Extended search
Ice jams forecasting on the Lena River using machine learning methods
Sсhmidt Institute of Physics of the Earth, Russian Academy of Sciences, Moscow, Russia
Journal: Geophysical processes and biosphere
Tome: 21
Number: 3
Year: 2022
Pages: 18–26
UDK: 556 504.453
Show citation
Malygin I.V., Aleshin
I.M. Ice jams forecasting on the Lena River using machine learning methods // . 2022. Т. 21. № 3. С. 18–26. DOI: https://doi.org/10.21455/GPB2022.3-3
@article{MalyginIce2022,
author = "Malygin, I. V. and Aleshin
, I. M.",
title = "Ice jams forecasting on the Lena River using machine learning methods",
journal = "Geophysical processes and biosphere",
year = 2022,
volume = "21",
number = "3",
pages = "18–26",
doi = "https://doi.org/10.21455/GPB2022.3-3",
language = "English"
}
Copy link
Copy BibTex
Keywords: ice jams forecasting, intelligent system, machine learning.
Аnnotation: This paper considers the application of the ice jams forecasting intelligent system, developed earlier for the river Northern Dvina, in a new region – the basin of the Lena River. The application of the developed technology to new data from another region became possible due to a similar formulation of the problem of forecasting and publishing new open datasets of hydrological and meteorological data for the period from 1985 to 2019. Based on observations at hydrological stations and meteorological stations, the system makes it possible to form a short-term forecast of the formation of ice jams powerful in the region under conditions of data incompleteness and data omissions. To prepare the initial data and eliminate gaps, interpolation methods based on machine learning were used. The performed calculations showed the performance of the forecast system. The estimated accuracy of forecasting was 76 %. Assessing the importance of the factors as a whole showed the commonality of the influence of groups of factors in different regions on ice jams formation process.
