Post-processing of ensemble photovoltaic power forecasts with distributional and quantile regression methods

\'Agnes Baran; Dazhi Yang; Martin J\'anos Mayer; Nina Horat; S\'andor Baran; Sebastian Lerch

REVIEW 3 major objections 3 minor 82 references

Post-processing of ensemble photovoltaic power forecasts with distributional and quantile regression methods

T0 review · 3 major / 3 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read The paper claims that for seven Hungarian utility-scale PV plants, every one of the seven tested post-processing methods outperforms the raw ensemble forecast, with nonlinear quantile regression models giving the best probabilistic predicti

desk verdict Plausible benchmark comparison of PV post-processing methods; central claims are credible but unverifiable from the corrupted full text provided, and the paper deserves a real referee. read the letter →

arxiv 2508.15508 v1 pith:ERINJQKC submitted 2025-08-21 stat.AP

Martin János Mayer , Ágnes Baran , Sebastian Lerch , Nina Horat , Dazhi Yang , Sándor Baran This is my paper · ORCID

classification stat.AP

keywords photovoltaicpowerforecastingensemblepost-processingquantileregressiondistributionalprobabilisticforecastcalibrationmachinelearningsolarenergy

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

This paper tests whether post-processing ensemble photovoltaic power forecasts with statistical and machine-learning methods improves accuracy relative to the raw ensemble, using data from seven utility-scale solar plants in Hungary. The authors aim to establish three claims: any of the seven post-processing methods improves predictive performance; non-parametric methods outperform parametric models; and advanced nonlinear quantile regression models, especially machine-learning variants, give the best results. If the claims hold, grid operators and energy traders can improve short-term solar power forecasts without changing the numerical weather model, simply by applying flexible quantile regression to the existing ensemble output.

What carries the argument

The central mechanism is post-processing as a function that maps raw ensemble members and covariates to a predictive distribution or a set of quantiles of PV power. The comparison is carried by proper scoring rules such as CRPS and quantile/pinball-style losses, which reward forecasts that are both sharp and calibrated. The seven methods differ in how flexible this mapping is, from parametric distributional regression to nonlinear quantile regression and machine-learning models, and the paper attributes the performance ranking to that flexibility.

What would settle it

Re-run all seven methods on an independent held-out year for the same plants, or on a new set of PV plants, using identical training windows and no additional tuning; if the raw ensemble achieves comparable CRPS or quantile scores to the post-processed methods, or if a parametric method outperforms the nonlinear quantile models, the paper's central ranking collapses.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is an empirical ranking based on out-of-sample evaluation: over the Hungarian plant data, every form of statistical post-processing of ensemble PV power forecasts yields better probabilistic forecasts than the raw ensemble. Within the tested methods, non-parametric approaches beat parametric distributional models, and nonlinear quantile regression methods—particularly machine-learning-based ones—show the best predictive performance. The paper presents this ranking as practical guidance for choosing post-processing methods in operational solar forecasting.

Load-bearing premise

The ranking holds only if all seven methods were trained and evaluated out-of-sample under identical conditions, with no method tuned on the verification period and no selection of plants or periods that favours the reported ordering.

Editorial extensions

If this is right

All seven post-processing methods can replace raw ensemble outputs at the studied plants with measurable gains in probabilistic forecast quality.
Non-parametric and machine-learning quantile methods appear to be the more reliable first choices for PV post-processing over parametric distributional models.
The same evaluation protocol can be transferred to other plants, seasons, and ensemble systems without requiring a new weather model.
The empirical ranking gives operators a concrete starting point when choosing between computational simplicity and forecast skill.
The gain from post-processing is likely to be largest where the model chain injects systematic bias—so the same methods should help in other solar forecasting chains with similar error structures.

Reading between the lines

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

The performance ranking may depend on plant-specific characteristics, such as cloud climatology and inverter saturation; nonlinear quantile models could show larger gains at plants with more nonlinear power curves.
A direct extension would be to test whether the best method stays best at longer lead times or when forecasts are aggregated across many plants, where smoothing might narrow the gaps.
For practical deployment, computational cost and training-data requirements may favour simpler methods if the skill gap is small; the paper's ranking is about forecast skill, not cost-adjusted utility.
A useful falsifying check would be to run the same seven methods on an independent set of PV plants in a different climate and see whether the ranking order persists.

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.

Desk Editor's Note

read the letter

Two things you should know. First, the abstract is a coherent, plausible benchmark claim: seven post-processing methods on seven Hungarian PV plants, with nonlinear quantile regression winning and any post-processing beating raw ensembles. Second, the full text we received is corrupted—mojibake—so none of the supporting protocol is visible to me. That means the headline ranking is credible but unaudited in the material I had.

