10.1007/s40095-022-00493-6

Global horizontal and direct normal solar irradiance modeling by the machine learning methods XGBoost and deep neural networks with CNN-LSTM layers: a case study using the GOES-16 satellite imagery

  • Paulo A. C. Rocha*1,
  • Victor Oliveira Santos1,
  1. Solar Energy and Natural Gas Laboratory, Mechanical Engineering Department, Technology Center, Federal University of Ceará, Fortaleza, Ceará, 60020-181, BR

Published in Issue 2022-04-19

How to Cite

Rocha, P. A. C., & Santos, V. O. (2022). Global horizontal and direct normal solar irradiance modeling by the machine learning methods XGBoost and deep neural networks with CNN-LSTM layers: a case study using the GOES-16 satellite imagery. International Journal of Energy and Environmental Engineering, 13(4 (December 2022). https://doi.org/10.1007/s40095-022-00493-6
Crossref
Scopus
Google Scholar
Europe PMC

Abstract

Abstract Restrictive legislations on the use of fossil fuels encourage the research and development of clean and renewable energies. Renewable energy is characterized by random behavior, which hampers its integration into the current energy base system. Thus, estimating solar irradiation is important for the adoption of renewable energies into the current energy matrix. In this paper, two machine learning estimation models for global horizontal (GHI) and direct normal solar irradiance (DNI) are proposed: the first uses XGBoost and the second employs a convolutional neural network (CNN) combined with a long short-term memory (LSTM) network, forming the hybrid CNN-LSTM model. The case studies apply both models to process images from the GOES-16 satellite, taken from the city of Petrolina, Pernambuco, Brazil. Their results are compared against the reference Copernicus Atmosphere Monitoring Service, Solcast and the Physical Solar Model (PSM) provided by the National Solar Radiation Database. For the GHI estimation, the PSM model achieved the lowest RMSE, 147.23 W/m 2 , while for DNI estimation, the CNN-LSTM model performed best, with an RMSE equal to 238.22 W/m 2 . In this case, the proposed models achieved lower RMSE for DNI estimation when compared against the benchmark models, improving by 2.89% and 1.70% for the CNN-LSTM and XGBoost models, respectively.

Keywords

  • Solar irradiance,
  • GOES-16 satellite,
  • Machine learning,
  • CNN-LSTM,
  • Keras R package,
  • Caret R package

