Application of machine learning to improving numerical weather prediction

А.Yu. Doroshenko, V.M. Shpyg, R.V. Kushnirenko


In this paper are presented a brief overview of trends in numerical weather prediction, difficulties and the nature of their occurrence, the existing and perspective ways to overcome them. The neural network architecture is proposed as a promising approach to increase the accuracy of the 2m temperature forecast by COSMO regional model. This architecture allows predicting errors of the atmospheric model forecasts with their further corrections. Experiments were conducted with different prehistories of regional model errors. The number of epochs was determined after which the increase of the so-called retraining of the network had place. It is shown that the proposed architecture makes it possible to achieve an improvement of 2m temperature forecast in approximately 50 % of cases.

Problems in programming 2020; 2-3: 375-383


neural network; numerical model; COSMO; forecast; air temperature


Anderson D. & Tannehill J. & Pletcher R. (1990). Computational fluid mechanics and heat transfer. Ed. by G. V. Podviz. Moscow: Mir. 726 p. (in Russian).

Belov P.N. & Borisenkov E.P. & Panin B.D. (1989) Methods of Numerical Weather Prediction. Leningrad.:Gidrometeoizdat. 376 p.

(in Russian).

Volodin E.M. (2007). Mathematical modeling of the general circulation of the atmosphere. Moscow: Institute of Computational Mathematics, Russian Academy of Sciences. 89 p. (in Russian).

Gill Adrian E. (1986) Atmosphere-Ocean Dynamics. Vol. 1,2. Moscow: Mir. 416 p. (in Russian).

Gladkyi A.V. & Skopetsky V.V. (2005) Methods of numerical modeling of ecological processes. Kyiv: Politekhnika. 152 p. (in Ukrainian).

(1988). Weather Dynamics. Ed. by S. Manabe. – Leningrad.:Gidrometeoizdat. 420 p. (in Russian).

Dymnikov V.P. (2007). Stability and predictability of large-scale atmospheric processes. Moscow: Institute of Computational Mathematics, Russian Academy of Sciences. 283 p. (in Russian).

kivganov A.F. et al. (2002). Hydrodynamic weather forecasting methods and grid methods for their implementation. Odessa: Odessa State Environmental University. 179 p. (in Ukrainian).

Kozakov O.L. (2003). Dynamical Meteorology. Odessa: Odessa State Environmental University. 148 p. (in Ukrainian).

Marchuk G.I. (1967). Numerical Methods in Weather Prediction. Leningrad.:Gidrometeoizdat. 353 p. (in Russian).

Prusov V.A. & Snizhko S.I. (2005). Mathematical modeling of atmospheric processes. Kyiv: Nika-Center. 496 p. (In Ukrainian).

Prusov V.A. & Doroshenko A.Yu. (2006). Modeling of natural and technogenic processes in the atmosphere. Кyiv: Naukova dumka. 542 p. (in Ukrainian).

Orlanski I. (1975). A Rational Subdivision of Scales for Atmospheric Processes. Bulletin of the American Meteorological Society. 56(5). P. 527-530. CrossRef

Prusov V.A. et. al. (2007). Effective difference scheme for the numerical solution of the convective diffusion problem. Cybernetics and Systems Analysis. 3. P. 64-74. CrossRef

Frnda J. et al. (2019). A Weather Forecast Model Accuracy Analysis and ECMWF Enhancement Proposal by Neural Network. Sensors. 19(23). P. 5144. CrossRef

Doms G. (2011). A Description of the Nonhydrostatic Regional COSMO-Model. Part I: Dynamics and Numerics. Offenbach: Deutscher Wetterdienst. 153 p.

Shpyg V. et al. (2013). The application of regional NWP models to operational weather forecasting in Ukraine. CAS Technical Conference (TECO) on "Responding to the Environmental Stressors of the 21st Century" Available from: [Accessed 27/02/2020].

Cho K. et al. (2014) Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv preprint arXiv:1406.1078. CrossRef

Hochreiter S. & Schmidhuber J. (1997). Long short-term memory. Neural computation, 9(8), P. 1735-1780. CrossRef

Katsalova L.M. & Shpyg V.M. (2016). The choice of optimal lag for Kriging interpolation of NWP model forecast. Meteorology, Hydrology and Water Management. 4(2). P. 23-28. CrossRef



  • There are currently no refbacks.