What the paper does well: it addresses a practical problem that matters. PV forecasting is operationally relevant, and systematic comparisons of state-of-the-art post-processing methods on multiple plants are useful to practitioners. The method list is sensible: parametric, non-parametric, statistical, ML. The authors are established in this literature, and the abstract's directional claims align with the broader trend favoring non-parametric and ML approaches. Nothing in the abstract suggests the result is built in; this is a normal empirical benchmark. If the actual full text is intact, this should be a solid contribution—a benchmark, not a new method, but benchmarks have value.

Soft spots: I can't check the evaluation design. The stress-test concern about fair comparison—same training/validation windows, no tuning on the test period, paired significance tests—is legitimate. The abstract says 'significantly improves' and 'outperform' without showing the statistical tests. That could be fine or not; I can't tell. Also, the full text corruption might be our pipeline's problem, not the authors', so I'm not holding it against them. One more minor point: the claims are plant- and region-specific; the abstract doesn't overclaim global generality, which is good.

Who this is for: anyone working on probabilistic forecasting post-processing, especially for solar power, and practitioners in Hungary or similar mid-latitude markets. It's a benchmark paper, not a theoretical breakthrough. If the full text holds up, it's a useful data point.

My recommendation: send it to peer review. The topic is important, the authors are credible, and the claims are specific and testable. A referee with the readable text can verify the protocol. If the evaluation design is sound, it's publishable. If not, the referee will catch it. I'd be surprised if this is a desk reject.

I would accept a review invitation if asked, and I'd bring it up in our reading group once I can see the actual full text.

Referee Report

3 major / 3 minor

Summary. The paper describes a systematic comparison of seven statistical post-processing methods for ensemble photovoltaic (PV) power forecasts, applied to data from seven utility-scale PV plants in Hungary. The abstract claims that every tested post-processing method significantly improves predictive performance over the raw ensemble, that non-parametric methods outperform parametric methods, that nonlinear quantile regression models perform best, and that machine-learning-based approaches outperform traditional statistical approaches. The full text supplied for review is almost entirely corrupted and unreadable: the sections on data, methods, experimental setup, results, and conclusions are present only as mojibake/placeholder-like text. Only the abstract and fragments of figures/tables can be discerned. Consequently, none of the evaluation protocol or quantitative results underlying the abstract's claims can be audited.

Significance. If the claims are correct, the paper would provide a useful empirical benchmark for PV power forecast post-processing, comparing a broad set of modern methods on a multi-site dataset. The comparative design described in the abstract is appropriate for such a study, and no circularity is apparent. However, the significance of the contribution cannot be assessed from the submitted version: the evidence supporting the central claims—the experimental design, the verification metrics, the statistical tests, and the numerical results—is completely invisible to the reviewer. No machine-checked proofs, reproducible code, or other verifiable artifacts are visible. The importance of the topic and the breadth of the comparison are the only assessable strengths.

major comments (3)

[Full text (all sections after the abstract)] The supplied manuscript is unreadable because of corrupted encoding. The evaluation protocol—data description, training/validation/test splits, forecast horizons, hyperparameter selection, verification metrics, and all results tables/figures—cannot be checked. The abstract makes strong empirical claims ('any form of statistical post-processing significantly improves', 'non-parametric methods outperform parametric models', 'machine learning-based approaches surpass their traditional counterparts'), but the supporting evidence is entirely absent from the submitted version. This is a load-bearing problem: the central claims are empirical rankings, and without an auditable evaluation section they are unsupported.
[Abstract, claim of statistical significance] The abstract uses the word 'significantly' to characterize the improvement of post-processed forecasts over the raw ensemble. No readable evidence of significance testing appears anywhere in the supplied text: no paired per-plant tests, no confidence intervals, no multiple-comparison corrections, and no effect sizes. The reader cannot determine whether this claim refers to pooled averages, per-site paired tests, or something else. The claim of significance is load-bearing for the abstract's headline conclusion, and it cannot be verified.
[Abstract / results (location unreadable)] The claimed universal ranking—non-parametric over parametric, ML over traditional—depends critically on all methods being evaluated under identical conditions: the same training window, the same verification period, the same error metric, and hyperparameters selected only on training/validation data. None of these conditions can be checked in the supplied manuscript. In particular, if the nonlinear quantile regression models were selected or tuned using the verification data, the reported ranking would not follow. The manuscript must disclose the exact split and model-selection protocol; currently this information is invisible.

