electricity Production --

    Kolbjørn Engeland, Marco Borga, Jean-Dominique Creutin, Baptiste François, Maria-Helena Ramos, and Jean-Philippe Vidal. Space-time variability of climate variables and intermittent renewable electricity production – a review.Renewable and Sustainable Energy Reviews, 79:600–617, 2017. ISSN 1364-0321. doi:https://doi.org/10.1016/j.rser.2017.05.046. URL http...

    work page doi:10.1016/j.rser.2017.05.046 2017

  15. [15]

    Schapire, Yoram Singer, and Manfred K

    Yoav Freund, Robert E. Schapire, Yoram Singer, and Manfred K. Warmuth. Using and combining predictors that specialize. InSymposium on the Theory of Computing, 1997. URLhttps://api.semanticscholar.org/CorpusID:156255

    work page 1997

  16. [16]

    opera: Online prediction by expert aggregation.URL: https://CRAN

    Pierre Gaillard, Yannig Goude, Laurent Plagne, Thibaut Dubois, and Benoit Thieurmel. opera: Online prediction by expert aggregation.URL: https://CRAN. R-project. org/package= opera. r package version, 1, 2016

    work page 2016

  17. [17]

    Probabilistic Forecasting.Annual Review of Statistics and Its Application, 1 (1):125–151, January 2014

    Tilmann Gneiting and Matthias Katzfuss. Probabilistic Forecasting.Annual Review of Statistics and Its Application, 1 (1):125–151, January 2014. ISSN 2326-8298, 2326-831X. doi:10.1146/annurev-statistics-062713-085831. URL https: //www.annualreviews.org/doi/10.1146/annurev-statistics-062713-085831

    work page doi:10.1146/annurev-statistics-062713-085831 2014

  18. [18]

    Strictly

    Tilmann Gneiting and Adrian E Raftery. Strictly Proper Scoring Rules, Prediction, and Estimation.Journal of the American Statistical Association, 102(477):359–378, March 2007. ISSN 0162-1459, 1537-274X. doi:10.1198/016214506000001437. URLhttp://www.tandfonline.com/doi/abs/10.1198/016214506000001437. Publisher: Informa UK Limited

    work page doi:10.1198/016214506000001437 2007

  19. [19]

    Raftery, Anton H

    Tilmann Gneiting, Adrian E. Raftery, Anton H. Westveld, and Tom Goldman. Calibrated Probabilistic Forecasting Using Ensemble Model Output Statistics and Minimum CRPS Estimation.Monthly Weather Review, 133(5):1098–1118, May 18 Daily S2S Probabilistic Forecasts of French Wind PowerPREPRINT

    work page

  20. [20]

    doi:10.1175/MWR2904.1

    ISSN 1520-0493, 0027-0644. doi:10.1175/MWR2904.1. URL http://journals.ametsoc.org/doi/10.1175/ MWR2904.1

    work page doi:10.1175/mwr2904.1

  21. [21]

    Intermittency and the Value of Renewable Energy, 2011

    Gautam Gowrisankaran, Stanley S Reynolds, and Mario Samano. Intermittency and the Value of Renewable Energy, 2011

    work page 2011

  22. [22]

    Methods to extrapolate wind resource to the turbine hub height based on power law: A 1- h wind speed vs

    Giovanni Gualtieri and Sauro Secci. Methods to extrapolate wind resource to the turbine hub height based on power law: A 1- h wind speed vs. Weibull distribution extrapolation comparison.Renewable Energy, 43:183–200, July 2012. ISSN 09601481. doi:10.1016/j.renene.2011.12.022. URLhttps://linkinghub.elsevier.com/retrieve/pii/S0960148112000109

    work page doi:10.1016/j.renene.2011.12.022 2012

  23. [23]

    Hans Hersbach, Bill Bell, Paul Berrisford, Shoji Hirahara, András Horányi, Joaquín Muñoz-Sabater, Julien Nicolas, Carole Peubey, Raluca Radu, Dinand Schepers, Adrian Simmons, Cornel Soci, Saleh Abdalla, Xavier Abellan, Gianpaolo Balsamo, Peter Bechtold, Gionata Biavati, Jean Bidlot, Massimo Bonavita, Giovanna De Chiara, Per Dahlgren, Dick Dee, Michail Dia...

    work page doi:10.1002/qj.3803 1999

  24. [24]

