Multivariate distributional regression forests for probabilistic nowcasting of wind profiles
-
Moritz Lang ; Georg von Mayr ; Lisa Schlosser [1] ; Thorsten Simon [1] ; Reto Stauffer [1] ; Achim Zeileis [1]
-
[1] University of Innsbruck
University of Innsbruck
Innsbruck, Austria
-
- Localización: Proceedings of the 35th International Workshop on Statistical Modelling : July 20-24, 2020 Bilbao, Basque Country, Spain / Itziar Irigoien Garbizu (ed. lit.), Dae-Jin Lee (ed. lit.), Joaquín Martínez Minaya (ed. lit.), María Xosé Rodríguez Álvarez (ed. lit.), 2020, ISBN 978-84-1319-267-3, págs. 142-147
- Idioma: inglés
- Enlaces
-
Resumen
This study presents statistical methods to probabilistically predict wind pro les along the approach path of an airport for one hour in advance.
Accurate nowcasts of wind pro les increase safety and facilitate optimal air traf- c management by timely re-routing of landing aircraft when wind direction shifts. Distributional regression trees and forests are enhanced to predict vertical wind pro les employing a multivariate normal distribution. To gain probabilistic forecasts for both wind speed and wind direction, the components of the twodimensional Cartesian wind vector are modeled simultaneously for several height levels of a measurement tower. The resulting tree-based models can capture nonlinear e ects and interactions, and automatically select the relevant covariates that are associated with changes in any of the parameters of the (possibly) highdimensional multivariate normal distribution employed. Extending the multivariate distributional regression trees to multivariate distributional regression forests can further improve the predictive performance by regularizing and smoothing the covariate e ects.