minor comments (3)

[Title/Abstract] The title mentions 'distributional and quantile regression methods', but the method definitions and equations are in the corrupted part of the text. A clean version must include legible model specifications and naming conventions.
[Figures and tables] All figures and tables appear as garbled blocks in the supplied version. Clean captions and legible numerical values are required for review.
[Conclusions/references] The concluding sections and reference list are partially visible but truncated and unreadable. Please ensure the final manuscript contains complete references and a properly formatted conclusions section.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity identified in available text; full text is corrupted so no reduction-by-construction can be quoted.

full rationale

The only readable portion of the manuscript is the abstract, which reports an empirical benchmark of seven post-processing methods against a raw PV power ensemble. Nothing in the abstract defines a method or parameter in terms of the target ranking, nor does any abstracted equation reduce a prediction to an input fit. The claims that post-processing improves on the raw ensemble and that nonlinear quantile regression performs best are ordinary comparative findings that require an evaluation protocol, but inability to audit that protocol is an evidentiary limitation, not circularity. The full-text body supplied is largely unreadable mojibake, so no specific equation, fitted parameter, or self-citation chain can be quoted to exhibit a circular step. Under the hard rule that circularity may only be claimed when a specific reduction can be quoted from the paper, no circularity finding is supported. A score of 0 is therefore the honest outcome.

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

No invented entities appear; post-processing algorithms inherit standard model assumptions from the cited literature. The only visible 'extra' is the choice of seven plants and one forecast period, which is a data-domain assumption rather than a new theoretical entity.

assumptions (3)

domain assumption The seven utility-scale PV plants in Hungary and the associated forecast period are representative enough for the paper's general comparative conclusions (non-parametric over parametric, ML over classical).
The abstract generalizes from a single-country, seven-plant dataset; if these plants are not representative, the headline ranking may not transfer.
domain assumption The raw ensemble PV power forecasts and PV production records are accurate enough to serve as ground truth and input, i.e., systematic errors are in the weather model chain and are correctable by post-processing.
The entire study presumes that observed power is trustworthy and that the ensemble input carries the predictive signal that post-processing can calibrate.
domain assumption The statistical properties of the forecast errors are constant over the evaluation period (stationarity), enabling any post-processing model fitted on past data to be valid on future data.
Standard for post-processing benchmarks; the abstract does not discuss non-stationarities such as seasonal or ageing effects of the plants.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Post-processing of ensemble photovoltaic power forecasts with distributional and quantile regression methods." pith.science (2026). https://pith.science/paper/ERINJQKC

@misc{pith2026250815508,
  author       = {Pith},
  title        = {Pith review of: Post-processing of ensemble photovoltaic power forecasts with distributional and quantile regression methods},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ERINJQKC}},
  note         = {Machine review of arXiv:2508.15508}
}

read the original abstract

Accurate and reliable forecasting of photovoltaic (PV) power generation is crucial for grid operations, electricity markets, and energy planning, as solar systems now contribute a significant share of the electricity supply in many countries. PV power forecasts are often generated by converting forecasts of relevant weather variables to power predictions via a model chain. The use of ensemble simulations from numerical weather prediction models results in probabilistic PV forecasts in the form of a forecast ensemble. However, weather forecasts often exhibit systematic errors that propagate through the model chain, leading to biased and/or uncalibrated PV power predictions. These deficiencies can be mitigated by statistical post-processing. Using PV production data and corresponding short-term PV power ensemble forecasts at seven utility-scale PV plants in Hungary, we systematically evaluate and compare seven state-of-the-art methods for post-processing PV power forecasts. These include both parametric and non-parametric techniques, as well as statistical and machine learning-based approaches. Our results show that compared to the raw PV power ensemble, any form of statistical post-processing significantly improves the predictive performance. Non-parametric methods outperform parametric models, with advanced nonlinear quantile regression models showing the best results. Furthermore, machine learning-based approaches surpass their traditional statistical counterparts.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

82 extracted references · 77 canonical work pages

[1]

and Koyama, M

Akiba, T., Sano, S., Yanase, T., Ohta, T. and Koyama, M. (2019) Optuna: A Next-generation Hyperparameter Optimization Framework . In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining , KDD '19, 2623--2631. New York, NY, USA: Association for Computing Machinery

2019
[2]

and Schmeits, M

Bakker, K., Whan, K., Knap, W. and Schmeits, M. (2019) Comparison of statistical post-processing methods for probabilistic NWP forecasts of solar radiation. Solar Energy, 191, 138–150