    Deep Learning for Postprocessing Global Probabilistic Forecasts on Subseasonal Time Scales.MONTHLY WEATHER REVIEW, 152, 2024

    Nina Horat and Sebastian Lerch. Deep Learning for Postprocessing Global Probabilistic Forecasts on Subseasonal Time Scales.MONTHLY WEATHER REVIEW, 152, 2024

    work page 2024

  25. [25]

    Improving Model Chain Approaches for Probabilistic Solar Energy Forecasting through Post-processing and Machine Learning.Advances in Atmospheric Sciences, 42(2):297–312, February

    Nina Horat, Sina Klerings, and Sebastian Lerch. Improving Model Chain Approaches for Probabilistic Solar Energy Forecasting through Post-processing and Machine Learning.Advances in Atmospheric Sciences, 42(2):297–312, February

    work page

  26. [26]

    doi:10.1007/s00376-024-4219-2

    ISSN 0256-1530, 1861-9533. doi:10.1007/s00376-024-4219-2. URL https://link.springer.com/10.1007/ s00376-024-4219-2

    work page doi:10.1007/s00376-024-4219-2

  27. [27]

    Renewable energy prospects for central and south- eastern europe energy connectivity (cesec)

    International Renewable Energy Agency (IRENA). Renewable energy prospects for central and south- eastern europe energy connectivity (cesec). https://www.irena.org/Publications/2020/Oct/ Renewable-Energy-Prospects-for-Central-and-South-Eastern-Europe-Energy-Connectivity-CESEC ,

    work page 2020

  28. [28]

    Accessed: August 26, 2024

    work page 2024

  29. [29]

    Systematic review of nowcasting approaches for solar energy production based upon ground-based cloud imaging.Solar Energy Advances, 2:100019, 2022

    Bruno Juncklaus Martins, Allan Cerentini, Sylvio Luiz Mantelli, Thiago Zimmermann Loureiro Chaves, Nicolas Moreira Branco, Aldo von Wangenheim, Ricardo Rüther, and Juliana Marian Arrais. Systematic review of nowcasting approaches for solar energy production based upon ground-based cloud imaging.Solar Energy Advances, 2:100019, 2022. ISSN 2667-

    work page 2022

  30. [30]

    URL https://www.sciencedirect.com/science/article/ pii/S2667113122000079

    doi:https://doi.org/10.1016/j.seja.2022.100019. URL https://www.sciencedirect.com/science/article/ pii/S2667113122000079

    work page doi:10.1016/j.seja.2022.100019 2022

  31. [31]

    C. G. Justus and Amir Mikhail. Height variation of wind speed and wind distributions statistics.Geophysical Research Letters, 3(5):261–264, May 1976. ISSN 0094-8276, 1944-8007. doi:10.1029/GL003i005p00261. URL https://agupubs. onlinelibrary.wiley.com/doi/10.1029/GL003i005p00261

    work page doi:10.1029/gl003i005p00261 1976

  32. [32]

    Daily prediction of solar power generation based on weather forecast information in korea.IET Renewable Power Generation, 11(10):1268–1273, 2017

    Jae-Gon Kim, Dong-Hyuk Kim, Woo-Sik Yoo, Joung-Yun Lee, and Yong Bae Kim. Daily prediction of solar power generation based on weather forecast information in korea.IET Renewable Power Generation, 11(10):1268–1273, 2017. ISSN 1752-1416, 1752-1424. doi:10.1049/iet-rpg.2016.0698. URL https://onlinelibrary.wiley.com/doi/10. 1049/iet-rpg.2016.0698

    work page doi:10.1049/iet-rpg.2016.0698 2017

  33. [33]

    Econometric Society Monographs

    Roger Koenker.Quantile Regression. Econometric Society Monographs. Cambridge University Press, 2005

    work page 2005

  34. [34]

    Forecasting solar energy production: A comparative study of machine learning algorithms.Energy Reports, 10:1004–1012,

    Younes Ledmaoui, Adila El Maghraoui, Mohamed El Aroussi, Rachid Saadane, Ahmed Chebak, and Abdellah Chehri. Forecasting solar energy production: A comparative study of machine learning algorithms.Energy Reports, 10:1004–1012,

    work page

  35. [35]

    doi:10.1016/j.egyr.2023.07.042

    ISSN 23524847. doi:10.1016/j.egyr.2023.07.042. URL https://linkinghub.elsevier.com/retrieve/pii/ S2352484723011228

    work page doi:10.1016/j.egyr.2023.07.042 2023

  36. [36]

