Vol. 38, issue 04, article # 7

Abstract:

In 2023, more than a third of dangerous meteorological events in the Siberian Federal District were associated with strong winds, which underscores the importance of improving the accuracy and timing of its forecasting. Modern numerical simulation and machine learning methods make it possible to improve forecasts, but the task of directly calculating wind gusts remains relevant due to the limited resolution of models. An original method is proposed for correcting the results of short-term forecast of wind gusts obtained on the basis of mesoscale models of numerical weather forecasting using advance measurements and artificial neural networks. The results show that the proposed correction method makes it possible to improve the forecast of wind gusts by various semi-empirical methods. The results can be applied in meteorology, energy, transportation, and other industries to minimize damage from dangerous weather events.

Keywords:

wind gust, mesoscale model TSUNM3, ultrasonic meteosite, numerical forecast correction, artificial neural network

References:

1. Doklad ob osobennostyakh klimata na territorii Rossiiskoi Federatsii za 2023 god. M.: Rosgidromet, 2024. 104 p.
2. Kurbatova M.M., Rubinshtein K.G. Gibridnyi metod prognoza poryvov vetra // Optika atmosf. i okeana. 2018. V. 31, N 7. P. 523–529. DOI: 10.15372/AOO20180704.
3. Starchenko A.V., Del' I.V., Odintsov S.L. Chislennoe prognozirovanie poryvov vetra v g. Tomske s pomoshch'yu modeli TSUNM3 // Optika atmosf. i okeana. 2024. V. 37, N 3. P. 225‒233. DOI: 10.15372/AOO20240306; Starchenko A.V., Del’ I.V., Odintsov S.L. Numerical prediction of wind gusts using the TSUNM3 model // Atmos. Ocean. Opt. 2024. V. 37, N 3. P. 429–437.
4. Schulz J.-P. Revision of the turbulent gust diagnostics in the COSMO model // COSMO Newletter. 2008. V. 8. P. 17–22.
5. Zjavka L. Wind speed forecast correction models using polynomial neural networks // Renewable Energy. 2015. V. 83. P. 998–1006. DOI: 10.1016/j.renene.2015.04.054.
6. Ding M., Zhou H., Xie H., Wu M., Nakanishi Y., Yokoyama R. A gated recurrent unit neural networks based wind speed error correction model for short-term wind power forecasting // Neurocomputing. 2019. V. 365. P. 54–61. DOI: 10.1016/j.neucom.2019.07.058.
7. Zhao X., Sun Q., Tang W., Yu S., Wang B. A comprehensive wind speed forecast correction strategy with an artificial intelligence algorithm // Front. Environ. Sci. 2022. V. 10. DOI: 10.3389/fenvs.2022.1034536.
8. Borisov D.V., Kuznetsova I.N. Postprotsessing chislennykh prognozov kontsentratsii prizemnogo ozona s ispol'zovaniem mashinnogo obucheniya // Gidrometeorologicheskie issledovaniya i prognozy. 2023. V. 390, N 4. P. 86–104. DOI: 10.37162/2618-9631-2023-4-86-104.
9. Starchenko A.V., Kizhner L.I., Danilkin E.A., Shel'mina E.A., Prokhanov S.A. Chislennoe modelirovanie pogody i kachestva atmosfernogo vozdukha v gorodakh. Tomsk: Izd-vo TGU, 2022. 140 p.
10. Khaikin S. Neironnye seti. Polnyi kurs. M.: Izdatel'skii dom «Vil'yams», 2006. 1104 p.
11. Gladkikh V.A., Nevzorova I.V., Odintsov S.L. Struktura poryvov vetra v prizemnom sloe atmosfery // Optika atmosf. i okeana. 2019. V. 32, N 4. P. 304–308. DOI: 10.15372/AOO20190408.
12. Tolstykh M.A., Shashkin V.V., Fadeev R.Yu., Shlyaeva A.V., Mizyak V.G., Rogutov V.S., Bogoslovskii N.N., Goiman G.S., Makhnorylova S.V., Yurova A.Yu. Sistema modelirovaniya atmosfery dlya besshovnogo prognoza. M.: Triada LTD, 2017. 166 p.
13. Benjamin S.G., Brown J.M., Brundage K.J., Dévényi D., Grell G.A., Kim D., Schwartz B.E., Smirnova T.G., Smith T.L., Weygandt S.S., Manikin G.S. NWS Technical Procedures. Bulletin. N 490. RUC20 – The 20 km Version of the Rapid Update Cycle. Boulder: NOAA/OAR Forecast Systems Laboratory, 2002.
14. Suarez M., Poffo D., Pierobon E., Martina A., Saffe J., Rodriguez A. Wind and gust forecasts assesment of Weather Research and Forecast (WRF) model in Cordoba, Argentina // AJAE. 2021. V. 16, N 1. P. 2021133. DOI: 10.5572/ajae.2021.133.
15. Smirnova M.M., Ignatov R.Yu. Analiz vertikal'noi struktury i turbulentnosti v pogranichnom sloe atmosfery v prognozakh regional'noi modeli // Uchenye zapiski RGGMU. 2011. N 21. P. 57–65.
16. Friedman J.H. Greedy function approximation: A gradient boosting machine // Ann. Stat. 2001. V. 29, N 5. P. 1189–1232. DOI: 10.1214/aos/1013203451.
17. Sholle F. Glubokoe obuchenie na Python. SPb.: Piter, 2018. 400 p.
18. Abadi M. Agarwal A., Barham P., Brevdo E., Chen Z., Citro C., Corrado G., Davis A., Dean J., Devin M., Ghemawat S., Goodfellow I., Harp A., Irving G., Isard M., Jia Y., Józefowicz R., Kaiser L., Kudlur M., Levenberg J., Mané D., Monga R., Moore Sh., Murray D., Olah C., Schuster M., Shlens J., Steiner B., Sutskever I., Talwar K., Tucker P., Vanhoucke V., Vasudevan V., Viégas F., Vinyals O., Warden P., Wattenberg M., Wicke M., Yu Yu., Zheng X. TensorFlow: A system for large-scale machine learning // 12th USENIX Symposium on Operating Systems Design and Implementation. 2016. P. 265–283.
19. Gardner M.W., Dorling S.R. Artificial neural networks (the multilayer perceptron). A review of applications in the atmospheric sciences // Atmos. Environ. 1998. V. 32. P. 2627–2636. DOI: 10.1016/s1352-2310(97)00447-0.
20. Bergstra J., Bengio Y. Random search for hyper-parameter optimization // J. Machine Learning Res. 2012. V. 13, N 10. P. 281–305. DOI: 10.5555/2503308.2188395.
21. He K., Zhang X., Ren S., Sun J. Deep residual learning for image recognition // 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2015. P. 770–778. DOI: 10.48550/arXiv.1512.03385.
22. Loshchilov I., Hutter F. Fixing weight decay regularization in Adam // 2018 International Conference on Learning Representations. 2018. DOI: 10.48550/arXiv.1711.05101.
23. Huber P.J. Robust estimation of a location parameter // Ann. Math. Stat. 1964. V. 35, N 1. P. 73–101.