2019
[3]

and Baran, S

Baran, \'A . and Baran, S. (2024) A two-step machine learning approach to statistical post-processing of weather forecasts for power generation. Quarterly Journal of the Royal Meteorological Society, 150, 1029--1047

work page 2024
[4]

and Lerch, S

Baran, S. and Lerch, S. (2015) Log-normal distribution based EMOS models for probabilistic wind speed forecasting. Quarterly Journal of the Royal Meteorological Society, 141, 2289--2299

work page 2015
[5]

C., Cuevas, O., D\' az, M., Szab\'o, M., Nicolis, O

Baran, S., Mar\' n, J. C., Cuevas, O., D\' az, M., Szab\'o, M., Nicolis, O. and Lakatos, M. (2025) Machine-learning-based probabilistic forecasting of solar irradiance in Chile . Advances in Statistical Climatology, Meteorology and Oceanography, 11, 89--105

work page 2025
[6]

and Nemoda, D

Baran, S. and Nemoda, D. (2016) Censored and shifted gamma distribution based EMOS model for probabilistic quantitative precipitation forecasting. Environmetrics, 27, 280--292

work page 2016
[7]

and Hochberg, Y

Benjamini, Y. and Hochberg, Y. (1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Methodological), 57, 289--300

work page 1995
[8]

and Tian, Q

Bi, K., Xie, L., Zhang, H., Chen, X., Gu, X. and Tian, Q. (2023) Accurate medium-range global weather forecasting with 3D neural networks . Nature, 619, 533--–538

work page 2023

Show all 82 references

[9]

Bremnes, J. B. (2020) Ensemble postprocessing using quantile function regression based on neural networks and Bernstein polynomials. Monthly Weather Review, 148, 403--414

2020
[10]

B., Nipen, T

Bremnes, J. B., Nipen, T. N. and Seierstad, I. A. (2024) Evaluation of forecasts by a global data-driven weather model with and without probabilistic post-processing at Norwegian stations. Nonlinear Processes in Geophysics, 31, 247--257

2024
[11]

(2009) Reliability, sufficiency, and the decomposition of proper scores

Br \"o cker, J. (2009) Reliability, sufficiency, and the decomposition of proper scores. Quarterly Journal of the Royal Meteorological Society, 135, 1512--1519

2009
[12]

(2012) Evaluating raw ensembles with the continuous ranked probability score

Bröcker, J. (2012) Evaluating raw ensembles with the continuous ranked probability score. Quarterly Journal of the Royal Meteorological Society, 138, 1611--1617

2012
[13]

(2018) Ensemble forecasting and the need for calibration

Buizza, R. (2018) Ensemble forecasting and the need for calibration. In Statistical Postprocessing of Ensemble Forecasts (eds. S. Vannitsem, D. S. Wilks and J. W. Messner), 15--48. Amsterdam: Elsevier

2018
[14]

and Lerch, S

B \"u lte, C., Horat, N., Quinting, J. and Lerch, S. (2025) Uncertainty quantification for data-driven weather models. Artificial Intelligence for the Earth Systems, in press

2025
[15]

Cannon, A. J. (2011) Quantile regression neural networks: Implementation in R and application to precipitation downscaling. Computers and Geosciences, 37, 1277--1284

2011
[16]

and Beckman, W

De Soto, W., Klein, S. and Beckman, W. (2006) Improvement and validation of a model for photovoltaic array performance. Solar Energy, 80, 78--88

2006
[17]

P., Zhou, Y., Deng, X

Delle Monache, L., Hacker, J. P., Zhou, Y., Deng, X. and Stull, R. B. (2006) Probabilistic aspects of meteorological and ozone regional ensemble forecasts. Journal of Geophysical Research: Atmospheres, 111, D24307

2006
[18]

and Vannitsem, S

Demaeyer, J., Bhend, J., Lerch, S., Primo, C., Van Schaeybroeck, B., Atencia, A., Ben Bouall \`e gue, Z., Chen, J., Dabernig, M., Evans, G., Faganeli Pucer, J., Hooper, B., Horat, N., Jobst, D., Mer s e, J., Mlakar, P., M \"o ller, A., Mestre, O., Taillardat, M. and Vannitsem,...