    Eloi Lindas, Yannig Goude, and Philippe Ciais. Toward accurate forecasting of renewable energy: Building datasets and benchmarking machine learning models for solar and wind power in france.Environmental Data Science, 4:e45, 2025. doi:10.1017/eds.2025.10021

    work page doi:10.1017/eds.2025.10021 2025

  37. [37]

    Lledó, V

    Ll. Lledó, V . Torralba, A. Soret, J. Ramon, and F.J. Doblas-Reyes. Seasonal forecasts of wind power generation.Renewable Energy, 143:91–100, December 2019. ISSN 09601481. doi:10.1016/j.renene.2019.04.135. URL https://linkinghub. elsevier.com/retrieve/pii/S0960148119306196

    work page doi:10.1016/j.renene.2019.04.135 2019

  38. [38]

    Lynch, David J

    Kieran J. Lynch, David J. Brayshaw, and Andrew Charlton-Perez. Verification of European Subseasonal Wind Speed Forecasts.Monthly Weather Review, 142(8):2978–2990, August 2014. ISSN 0027-0644, 1520-0493. doi:10.1175/MWR-D- 13-00341.1. URLhttp://journals.ametsoc.org/doi/10.1175/MWR-D-13-00341.1

    work page doi:10.1175/mwr-d- 2014

  39. [39]

    The new ECMWF interpolation package MIR, 2017

    Pedro Maciel, Tiago Quintino, Umberto Modigliani, Paul Dando, Baudouin Raoult, Willem Deconinck, Florian Rathgeber, and Cristian Simarro. The new ECMWF interpolation package MIR, 2017. URL https://www.ecmwf.int/node/18197. Pages: 36-39 Publication Title: ECMWF newsletter Section: Computing

    work page 2017

  40. [40]

    Quantile Regression Forests.Journal of Machine Learning Research, 2006

    Nicolai Meinshausen. Quantile Regression Forests.Journal of Machine Learning Research, 2006. 19 Daily S2S Probabilistic Forecasts of French Wind PowerPREPRINT

    work page 2006

  41. [41]

    SubseasonalClimateUSA: A Dataset for Subseasonal Forecasting and Benchmarking, 2023

    Soukayna Mouatadid, Paulo Orenstein, Genevieve Flaspohler, Miruna Oprescu, Judah Cohen, Franklyn Wang, Sean Knight, Maria Geogdzhayeva, Sam Levang, Ernest Fraenkel, and Lester Mackey. SubseasonalClimateUSA: A Dataset for Subseasonal Forecasting and Benchmarking, 2023

    work page 2023

  42. [42]

    Better seasonal forecasts for the renewable energy industry.Nature Energy, 5(2):108–110, February 2020

    Anton Orlov, Jana Sillmann, and Ilaria Vigo. Better seasonal forecasts for the renewable energy industry.Nature Energy, 5(2):108–110, February 2020. ISSN 2058-7546. doi:10.1038/s41560-020-0561-5. URL https://www.nature.com/ articles/s41560-020-0561-5

    work page doi:10.1038/s41560-020-0561-5 2020

  43. [43]

    Impact of forecasting on energy system optimization.Advances in Applied Energy, 15:100181, 2024

    Florian Peterssen, Marlon Schlemminger, Clemens Lohr, Raphael Niepelt, Richard Hanke-Rauschenbach, and Rolf Brendel. Impact of forecasting on energy system optimization.Advances in Applied Energy, 15:100181, 2024. ISSN 2666-7924. doi:https://doi.org/10.1016/j.adapen.2024.100181. URL https://www.sciencedirect.com/science/article/pii/ S2666792424000192

    work page doi:10.1016/j.adapen.2024.100181 2024

  44. [44]

    Evaluating ensemble post-processing for wind power forecasts.Wind Energy, 25(8):1379–1405, 2022

    Kaleb Phipps, Sebastian Lerch, Maria Andersson, Ralf Mikut, Veit Hagenmeyer, and Nicole Ludwig. Evaluating ensemble post-processing for wind power forecasts.Wind Energy, 25(8):1379–1405, 2022. doi:https://doi.org/10.1002/we.2736. URL https://onlinelibrary.wiley.com/doi/abs/10.1002/we.2736

    work page doi:10.1002/we.2736 2022

  45. [45]