References

  1. Agreement, P.: United nations. United Nations Treaty Collect, 1–27 (2015)
  2. Saidi and Omri (2020) The impact of renewable energy on carbon emissions and economic growth in 15 major renewable energy-consuming countries https://doi.org/10.1016/j.envres.2020.109567
  3. IRENA: Renewable Energy Statistics 2021. International Renewable Energy Agency (IRENA), Abu Dhabi (2021)
  4. Ministry of Mines and Energy: Boletim Mensal de Monitoramento do Sistema Eletrico Brasileiro, Novembro de 2021(2021). Available at:
  5. https://www.gov.br/mme/pt-br/assuntos/secretarias/energia-eletrica/publicacoes/boletim-de-monitoramento-do-sistema-eletrico/2021/boletim-de-monitoramento-do-sistema-eletrico-novembro-2021.pdf
  6. . Accessed: 2022–01–17
  7. Abatan, A.A., Tett, S.F., Dong, B., Cunningham, C., Rudorff, C.M., Klingaman, N.P., de Abreu, R.C.: Drivers and physical processes of drought events over the state of São Paulo, Brazil. Clim. Dyn., 1–15 (2022)
  8. Vourlitis et al. (2022) Tree growth responses to climate variation in upland and seasonally flooded forests and woodlands of the cerrado-pantanal transition of Brazil https://doi.org/10.1016/j.foreco.2021.119917
  9. Hunt et al. (2022) Energy crisis in Brazil: impact of hydropower reservoir level on the river flow https://doi.org/10.1016/j.energy.2021.121927
  10. Tupinambá-Simões et al. (2022) Assessment of drought effects on survival and growth dynamics in eucalypt commercial forestry using remote sensing photogrammetry. A showcase in Mato Grosso Brazil https://doi.org/10.1016/j.foreco.2021.119930
  11. Freitas et al. (2022) Drought assessment in São Francisco river basin, Brazil: characterization through SPI and associated anomalous climate patterns 13(1) https://doi.org/10.3390/atmos13010041
  12. Larson et al. (2020) Scope: Spectral cloud optical property estimation using real-time goes-r longwave imagery 12(2) https://doi.org/10.1063/1.5144350
  13. Rajagukguk et al. (2020) A review on deep learning models for forecasting time series data of solar irradiance and photovoltaic power 13(24) https://doi.org/10.3390/en13246623
  14. Kaur et al. (2016) Benefits of solar forecasting for energy imbalance markets (pp. 819-830) https://doi.org/10.1016/j.renene.2015.09.011
  15. Law et al. (2016) Calculating the financial value of a concentrated solar thermal plant operated using direct normal irradiance forecasts (pp. 267-281) https://doi.org/10.1016/j.solener.2015.12.031
  16. Yang et al. (2018) History and trends in solar irradiance and pv power forecasting: a preliminary assessment and review using text mining (pp. 60-101) https://doi.org/10.1016/j.solener.2017.11.023
  17. Voyant et al. (2017) Machine learning methods for solar radiation forecasting: a review (pp. 569-582) https://doi.org/10.1016/j.renene.2016.12.095
  18. Rocha et al. (2019) Estimation of daily, weekly and monthly global solar radiation using ANNs and a long data set: a case study of Fortaleza, in brazilian northeast region 10(3) (pp. 319-334) https://doi.org/10.1007/s40095-019-0313-0
  19. Marzo et al. (2017) Daily global solar radiation estimation in desert areas using daily extreme temperatures and extraterrestrial radiation (pp. 303-311) https://doi.org/10.1016/j.renene.2017.01.061
  20. Cornejo-Bueno et al. (2019) Machine learning regressors for solar radiation estimation from satellite data (pp. 768-775) https://doi.org/10.1016/j.solener.2019.03.079
  21. Chen et al. (2021) Estimating half-hourly solar radiation over the continental United States using goes-16 data with iterative random forest (pp. 916-929) https://doi.org/10.1016/j.renene.2021.06.129
  22. Jebli et al. (2021) Prediction of solar energy guided by pearson correlation using machine learning https://doi.org/10.1016/j.energy.2021.120109
  23. Fogno Fotso et al. (2021) A novel hybrid model based on weather variables relationships improving applied for wind speed forecasting (pp. 1-14)
  24. Rodríguez et al. (2022) Forecasting intra-hour solar photovoltaic energy by assembling wavelet based time-frequency analysis with deep learning neural networks https://doi.org/10.1016/j.ijepes.2021.107777
  25. Bastos et al. (2021) U-convolutional model for spatio-temporal wind speed forecasting 37(2) (pp. 949-970) https://doi.org/10.1016/j.ijforecast.2020.10.007
  26. Feng and Zhang (2020) Solarnet: a sky image-based deep convolutional neural network for intra-hour solar forecasting (pp. 71-78) https://doi.org/10.1016/j.solener.2020.03.083
  27. Huertas-Tato et al. (2021) Using a multi-view convolutional neural network to monitor solar irradiance https://doi.org/10.1007/s00521-021-05959-y
  28. Dong et al. (2020) A novel convolutional neural network framework based solar irradiance prediction method https://doi.org/10.1016/j.ijepes.2019.105411
  29. Feng et al. (2022) Convolutional neural networks for intra-hour solar forecasting based on sky image sequences https://doi.org/10.1016/j.apenergy.2021.118438
  30. Ghimire et al. (2019) Deep solar radiation forecasting with convolutional neural network and long short-term memory network algorithms https://doi.org/10.1016/j.apenergy.2019.113541
  31. Zang et al. (2020) Short-term global horizontal irradiance forecasting based on a hybrid cnn-lstm model with spatiotemporal correlations (pp. 26-41) https://doi.org/10.1016/j.renene.2020.05.150
