254-25

Assessment of Random Forest Algorithm Efficacy for Photovoltaic Production Prediction During Saharan Dust Intrusion Episodes: A Case Study Including Wet Deposition

S. Alonso-Pérez(1), J. López-Solano(1), C. Montes(2), L. Ocaña(2), E. Llarena(2), B. González-Díaz(1)

1.- Departamento de Ingeniería Industrial. Escuela Superior de Ingeniería y Tecnología, Universidad de La Laguna. San Cristóbal de La Laguna, Tenerife (Spain)

2.- Instituto Tecnológico y de Energías Renovables, S.A. Pol. Industrial Granadilla, Tenerife (Spain)

2025-07-25

This study assesses the efficacy of the Random Forest (RF) algorithm in predicting photovoltaic (PV) energy production during Saharan dust intrusion episodes, including days of wet deposition. The research focuses on a PV plant in Tenerife, Canary Islands, during the period July 24 to September 20, 2015, encompassing a significant dust event. The RF model was applied to both daily and hourly energy production forecasts, incorporating various meteorological and atmospheric parameters as input features. For daily predictions, the model achieved high accuracy (97.58%) using only four key features: solar insolation hours, incident shortwave radiation, wind velocity, and maximum gust velocity. Hourly predictions required ten features to attain comparable accuracy (95.86%). The study demonstrates the RF algorithm's robustness in forecasting PV output under complex atmospheric conditions, even with limited non-in-situ data.

Key words: African dust, Random Forest, Photovoltaic energy, energy production, prediction

[1] J. Antonanzas, N. Osorio, R. Escobar, R. Urraca, F. J. Martinez-de-Pison and F. Antonanzas-Torres, "Review of photovoltaic power forecasting", Solar Energy (2016), Vol. 136, pp. 78-111, doi: 10.1016/j.solener.2016.06.069.

[2] D. J. V. Kumar, K. A. Ritter III, J. R. Raush, F. Ferdowsi, R. Gottumukkala and T. L. Chambers, "Optimizing Photovoltaic Soiling Loss Predictions in Louisiana: A Comparative Study of Measured and Modeled Data Using a Novel Approach", Progress in Photovoltaics (2025), Vol. 33 (4), pp. 560-579, doi: 10.1002/pip.3891.

[3] S. Fan, M. He and Z. Zhang, "Daily prediction method of dust accumulation on photovoltaic (PV) panels using echo state network with delay output", Applied Soft Computing (2023), Vol. 144, p. 110528, doi: 10.1016/j.asoc.2023.110528.

[4] J. Hu et al., "A Comprehensive Review of Artificial Intelligence Applications in the Photovoltaic Systems", CAAI Artificial Intelligence Research (2024), Vol. 3, doi: 10.26599/AIR.2024.9150031.

[5] M. Tucci, A. Piazzi and D. Thomopulos, "Machine Learning Models for Regional Photovoltaic Power Generation Forecasting with Limited Plant-Specific Data", Energies (2024), Vol. 17 (10), p. 2346, doi: 10.3390/en17102346.

[6] D. Suárez-Molina et al., "Dust events characterization from visibility, trends and Dust Adversity Index in the Canary Islands for the period 1980–2022", Heliyon (2024), Vol. 10 (10), pp. 1-19, doi: 10.1016/J.HELIYON.2024.E31262.

[7] C. Tsamalis, A. Chédin, J. Pelon and V. Capelle, "The seasonal vertical distribution of the Saharan Air Layer and its modulation by the wind", Atmospheric Chemistry and Physics (2013), Vol. 13 (22), pp. 11235-11257, doi: 10.5194/acp-13-11235-2013.

[8] Á. Barreto et al., "Long-term characterisation of the vertical structure of the Saharan Air Layer over the Canary Islands using lidar and radiosonde profiles: implications for radiative and cloud processes over the subtropical Atlantic Ocean", Atmospheric Chemistry and Physics (2022), Vol. 22 (2), pp. 739-763, doi: 10.5194/acp-22-739-2022.

[9] L. Breiman, J. Friedman, R. A. Olshen and C. J. Stone, Classification and Regression Trees. Chapman and Hall/CRC , New York (2017), doi: 10.1201/9781315139470.

[10] J. Holloway-Brown, K. J. Helmstedt and K. L. Mengersen, "Stochastic spatial random forest (SS-RF) for interpolating probabilities of missing land cover data", Journal of Big Data (2020), Vol. 7 (1), p. 55, doi: 10.1186/s40537-020-00331-8.