    Optimal energy transition with variable and intermittent renewable electricity generation.Journal of Economic Dynamics and Control, 134:104273, 2022

    Aude Pommeret and Katheline Schubert. Optimal energy transition with variable and intermittent renewable electricity generation.Journal of Economic Dynamics and Control, 134:104273, 2022. ISSN 0165-1889. doi:https://doi.org/10.1016/j.jedc.2021.104273. URL https://www.sciencedirect.com/science/article/pii/ S0165188921002086

    work page doi:10.1016/j.jedc.2021.104273 2022

  46. [46]

    Comparison of Model Output Statistics and Neural Networks to Postprocess Wind Gusts, January 2024

    Cristina Primo, Benedikt Schulz, Sebastian Lerch, and Reinhold Hess. Comparison of Model Output Statistics and Neural Networks to Postprocess Wind Gusts, January 2024. URL http://arxiv.org/abs/2401.11896. arXiv:2401.11896 [stat]

    work page arXiv 2024

  47. [47]

    Using Bayesian Model Averaging to Calibrate Forecast Ensembles.MONTHLY WEATHER REVIEW, 133, 2005

    Adrian E Raftery, Tilmann Gneiting, Fadoua Balabdaoui, and Michael Polakowski. Using Bayesian Model Averaging to Calibrate Forecast Ensembles.MONTHLY WEATHER REVIEW, 133, 2005

    work page 2005

  48. [48]

    Neural networks for post-processing ensemble weather forecasts

    Stephan Rasp and Sebastian Lerch. Neural networks for post-processing ensemble weather forecasts.Monthly Weather Review, 146(11):3885–3900, November 2018. ISSN 0027-0644, 1520-0493. doi:10.1175/MWR-D-18-0187.1. URL http://arxiv.org/abs/1805.09091. arXiv:1805.09091 [stat]

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1175/mwr-d-18-0187.1 2018

  49. [49]

    J. Ryu, H. Kim, S.-Y . S. Wang, and J.-H. Yoon. Increasing resolution and accuracy in sub-seasonal forecasting through 3d u-net: the western us.EGUsphere, 2025:1–18, 2025. doi:10.5194/egusphere-2025-308. URL https://egusphere. copernicus.org/preprints/2025/egusphere-2025-308/

    work page doi:10.5194/egusphere-2025-308 2025

  50. [50]

    Machine learning methods for postprocessing ensemble forecasts of wind gusts: A systematic comparison.Monthly Weather Review, 150(1):235–257, January 2022

    Benedikt Schulz and Sebastian Lerch. Machine learning methods for postprocessing ensemble forecasts of wind gusts: A systematic comparison.Monthly Weather Review, 150(1):235–257, January 2022. ISSN 0027-0644, 1520-0493. doi:10.1175/MWR-D-21-0150.1. URLhttp://arxiv.org/abs/2106.09512. arXiv:2106.09512 [stat]

    work page doi:10.1175/mwr-d-21-0150.1 2022

  51. [51]

    A review on seasonal climate prediction.Advances in Atmospheric Sciences, 18:197–208, 03 2001

    Wang Shaowu and Zhu Jinhong. A review on seasonal climate prediction.Advances in Atmospheric Sciences, 18:197–208, 03 2001. doi:10.1007/s00376-001-0013-5

    work page doi:10.1007/s00376-001-0013-5 2001

  52. [52]

    Predicting solar generation from weather forecasts using machine learning

    Navin Sharma, Pranshu Sharma, David Irwin, and Prashant Shenoy. Predicting solar generation from weather forecasts using machine learning. In2011 IEEE International Conference on Smart Grid Communications (SmartGridComm), pages 528–533. IEEE, 2011. ISBN 978-1-4577-1702-4 978-1-4577-1704-8. doi:10.1109/SmartGridComm.2011.6102379. URL http://ieeexplore.ieee...

    work page doi:10.1109/smartgridcomm.2011.6102379 2011

  53. [53]

    Subseasonal solar power forecasting via deep sequence learning

    Saumya Sinha, Bri-Mathias S Hodge, and Claire Monteleoni. Subseasonal solar power forecasting via deep sequence learning. InNeurIPS 2021 Workshop on Tackling Climate Change with Machine Learning, 2021. URL https://www. climatechange.ai/papers/neurips2021/23

    work page 2021

  54. [54]