2023
[19]

Diebold, F. X. and Mariano, R. S. (1995) Comparing predictive accuracy. Journal of Business and Economic Statistics, 13, 253--263

1995
[20]

and Harrison, S

Driesse, A., Jain, P. and Harrison, S. (2008) Beyond the curves: Modeling the electrical efficiency of photovoltaic inverters. In 2008 33rd IEEE Photovolatic Specialists Conference , 1--6. IEEE

2008
[21]

(2011) Making and evaluating point forecasts

Gneiting, T. (2011) Making and evaluating point forecasts. Journal of the American Statistical Association, 106, 746--762

2011
[22]

and Schulz, B

Gneiting, T., Lerch, S. and Schulz, B. (2023 a ) Probabilistic solar forecasting: Benchmarks, post-processing, verification. Solar Energy, 252, 72--80

2023
[23]

and Raftery, A

Gneiting, T. and Raftery, A. E. (2007) Strictly proper scoring rules, prediction and estimation. Journal of the American Statistical Association, 102, 359--378

2007
[24]

E., Westveld, A

Gneiting, T., Raftery, A. E., Westveld, A. H. and Goldman, T. (2005) Calibrated probabilistic forecasting using ensemble model output statistics and minimum CRPS estimation. Monthly Weather Review, 133, 1098--1118

2005
[25]

and Ranjan, R

Gneiting, T. and Ranjan, R. (2011) Comparing density forecasts using thresholdand quantile-weighted scoring rules. Journal of Business and Economic Statistics, 29, 411--422

2011
[26]

I., Lerch, S., Phipps, K

Gneiting, T., Wolffram, D., Resin, J., Kraus, K., Bracher, J., Dimitriadis, T., Hagenmeyer, V., Jordan, A. I., Lerch, S., Phipps, K. et al. (2023 b ) Model diagnostics and forecast evaluation for quantiles. Annual Review of Statistics and Its Application, 10, 597--621

2023
[27]

E., Chapman, W., Adams, S

Haupt, S. E., Chapman, W., Adams, S. V., Kirkwood, C., Hosking, J. S., Robinson, N. H., Lerch, S. and Subramanian, A. C. (2021) Towards implementing artificial intelligence post-processing in weather and climate: Proposed actions from the Oxford 2019 workshop. Philosophical Tr...

2021
[28]

Henzi, A., Ziegel, J. F. and Gneiting, T. (2021) Isotonic distributional regression. Journal of the Royal Statistical Society Series B: Statistical Methodology, 83, 963--993

2021
[29]

(2000) Decomposition of the continuous ranked probability score for ensemble prediction systems

Hersbach, H. (2000) Decomposition of the continuous ranked probability score for ensemble prediction systems. Weather and Forecasting, 15, 559--570

2000
[30]

and Zareipour, H

Hong, T., Pinson, P., Wang, Y., Weron, R., Yang, D. and Zareipour, H. (2020) Energy forecasting: A review and outlook. IEEE Open Access Journal of Power and Energy, 7, 376--388

2020
[31]

and Lerch, S

Horat, N., Klerings, S. and Lerch, S. (2025) Improving model chain approaches for probabilistic solar energy forecasting through post-processing and machine learning. Advances in Atmospheric Sciences, 42, 297--312

2025
[32]

and Lerch, S

Jordan, A., Kr \"u ger, F. and Lerch, S. (2019) Evaluating probabilistic forecasts with scoringRules . Journal of Statistical Software, 90, 1--37

2019
[33]

(2005) Quantile Regression

Koenker, R. (2005) Quantile Regression. Cambridge, UK: Cambridge University Press

2005
[34]

(2020) Why the “best” point forecast depends on the error or accuracy measure

Kolassa, S. (2020) Why the “best” point forecast depends on the error or accuracy measure. International Journal of Forecasting, 36, 208--211

2020
[35]

and Gneiting, T

Kr\"uger, F., Lerch, S., Thorarinsdottir, T. and Gneiting, T. (2021) Predictive inference based on Markov Chain Monte Carlo output. International Statistical Review, 89, 274--301

2021
[36]

and Battaglia, P

Lam, R., Sanchez-Gonzalez, A., Willson, M., Wirnsberger, P., Fortunato, M., Alet, F., Ravuri, S., Ewalds, T., Eaton-Rosen, Z., Hu, W., Merose, A., Hoyer, S., Holland, G., Vinyals, O., Stott, J., Pritzel, A., Mohamed, S. and Battaglia, P. (2023) Learning skillful medium-range g...

2023
[37]

Lang, S., Alexe, M., Chantry, M., Dramsch, J., Pinault, F., Raoult, B., Clare, M. C. A., Lessig, C., Maier-Gerber, M., Magnusson, L., Bouallègue, Z. B., Nemesio, A. P., Dueben, P. D., Brown, A., Pappenberger, F. and Rabier, F. (2024 a ) AIFS -- ECMWF 's data-driven forecasting...