[11] E. Scornet, "Trees, forests, and impurity-based variable importance in regression", Annales de l'Institut Henri Poincaré, Probabilités et Statistiques (2023), Vol. 59 (1), doi: 10.1214/21-AIHP1240.

[12] "Spanish Meteorological Agency OpenData". [On-line]. Available at: https://opendata.aemet.es/centrodedescargas/inicio

[13] University of Wyoming, "METeorological Aerodrome Reports (METAR)". [On-line]. Available at: https://weather.uwyo.edu/surface/meteorogram/index.shtml

[14] National Aeronautics and Space Administration, "Giovanni database". [On-line]. Available at: https://giovanni.gsfc.nasa.gov/giovanni/

[15] Q. Wen et al., "Time Series Data Augmentation for Deep Learning: A Survey", Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, Montreal, Canada: International Joint Conferences on Artificial Intelligence Organization (2021), pp. 4653-4660. doi: 10.24963/ijcai.2021/631.

[16] L. Bottou, "Large-Scale Machine Learning with Stochastic Gradient Descent", Proceedings of COMPSTAT’2010, Y. Lechevallier and G. Saporta, Eds., Heidelberg (2019), pp. 177-186, doi: 10.1007/978-3-7908-2604-3_16.

[17] D. Goossens and Z. Y. Offer, "Comparisons of day-time and night-time dust accumulation in a desert region", Journal of Arid Environments (1995), Vol. 31 (3), pp. 253-281, doi: 10.1016/S0140-1963(05)80032-1.

[18] G. A. Cruz and P. Vorobiev, "Dust adhesion physics on photovoltaic surfaces and its application on assisted natural cleaning of solar trackers", Journal of Physics: Conferences Series (2022), Vol. 2238 (1), p. 012013, doi: 10.1088/1742-6596/2238/1/012013.

[19] P. Borah, L. Micheli and N. Sarmah, "Analysis of Soiling Loss in Photovoltaic Modules: A Review of the Impact of Atmospheric Parameters, Soil Properties, and Mitigation Approaches", Sustainability (2023), Vol. 15 (24), p. 16669, doi: 10.3390/su152416669

[20] B. R. Paudyal and S. R. Shakya, "Dust accumulation effects on efficiency of solar PV modules for off grid purpose: A case study of Kathmandu", Solar Energy (2016), Vol. 135, pp. 103-110, doi: 10.1016/j.solener.2016.05.046.

[21] R. Li, S. Ti, L. Li, J. Song and T. Chen, "Diurnal variations of atmospheric electric fields on fair weather days and its correlations with aerosols, wind speed, irradiance, and relative humidity", Scientific Reports (2025), Vol. 15 (1), p. 7030, doi: 10.1038/s41598-025-89788-2.

[22] T. Sarver, A. Al-Qaraghuli, and L. L. Kazmerski, "A comprehensive review of the impact of dust on the use of solar energy: History, investigations, results, literature, and mitigation approaches", Renewable and Sustainable Energy Reviews (2013), Vol. 22, pp. 698-733, doi: 10.1016/j.rser.2012.12.065.

[23] P. Hooshyar, A. Moosavi and A. N. Borujerdi, "Enhanced dust reduction method for solar panels application", Scientific Reports (2024), Vol. 14 (1), p. 30351, doi: 10.1038/s41598-024-81183-7.

[24] D. Abdulai, S. Gyamfi, F. A. Diawuo and P. Acheampong, "Data analytics for prediction of solar PV power generation and system performance: A real case of Bui Solar Generating Station, Ghana", Scientific African (2023), Vol. 21, p. e01894, doi: 10.1016/j.sciaf.2023.e01894.

[25] K. Mahmud, S. Azam, A. Karim, S. Zobaed, B. Shanmugam and D. Mathur, "Machine Learning Based PV Power Generation Forecasting in Alice Springs", IEEE Access (2021), Vol. 9, pp. 46117-46128, doi: 10.1109/ACCESS.2021.3066494.

[26] N. Quang, L. Duy, B. Van and Q. Dinh, "Applying Artificial Intelligence in Forecasting the Output of Industrial Solar Power Plant in Vietnam", EAI Endorsed Transactions on Energy Web (2018), Vol. 8 (36), p. 169166, doi: 10.4108/eai.29-3-2021.169166.

[27] M. Meng and C. Song, "Daily Photovoltaic Power Generation Forecasting Model Based on Random Forest Algorithm for North China in Winter", Sustainability (2020), Vol. 12 (6), p. 2247, doi: 10.3390/su12062247.