    Climate change impacts on renewable energy generation

    Kepa Solaun and Emilio Cerdá. Climate change impacts on renewable energy generation. a review of quantitative projections.Renewable and Sustainable Energy Reviews, 116:109415, 2019. ISSN 1364-0321. doi:https://doi.org/10.1016/j.rser.2019.109415. URL https://www.sciencedirect.com/science/article/pii/ S1364032119306239

    work page doi:10.1016/j.rser.2019.109415 2019

  55. [55]

    Sub-seasonal to seasonal climate predictions for wind energy forecasting.Journal of Physics: Conference Series, 1222(1):012009, May 2019

    A Soret, V Torralba, N Cortesi, I Christel, Ll Palma, A Manrique-Suñén, Ll Lledó, N González-Reviriego, and F J Doblas-Reyes. Sub-seasonal to seasonal climate predictions for wind energy forecasting.Journal of Physics: Conference Series, 1222(1):012009, May 2019. ISSN 1742-6588, 1742-6596. doi:10.1088/1742-6596/1222/1/012009. URL https: //iopscience.iop.o...

    work page doi:10.1088/1742-6596/1222/1/012009 2019

  56. [56]

    An application of the ECMWF Ensemble Predic- tion System for short-term solar power forecasting.Solar Energy, 133:437–450, August 2016

    Simone Sperati, Stefano Alessandrini, and Luca Delle Monache. An application of the ECMWF Ensemble Predic- tion System for short-term solar power forecasting.Solar Energy, 133:437–450, August 2016. ISSN 0038092X. doi:10.1016/j.solener.2016.04.016. URL https://linkinghub.elsevier.com/retrieve/pii/S0038092X1630041X

    work page doi:10.1016/j.solener.2016.04.016 2016

  57. [57]

    From research to applications – examples of operational ensemble post-processing in France using machine learning.Nonlinear Processes in Geophysics, 27(2):329–347, May 2020

    Maxime Taillardat and Olivier Mestre. From research to applications – examples of operational ensemble post-processing in France using machine learning.Nonlinear Processes in Geophysics, 27(2):329–347, May 2020. ISSN 1607-7946. doi:10.5194/npg-27-329-2020. URLhttps://npg.copernicus.org/articles/27/329/2020/. 20 Daily S2S Probabilistic Forecasts of French ...

    work page doi:10.5194/npg-27-329-2020 2020

  58. [58]

    Calibrated Ensemble Forecasts Using Quantile Regression Forests and Ensemble Model Output Statistics.Monthly Weather Review, 144(6):2375–2393, June 2016

    Maxime Taillardat, Olivier Mestre, Michaël Zamo, and Philippe Naveau. Calibrated Ensemble Forecasts Using Quantile Regression Forests and Ensemble Model Output Statistics.Monthly Weather Review, 144(6):2375–2393, June 2016. ISSN 0027-0644, 1520-0493. doi:10.1175/MWR-D-15-0260.1. URL http://journals.ametsoc.org/doi/10.1175/ MWR-D-15-0260.1

    work page doi:10.1175/mwr-d-15-0260.1 2016

  59. [59]

    A Quantile Regression Neural Network Approach to Estimating.Journal of F orecasting, 19(4):273–293, 2000

    James W Taylor. A Quantile Regression Neural Network Approach to Estimating.Journal of F orecasting, 19(4):273–293, 2000

    work page 2000

  60. [60]

    Thorarinsdottir and Tilmann Gneiting

    Thordis L. Thorarinsdottir and Tilmann Gneiting. Probabilistic Forecasts of Wind Speed: Ensemble Model Output Statistics by using Heteroscedastic Censored Regression.Journal of the Royal Statistical Society Series A: Statistics in Society, 173(2):371–388, April 2010. ISSN 0964-1998, 1467-985X. doi:10.1111/j.1467-985X.2009.00616.x. URL https://academic.oup...

    work page doi:10.1111/j.1467-985x.2009.00616.x 2010

  61. [61]

    The influence of weather regimes on European renewable energy production and demand.Environmental Research Letters, 14(9):094010, September 2019

