The need to appropriately deal with spatio-temporal data, especially in the context of weather forecasting, has become fundamental for a modern statistician. This is backed by the media attention that is nowadays focused on the effects of climate change, as well as on the reporting of increasingly frequent extreme weather phenomena. In this paper, some of the most commonly used spatio-temporal data models are described; particular emphasis is placed on hidden Markov models, which make it possible to capture the dependency structure between observations induced by their contiguity in space and/or time. The models are then trained on two datasets constructed from reanalysis data made available by the European Centre Medium Weather Forecast (ECMWF).
Saper analizzare nella maniera più opportuna dati di natura spazio-temporale, specialmente nel contesto delle previsioni meteorologiche, è diventato ormai imprescindibile per lo statistico moderno. L’attenzione mediatica che, al giorno d’oggi, viene riservata agli effetti dei cambiamenti climatici, nonché al racconto dei sempre più frequenti fenomeni atmosferici estremi, ne è la giustificazione. Nel presente elaborato si descrivono alcuni tra i modelli per dati spazio-temporali più utilizzati; particolare enfasi viene posta sugli hidden Markov models, i quali permettono di cogliere la struttura di dipendenza tra osservazioni indotta dalla loro contiguità nello spazio e/o nel tempo. I modelli vengono applicati su due dataset costruiti a partire dai dati di rianalisi messi a disposizione dal Centro europeo per le previsioni meteorologiche a medio termine (ECMWF).
Modelli statistici per dati meteorologici
SCALABRIN, FILIPPO
2023/2024
Abstract
The need to appropriately deal with spatio-temporal data, especially in the context of weather forecasting, has become fundamental for a modern statistician. This is backed by the media attention that is nowadays focused on the effects of climate change, as well as on the reporting of increasingly frequent extreme weather phenomena. In this paper, some of the most commonly used spatio-temporal data models are described; particular emphasis is placed on hidden Markov models, which make it possible to capture the dependency structure between observations induced by their contiguity in space and/or time. The models are then trained on two datasets constructed from reanalysis data made available by the European Centre Medium Weather Forecast (ECMWF).File | Dimensione | Formato | |
---|---|---|---|
Scalabrin_Filippo.pdf
accesso aperto
Dimensione
1.78 MB
Formato
Adobe PDF
|
1.78 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/68409