2024 arXiv
[38]

Lang, S., Alexe, M., Clare, M. C. A., Roberts, C., Adewoyin, R., Bouallègue, Z. B., Chantry, M., Dramsch, J., Dueben, P. D., Hahner, S., Maciel, P., Prieto-Nemesio, A., O'Brien, C., Pinault, F., Polster, J., Raoult, B., Tietsche, S. and Leutbecher, M. (2024 b ) AIFS-CRPS : Ens...

2024 arXiv
[39]

and Pinson, P

Lauret, P., David, M. and Pinson, P. (2019) Verification of solar irradiance probabilistic forecasts. Solar Energy, 194, 254--271

2019
[40]

and Lauret, P

Le Gal La Salle , J., Badosa, J., David, M., Pinson, P. and Lauret, P. (2020) Added-value of ensemble prediction system on the quality of solar irradiance probabilistic forecasts. Renewable Energy, 162, 1321--1339

2020
[41]

Lefèvre, M., Oumbe, A., Blanc, P., Espinar, B., Gschwind, B., Qu, Z., Wald, L., Schroedter-Homscheidt, M., Hoyer-Klick, C., Arola, A., Benedetti, A., Kaiser, J. W. and Morcrette, J. J. (2013) McClear : A new model estimating downwelling solar radiation at ground level in clear...

2013
[42]

and Ruiz, J

Martin, N. and Ruiz, J. (2001) Calculation of the PV modules angular losses under field conditions by means of an analytical model. Solar Energy Materials and Solar Cells, 70, 25--38

2001
[43]

and Poggi, P

Mattei, M., Notton, G., Cristofari, C., Muselli, M. and Poggi, P. (2006) Calculation of the polycrystalline PV module temperature using a simple method of energy balance. Renewable Energy, 31, 553--567

2006
[44]

Mayer, M. J. (2022) Impact of the tilt angle, inverter sizing factor and row spacing on the photovoltaic power forecast accuracy. Applied Energy, 323, 119598

2022
[45]

Mayer, M. J. and Gr \'o f, G. (2020) Techno-economic optimization of grid-connected, ground-mounted photovoltaic power plants by genetic algorithm based on a comprehensive mathematical model. Solar Energy, 202, 210--226

2020
[46]

Applied Energy, 283, 116239

--- (2021) Extensive comparison of physical models for photovoltaic power forecasting. Applied Energy, 283, 116239

2021
[47]

Mayer, M. J. and Yang, D. (2022) Probabilistic photovoltaic power forecasting using a calibrated ensemble of model chains. Renewable and Sustainable Energy Reviews, 168, 112821

2022
[48]

International Journal of Forecasting, 39, 981--991

--- (2023) Calibration of deterministic NWP forecasts and its impact on verification. International Journal of Forecasting, 39, 981--991

2023
[49]

Applied Energy, 371, 123681

--- (2024) Optimal place to apply post-processing in the deterministic photovoltaic power forecasting workflow. Applied Energy, 371, 123681

2024
[50]

W., Mayr, G

Messner, J. W., Mayr, G. J. and Zeileis, A. (2016) Heteroscedastic censored and truncated regression with crch . The R Journal , 8, 173--181

2016
[51]

Monthly Weather Review, 145, 137--147

--- (2017) Nonhomogeneous boosting for predictor selection in ensemble postprocessing. Monthly Weather Review, 145, 137--147

2017
[52]

and Faganeli Pucer, J

Mlakar, P., Merše, J. and Faganeli Pucer, J. (2024) Ensemble weather forecast post‐processing with a flexible probabilistic neural network approach. Quarterly Journal of the Royal Meteorological Society, 150, 4156–4177

2024
[53]

and Stewart, R

Perez, R., Ineichen, P., Seals, R., Michalsky, J. and Stewart, R. (1990) Modeling daylight availability and irradiance components from direct and global irradiance. Solar Energy, 44, 271--289

1990
[54]

and Ludwig, N

Phipps, K., Lerch, S., Andersson, M., Mikut, R., Hagenmeyer, V. and Ludwig, N. (2022) Evaluating ensemble post-processing for wind power forecasts. Wind Energy, 25, 1379--1405

2022
[55]

Politis, D. N. and Romano, J. P. (1994) The stationary bootstrap. Journal of the American Statistical Association, 89, 1303--1313

1994
[56]

R., El-Kadi, A., Masters, D., Ewalds, T., Stott, J., Mohamed, S., Battaglia, P., Lam, R

