The scientific goal of this thesis consists in the reconstruction of past changes in climate in the geographical region known as Fennoscandia during a time period of approximately 5000 years spanning from 4486 BC and AD 900. In order to achieve this goal, we use annually laminated varved sediments collected from the bottom of three lakes from Sweden (lake Kassjön) and from Finland (lakes Korttajärvi and Nautajärvi). Understanding the climatological history pertaining to the Fennoscandian Peninsula might be particularly useful to scientists, considering the lack of systematic records about the climate history of this region during past millennia. Throughout this thesis, statistical models rooted in functional data analysis and spatial statistics have been designed, implemented and compared, with the goal of clustering dependent and misaligned functional data, collected from multiple sources (i.e. the lakes). In particular, we show that it is possible to uncover a common message across the lakes, which consists in a partition of the years under study in five climate types, which are shared by all lakes and which appear gradually and coherently one after the other throughout the millennia. Lastly, we show that the shared information in the lakes consists precisely in the timing and occurrence of the five types of climate, which, on the other hand, are interpreted differently according to each lake.
L'interesse scientifico di questa tesi consiste nella ricostruzione del clima nella penisola finno-scandinava in un arco di tempo di circa 5000 anni compreso tra il 4486 a.C. e il 900 d.C. facendo uso di strati annuali di sedimenti provenienti dai fondali di tre laghi in Svezia (lago Kassjön) e Finlandia (lago Korttajärvi e lago Nautajärvi). L'utilità delle analisi condotte in questo lavoro risulta particolarmente evidente considerando la mancanza di sistematiche documentazioni storiche in relazione al clima nella penisola finno-scandinava nei millenni passati. In questa tesi sono stati sviluppati, implementati e comparati modelli statistici basati su functional data analysis e spatial statistics per la classificazione non supervisionata di dati funzionali dipendenti tra di loro, disallineati e provenienti da diverse fonti. In particolare, in questa tesi mostriamo che è possibile individuare un messaggio comune ai tre laghi, che si concretizza nella suddivisione degli anni qui analizzati in cinque tipologie climatiche, valide congiuntamente per tutti i tre laghi e che si susseguono gradualmente e coerentemente tra di loro. Infine, mostriamo che ciò che accomuna i tre laghi consiste precisamente nell'occorrenza dei cambiamenti del clima, il quale, invece, si manifesta diversamente in ciascuno di essi.
Joint clustering of functional data from multiple sources to reconstruct historical climatic patterns in Fennoscandia
Di Sabato, Michele
2021/2022
Abstract
The scientific goal of this thesis consists in the reconstruction of past changes in climate in the geographical region known as Fennoscandia during a time period of approximately 5000 years spanning from 4486 BC and AD 900. In order to achieve this goal, we use annually laminated varved sediments collected from the bottom of three lakes from Sweden (lake Kassjön) and from Finland (lakes Korttajärvi and Nautajärvi). Understanding the climatological history pertaining to the Fennoscandian Peninsula might be particularly useful to scientists, considering the lack of systematic records about the climate history of this region during past millennia. Throughout this thesis, statistical models rooted in functional data analysis and spatial statistics have been designed, implemented and compared, with the goal of clustering dependent and misaligned functional data, collected from multiple sources (i.e. the lakes). In particular, we show that it is possible to uncover a common message across the lakes, which consists in a partition of the years under study in five climate types, which are shared by all lakes and which appear gradually and coherently one after the other throughout the millennia. Lastly, we show that the shared information in the lakes consists precisely in the timing and occurrence of the five types of climate, which, on the other hand, are interpreted differently according to each lake.| File | Dimensione | Formato | |
|---|---|---|---|
|
2023_05_DiSabato_Tesi_01.pdf
accessibile in internet per tutti
Descrizione: Testo tesi
Dimensione
37.67 MB
Formato
Adobe PDF
|
37.67 MB | Adobe PDF | Visualizza/Apri |
|
2023_05_DiSabato_Executive_Summary_02.pdf
accessibile in internet per tutti
Descrizione: Executive summary
Dimensione
1.05 MB
Formato
Adobe PDF
|
1.05 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/10589/210346