The understanding of the compound events represents one of the most interesting challenges in the framework of the studies related to the atmospheric circulation. A group of researchers of the Department of Civil and Environmental Engineering of the Politecnico University of Milan has developed a new model designed to the study of the phenomenology which can be the cause of the compound events. In summary, the method is able to determine the dependence among two first order, two state Markov chains. Six coefficients are computed and they analyse either synchronism or the asynchronism of two temporal series and their values are compared with the confidence interval of the corresponding independent events. The following master thesis proposes the validation of a new version of the model. A statistical analysis was carried out in order to improve the computation of the confidence interval, arranging an empirical calculation through a Montecarlo simulation. Then, a study of the literature was carried out and thus it was proved the homogeneity, the stationarity and the aperiodicity of the annual Markov chain. Subsequently, two model’s implementation for the computation of the probabilities of the random samplings were proposed. The two solutions share an earlier phase of data pruning: the coefficients are compared with the confidence interval based on the assumption of an independent distribution of Bernoulli by the temporal series. The new model has been applied to the meteorological network in Europe and, with respect to the old one, a detailed analysis of the temporal series mostly affected by the statistical error was performed. The new model will move towards a better comprehension of the compound events: a study of the seasonal connection may be carried out in order to achieve a better knowledge of the European and global atmospheric circulation.
La comprensione degli eventi composti rappresenta una delle sfide più interessanti nel panorama degli studi legati alla circolazione atmosferica. Un team di ricercatori del Dipartimento di Ingegneria Civile e Ambientale del Politecnico di Milano ha sviluppato un modello volto all’indagine delle dinamiche fenomenologiche che possono essere alla base degli eventi composti. Esso individua una dipendenza tra due processi markoviani del primo ordine a due stati attraverso sei coefficienti che valutano la sincronia o la asincronia della stessa, che vengono confrontati con l’intervallo di confidenza dei rispettivi eventi congiunti indipendenti. Il seguente elaborato di tesi propone la validazione di una nuova versione del modello. Grazie ad una analisi statistica è stato perfezionato il calcolo dell’intervallo di confidenza, proponendo il calcolo per via empirica attraverso una simulazione Montecarlo. Successivamente è stata effettuata una analisi della letteratura dedicata, che ha evidenziato l’omogeneità, la stazionarietà e l’aperiodicità delle serie markoviane su scala annuale. La fase successiva è stata quella dell’implementazione del nuovo modello, in cui sono state proposte due soluzioni di calcolo per la stima delle probabilità dei campioni sintetici. Entrambe le soluzioni sono comunque accumunate da una antecedente fase di filtraggio dei dati, attraverso un confronto dei coefficienti con l’intervallo di confidenza ottenuto sotto l’ipotesi di distribuzione bernoullina delle serie di dati. Il nuovo modello è stato quindi applicato alla rete pluviometrica europea e, rispetto alla sua prima versione, è stata dedicata una analisi sulle serie di dati maggiormente influenzabili dagli effetti dell’errore statistico. Il nuovo modello potrà proseguire nella direzione inizialmente intrapresa, finalizzandosi allo studio delle connessioni stagionali con l’intento di migliorare la comprensione della dinamica della circolazione atmosferica europea e globale.
Studio delle connessioni in una rete meteorologica : analisi di letteratura, intervallo di confidenza ed errori statistici
BIANCHI, PAOLO
2018/2019
Abstract
The understanding of the compound events represents one of the most interesting challenges in the framework of the studies related to the atmospheric circulation. A group of researchers of the Department of Civil and Environmental Engineering of the Politecnico University of Milan has developed a new model designed to the study of the phenomenology which can be the cause of the compound events. In summary, the method is able to determine the dependence among two first order, two state Markov chains. Six coefficients are computed and they analyse either synchronism or the asynchronism of two temporal series and their values are compared with the confidence interval of the corresponding independent events. The following master thesis proposes the validation of a new version of the model. A statistical analysis was carried out in order to improve the computation of the confidence interval, arranging an empirical calculation through a Montecarlo simulation. Then, a study of the literature was carried out and thus it was proved the homogeneity, the stationarity and the aperiodicity of the annual Markov chain. Subsequently, two model’s implementation for the computation of the probabilities of the random samplings were proposed. The two solutions share an earlier phase of data pruning: the coefficients are compared with the confidence interval based on the assumption of an independent distribution of Bernoulli by the temporal series. The new model has been applied to the meteorological network in Europe and, with respect to the old one, a detailed analysis of the temporal series mostly affected by the statistical error was performed. The new model will move towards a better comprehension of the compound events: a study of the seasonal connection may be carried out in order to achieve a better knowledge of the European and global atmospheric circulation.File | Dimensione | Formato | |
---|---|---|---|
2020_04_Bianchi.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
4.85 MB
Formato
Adobe PDF
|
4.85 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/154068