Price, I., Sanchez-Gonzalez, A., Alet, F., Andersson, T. R., El-Kadi, A., Masters, D., Ewalds, T., Stott, J., Mohamed, S., Battaglia, P., Lam, R. and Willson, M. (2025) Probabilistic weather forecasting with machine learning. Nature, 637, 84--90

2025
[57]

and Lerch, S

Rasp, S. and Lerch, S. (2018) Neural networks for postprocessing ensemble weather forecasts. Monthly Weather Review, 146, 3885--3900

2018
[58]

and Andreas, A

Reda, I. and Andreas, A. (2004) Solar position algorithm for solar radiation applications. Solar Energy, 76, 577--589

2004
[59]

J., Mendiburu Zevallos, A

Roberts, J. J., Mendiburu Zevallos, A. A. and Cassula, A. M. (2017) Assessment of photovoltaic performance models for system simulation. Renewable and Sustainable Energy Reviews, 72, 1104–1123

2017
[60]

(2014) Probabilistic quantitative precipitation forecasting using ensemble model output statistics

Scheuerer, M. (2014) Probabilistic quantitative precipitation forecasting using ensemble model output statistics. Quarterly Journal of the Royal Meteorological Society, 140, 1086--1096

2014
[61]

and Baran, S

Schulz, B., El Ayari, M., Lerch, S. and Baran, S. (2021) Post-processing numerical weather prediction ensembles for probabilistic solar irradiance forecasting. Solar Energy, 220, 1016--1031

2021
[62]

and Lerch, S

Schulz, B., Köhler, L. and Lerch, S. (2022) Aggregating distribution forecasts from deep ensembles. Preprint, available at https://arxiv.org/abs/2204.02291

2022 arXiv
[63]

and Lerch, S

Schulz, B. and Lerch, S. (2022) Machine learning methods for postprocessing ensemble forecasts of wind gusts: A systematic comparison. Monthly Weather Review, 150, 235--257

2022
[64]

M., Bright, J

Song, M., Yang, D., Lerch, S., Xia, X., Yagli, G. M., Bright, J. M., Shen, Y., Liu, B., Liu, X. and Mayer, M. J. (2024) Non-crossing quantile regression neural network as a calibration tool for ensemble weather forecasts. Advances in Atmospheric Sciences, 41, 1417--1437

2024
[65]

and Naveau, P

Taillardat, M., Mestre, O., Zamo, M. and Naveau, P. (2016) Calibrated ensemble forecasts using quantile regression forests and ensemble model output statistics. Monthly Weather Review, 144, 2375--2393

2016
[66]

Taylor, J. W. (2000) A quantile regression neural network approach to estimating the conditional density of multiperiod returns. Journal of Forecasting, 19, 299--311

2000
[67]

and Georghiou, G

Theocharides, S., Makrides, G., Livera, A., Theristis, M., Kaimakis, P. and Georghiou, G. E. (2020) Day-ahead photovoltaic power production forecasting methodology based on machine learning and statistical post-processing. Applied Energy, 268, 115023

2020
[68]

Thorarinsdottir, T. L. and Gneiting, T. (2010) 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), 173A, 371--388

2010
[69]

and Vannitsem, S

Van Schaeybroeck, B. and Vannitsem, S. (2015) Ensemble post-processing using member-by-member approaches: Theoretical aspects. Quarterly Journal of the Royal Meteorological Society, 141, 807--818

2015
[70]

B., Demaeyer, J., Evans, G

Vannitsem, S., Bremnes, J. B., Demaeyer, J., Evans, G. R., Flowerdew, J., Hemri, S., Lerch, S., Roberts, N., Theis, S., Atencia, A., Ben Bouallègue , Z., Bhend, J., Dabernig, M., Cruz, L. D., Hieta, L., Mestre, O., Moret, L., Plenković, I. O., Schmeits, M., Taillardat, M., den...

2021
[71]

and Kleissl, J

Wang, W., Yang, D., Hong, T. and Kleissl, J. (2022) An archived dataset from the ECMWF ensemble prediction system for probabilistic solar power forecasting. Solar Energy, 248, 64--75

2022
[72]

The Stippling Shows Statistically Significant Grid Points

Wilks, D. S. (2016) "The Stippling Shows Statistically Significant Grid Points" : How research results are routinely overstated and overinterpreted, and what to do about it. Bulletin of the American Meteorological Society, 97, 2263--2273

2016
[73]

Amsterdam: Elsevier, 4th edn

--- (2019) Statistical Methods in the Atmospheric Sciences. Amsterdam: Elsevier, 4th edn

2019
[74]