  32. Kumari and Toshniwal (2021) Long short term memory–convolutional neural network based deep hybrid approach for solar irradiance forecasting https://doi.org/10.1016/j.apenergy.2021.117061
  33. Akhter et al. (2022) A hybrid deep learning method for an hour ahead power output forecasting of three different photovoltaic systems https://doi.org/10.1016/j.apenergy.2021.118185
  34. Sengupta et al. (2018) The national solar radiation data base (NSRDB) (pp. 51-60) https://doi.org/10.1016/j.rser.2018.03.003
  35. Habte et al. (2017) Technical report National Renewable Energy Lab(NREL)
  36. Yang (2018) Kriging for NSRDB PSM version 3 satellite-derived solar irradiance (pp. 876-883) https://doi.org/10.1016/j.solener.2018.06.055
  37. Yang (2018) A correct validation of the national solar radiation data base (NSRDB) (pp. 152-155) https://doi.org/10.1016/j.rser.2018.08.023
  38. Bright (2019) Solcast: validation of a satellite-derived solar irradiance dataset (pp. 435-449) https://doi.org/10.1016/j.solener.2019.07.086
  39. Yang and Bright (2020) Worldwide validation of 8 satellite-derived and reanalysis solar radiation products: a preliminary evaluation and overall metrics for hourly data over 27 years (pp. 3-19) https://doi.org/10.1016/j.solener.2020.04.016
  40. Hillger, D.W., Schmit, T.: Quantization noise for goes-r abi bands. In: Proceedings of the 13th Conference on Satellite Meteorology and Oceanography Norfolk, VA, pp. 20–23, American Meteorological Society (2004)
  41. Padula, F., Cao, C.: Cwg analysis: Abi max/min radiance characterization and validation. In: Proceedings of the 13th Conference on Satellite Meteorology and Oceanography Norfolk, VA, pp. 20–23, American Meteorological Society (2015)
  42. Kalluri et al. (2018) From photons to pixels: processing data from the advanced baseline imager 10(2) https://doi.org/10.3390/rs10020177
  43. Peel et al. (2007) Updated world map of the köppen-geiger climate classification 11(5) (pp. 1633-1644) https://doi.org/10.5194/hess-11-1633-2007
  44. Google Earth V7.3.4.8248: Geographic location of Petrolina SONDA station. 9°4’8”S, 40°19’11”W. Eye altitude 28 km. Maxar Technologies 2021, CNES/Airbus 2021 (2021). Available at:
  45. https://earth.google
  46. . com/web/. Accessed: 2021–08–31 (2021)
  47. Gao et al. (2020) Hourly forecasting of solar irradiance based on ceemdan and multi-strategy cnn-lstm neural networks (pp. 1665-1683) https://doi.org/10.1016/j.renene.2020.09.141
  48. Huang et al. (2021) Hybrid deep neural model for hourly solar irradiance forecasting (pp. 1041-1060) https://doi.org/10.1016/j.renene.2021.02.161
  49. Liu et al. (2014) Short-term wind speed forecasting using wavelet transform and support vector machines optimized by genetic algorithm (pp. 592-597) https://doi.org/10.1016/j.renene.2013.08.011
  50. Wang et al. (2016) Deep belief network based deterministic and probabilistic wind speed forecasting approach (pp. 80-93) https://doi.org/10.1016/j.apenergy.2016.08.108
  51. Kuhn and Johnson (2013) Springer https://doi.org/10.1007/978-1-4614-6849-3
  52. LeCun et al. (2015) Deep learning 521(7553) (pp. 436-444) https://doi.org/10.1038/nature14539
  53. Hochreiter and Schmidhuber (1997) Long short-term memory 9(8) (pp. 1735-1780) https://doi.org/10.1162/neco.1997.9.8.1735
  54. Wang et al. (2019) A hybrid ensemble method for pulsar candidate classification 364(8) (pp. 1-13) https://doi.org/10.1007/s10509-019-3602-4
  55. Chen, T., Guestrin, C.: Xgboost: A scalable tree boosting system. In: Proceedings of the 22nd Acm Sigkdd International Conference on Knowledge Discovery and Data Mining, pp. 785–794 (2016)
  56. Bondi, A.B.: Characteristics of scalability and their impact on performance. In: Proceedings of the 2nd International Workshop on Software and Performance, pp. 195–203 (2000)
  57. Si et al. (2021) Photovoltaic power forecast based on satellite images considering effects of solar position https://doi.org/10.1016/j.apenergy.2021.117514
  58. Fan et al. (2020) Predicting daily diffuse horizontal solar radiation in various climatic regions of China using support vector machine and tree-based soft computing models with local and extrinsic climatic data https://doi.org/10.1016/j.jclepro.2019.119264
  59. Koo et al. (2020) Estimation and mapping of solar irradiance for korea by using coms mi satellite images and an artificial neural network model 13(2) https://doi.org/10.3390/en13020301
  60. Kamath and Srinivasan (2020) Validation of global irradiance derived from insat-3d over India (pp. 45-54) https://doi.org/10.1016/j.solener.2020.03.084
  61. Nonnenmacher et al. (2014) Verification of the suny direct normal irradiance model with ground measurements (pp. 246-258) https://doi.org/10.1016/j.solener.2013.11.010
  62. Zhu et al. (2019) Clear-sky direct normal irradiance estimation based on adjustable inputs and error correction 11(5) https://doi.org/10.1063/1.5094808
  63. Kuhn et al. (2019) Shadow-camera based solar nowcasting system for shortest-term forecasts https://doi.org/10.1127/metz/2019/0954
  64. Paulescu et al. (2014) Nowcasting solar irradiance using the sunshine number (pp. 690-697) https://doi.org/10.1016/j.enconman.2013.12.048
  65. Nouri et al. (2021) A hybrid solar irradiance nowcasting approach: combining all sky imager systems and persistence irradiance models for increased accuracy https://doi.org/10.1002/solr.202100442