Hybrid Approach Combining Model-Based Method with the Technology of Machine Learning for Forecasting of Dangerous Weather Phenomena

Stankova, Elena N.; Grechko, Irina A.; Kachalkina, Yana N.; Khvatkov, Evgeny V.

doi:10.1007/978-3-319-62404-4_37

Cited by 9 publications

(1 citation statement)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our work, we used the so-called “hybrid” approach [ 6 , 7 ], combining numerical simulation and machine learning methods to identify the dependence of dynamic, microphysical and electrical parameters of convective clouds. This kind of identification is quite important for forecasting thunderstorm with the help of the models which do not have the block describing electrical processes.…”

Section: Introductionmentioning

confidence: 99%

On the Effectiveness of Using Various Machine Learning Methods for Forecasting Dangerous Convective Phenomena

Stankova

Tokareva

Dyachenko

2020

Computational Science and Its Applications – ICCSA 2020

View full text Add to dashboard Cite

The paper considers the possibility of thunderstorm forecasting using only dynamical and microphysical parameters of the cloud, simulated by the 1.5D model with further processing by machine learning methods. The problem of feature selection is discussed in two aspects: selection of the optimal values of time and height when and where the output model data are fixed and selection of fixed set of the most representative cloud parameters (features) among all output cloud characteristics. Five machine learning methods are considered: Support Vector Machine (SVM), Logistic Regression, Ridge Regression, boosted k-nearest neighbour algorithm and neural networks. It is shown that forecast accuracy of all five methods reaches values exceeding 90%.

show abstract

Section: Introductionmentioning

confidence: 99%