(2016) Solar radiation on inclined surfaces: Corrections and benchmarks

Yang, D. (2016) Solar radiation on inclined surfaces: Corrections and benchmarks. Solar Energy, 136, 288--302

2016
[75]

Journal of Renewable and Sustainable Energy, 13, 056101

--- (2021) Temporal-resolution cascade model for separation of 1-min beam and diffuse irradiance. Journal of Renewable and Sustainable Energy, 13, 056101

2021
[76]

Renewable and Sustainable Energy Reviews, 159, 112195

--- (2022) Estimating 1-min beam and diffuse irradiance from the global irradiance: A review and an extensive worldwide comparison of latest separation models at 126 stations. Renewable and Sustainable Energy Reviews, 159, 112195

2022
[77]

J., Gueymard, C

Yang, D., Gu, Y., Mayer, M. J., Gueymard, C. A., Wang, W., Kleissl, J., Li, M., Chu, Y. and Bright, J. M. (2024 a ) Regime-dependent 1-min irradiance separation model with climatology clustering. Renewable and Sustainable Energy Reviews, 189, 113992

2024
[78]

and van der Meer, D

Yang, D. and van der Meer, D. (2021) Post-processing in solar forecasting: Ten overarching thinking tools. Renewable and Sustainable Energy Reviews, 140, 110735

2021
[79]

A., Hong, T., Kleissl, J., Huang, J., Perez, M

Yang, D., Wang, W., Gueymard, C. A., Hong, T., Kleissl, J., Huang, J., Perez, M. J., Perez, R., Bright, J. M., Xia, X., van der Meer, D. and Peters, I. M. (2022 a ) A review of solar forecasting, its dependence on atmospheric sciences and implications for grid integration: Tow...

2022
[80]

and Xia, X

Yang, D., Wang, W. and Xia, X. (2022 b ) A concise overview on solar resource assessment and forecasting. Advances in Atmospheric Sciences, 39, 1239--1251

2022
[81]

and Mayer, M

Yang, D., Xia, X. and Mayer, M. J. (2024 b ) A tutorial review of the solar power curve: Regressions, model chains, and their hybridization and probabilistic extensions. Advances in Atmospheric Sciences, 41, 1023--1067

2024
[82]

and Li, H

Zhong, X., Chen, L., Fan, X., Qian, W., Liu, J. and Li, H. (2024) FuXi-2.0: Advancing machine learning weather forecasting model for practical applications . Preprint, available at https://arxiv.org/abs/2409.07188

2024 arXiv

Pith tools

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

[1] [1]

and Koyama, M

Akiba, T., Sano, S., Yanase, T., Ohta, T. and Koyama, M. (2019) Optuna: A Next-generation Hyperparameter Optimization Framework . In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining , KDD '19, 2623--2631. New York, NY, USA: Association for Computing Machinery

2019

[2] [2]

and Schmeits, M

Bakker, K., Whan, K., Knap, W. and Schmeits, M. (2019) Comparison of statistical post-processing methods for probabilistic NWP forecasts of solar radiation. Solar Energy, 191, 138–150

2019

[3] [3]

and Baran, S

Baran, \'A . and Baran, S. (2024) A two-step machine learning approach to statistical post-processing of weather forecasts for power generation. Quarterly Journal of the Royal Meteorological Society, 150, 1029--1047

work page 2024

[4] [4]

and Lerch, S

Baran, S. and Lerch, S. (2015) Log-normal distribution based EMOS models for probabilistic wind speed forecasting. Quarterly Journal of the Royal Meteorological Society, 141, 2289--2299

work page 2015

[5] [5]

C., Cuevas, O., D\' az, M., Szab\'o, M., Nicolis, O

Baran, S., Mar\' n, J. C., Cuevas, O., D\' az, M., Szab\'o, M., Nicolis, O. and Lakatos, M. (2025) Machine-learning-based probabilistic forecasting of solar irradiance in Chile . Advances in Statistical Climatology, Meteorology and Oceanography, 11, 89--105

work page 2025

[6] [6]

and Nemoda, D

Baran, S. and Nemoda, D. (2016) Censored and shifted gamma distribution based EMOS model for probabilistic quantitative precipitation forecasting. Environmetrics, 27, 280--292

work page 2016

[7] [7]

and Hochberg, Y

Benjamini, Y. and Hochberg, Y. (1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Methodological), 57, 289--300

work page 1995

[8] [8]

and Tian, Q

Bi, K., Xie, L., Zhang, H., Chen, X., Gu, X. and Tian, Q. (2023) Accurate medium-range global weather forecasting with 3D neural networks . Nature, 619, 533--–538

work page 2023