    Karin Van Der Wiel, Hannah C Bloomfield, Robert W Lee, Laurens P Stoop, Russell Blackport, James A Screen, and Frank M Selten. The influence of weather regimes on European renewable energy production and demand.Environmental Research Letters, 14(9):094010, September 2019. ISSN 1748-9326. doi:10.1088/1748-9326/ab38d3. URL https: //iopscience.iop.org/articl...

    work page doi:10.1088/1748-9326/ab38d3 2019

  62. [62]

    Ensemble post-processing using member-by-member approaches: theo- retical aspects.Quarterly Journal of the Royal Meteorological Society, 141(688):807–818, April 2015

    Bert Van Schaeybroeck and Stéphane Vannitsem. Ensemble post-processing using member-by-member approaches: theo- retical aspects.Quarterly Journal of the Royal Meteorological Society, 141(688):807–818, April 2015. ISSN 0035-9009, 1477-870X. doi:10.1002/qj.2397. URL https://rmets.onlinelibrary.wiley.com/doi/10.1002/qj.2397. Pub- lisher: Wiley

    work page doi:10.1002/qj.2397 2015

  63. [63]

    Wilks, and Jakob W

    Stéphane Vannitsem, Daniel S. Wilks, and Jakob W. Messner, editors.Statistical postprocessing of ensemble forecasts. Elsevier, Amsterdam, Netherlands Oxford, United Kingdom Cambridge, MA, United States, 2018. ISBN 978-0-12-812372-0

    work page 2018

  64. [64]

    Stéphane Vannitsem, John Bjørnar Bremnes, Jonathan Demaeyer, Gavin R. Evans, Jonathan Flowerdew, Stephan Hemri, Sebastian Lerch, Nigel Roberts, Susanne Theis, Aitor Atencia, Zied Ben Bouallègue, Jonas Bhend, Markus Dabernig, Lesley De Cruz, Leila Hieta, Olivier Mestre, Lionel Moret, Iris Odak Plenkovi´c, Maurice Schmeits, Maxime Taillardat, Joris Van Den ...

    work page doi:10.1175/bams-d-19-0308.1 2021

  65. [65]

    Vitart, C

    F. Vitart, C. Ardilouze, A. Bonet, A. Brookshaw, M. Chen, C. Codorean, M. Déqué, L. Ferranti, E. Fucile, M. Fuentes, H. Hendon, J. Hodgson, H.-S. Kang, A. Kumar, H. Lin, G. Liu, X. Liu, P. Malguzzi, I. Mallas, M. Manoussakis, D. Mas- trangelo, C. MacLachlan, P. McLean, A. Minami, R. Mladek, T. Nakazawa, S. Najm, Y . Nie, M. Rixen, A. W. Robertson, P. Ruti...

    work page doi:10.1175/bams-d-16-0017.1 2017

  66. [66]

    Aggregating strategies

    Vladimir V ovk. Aggregating strategies. InAnnual Conference Computational Learning Theory, 1990. URL https: //api.semanticscholar.org/CorpusID:5226318

    work page 1990

  67. [67]

    A review of deep learning for renewable energy forecasting.Energy Conversion and Management, 198:111799, 2019

    Huaizhi Wang, Zhenxing Lei, Xian Zhang, Bin Zhou, and Jianchun Peng. A review of deep learning for renewable energy forecasting.Energy Conversion and Management, 198:111799, 2019. ISSN 01968904. doi:10.1016/j.enconman.2019.111799. URLhttps://linkinghub.elsevier.com/retrieve/pii/S0196890419307812

    work page doi:10.1016/j.enconman.2019.111799 2019

  68. [68]

    White, Henrik Carlsen, Andrew W

    Christopher J. White, Henrik Carlsen, Andrew W. Robertson, Richard J.T. Klein, Jeffrey K. Lazo, Arun Kumar, Frederic Vitart, Erin Coughlan de Perez, Andrea J. Ray, Virginia Murray, Sukaina Bharwani, Dave MacLeod, Rachel James, Lora Fleming, Andrew P. Morse, Bernd Eggen, Richard Graham, Erik Kjellström, Emily Becker, Kathleen V . Pegion, Neil J. Holbrook, ...

    work page

  69. [69]

    URL https://rmets.onlinelibrary.wiley.com/doi/abs/10.1002/ met.1654

    doi:https://doi.org/10.1002/met.1654. URL https://rmets.onlinelibrary.wiley.com/doi/abs/10.1002/ met.1654. 21

    work page doi:10.1002/met.1654