Institutional Repository
Technical University of Crete
EN  |  EL



My Space

Investigations of causality relationships between environmental variables based on time series analysis

Karekla Anastasia

Full record

Year 2022
Type of Item Diploma Work
Bibliographic Citation Anastasia Karekla, "Investigations of causality relationships between environmental variables based on time series analysis", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2022
Appears in Collections


Extreme precipitation in Europe is modulated by large scale atmospheric processes. These processes operate over many decades and carry huge amounts of water from the oceans. Extreme rainfall phenomena can cause humanitarian and economic disasters. Various climate model scenarios predict desertification of certain areas of the Mediterranean basin. In light of such forecasts, it is important to better understand the phenomenon of rainfall and the factors that control it.Many studies have been conducted regarding the temporal and spatial variability of rainfall; however, many questions still remain unanswered. The North Atlantic Oscillation (NAO) index reflects the difference in atmospheric pressure at sea level between Iceland and the Azores, and it has been identified as the main source of variability in the European climate. This thesis aims to examine the impact of the NAO index on the monthly rainfall on the Greek islands of Rhodes and Crete. Monthly data for the period 1980-2020 were used to investigate the causal relationship between NAO index and rainfall. Rainfall data at stations on the islands of Rhodes and Crete come from the reanalysis database MERRA-2, while data for the NAO index were obtained from NOAA (National Oceanic and Atmospheric Administration).Initially, an exploratory statistical analysis of the data was performed, in order to assess basic statistical characteristics of the NAO index and monthly rainfall. The analysis of the cross correlation between NAO index and rainfall showed a statistically significant relation between the two variables. However, the correlation between NAO index and rainfall does not ensure that NAO index is causally related to precipitation. In order to investigate the existence of a causal relationship, Granger causality (GC) analysis was conducted. GC is a widely used method for investigating causal relationships between time series. In addition, a new method of causality analysis (developed by Liang) was used; this method is based on the calculation of the information flow rate between time series.It was established in this thesis through Granger causality analysis (using autoregressive models of orders between 1 and 12) that the NAO index has an impact on rainfall at every station that was examined. This result corresponds to an effect of NAO on rainfall with a temporal lag between 1 and 12 months. The information flow method, on the other hand, detected a small but statistically significant information flow between NAO and rainfall, only for some cases (that is, stations and time lags). The results of Granger causality are consistent with findings from other studies, in which the same method was applied. The results obtained from the two methods (Granger causality and information flow rate) are different, as they are based on different mathematical frameworks. The difference may be due to various factors, such as the nonstationarity and the lack of normal distribution of the rainfall time series, or the existence of statistical biases in the reanalysis precipitation data. Therefore, further investigation is necessary in order to determine with greater certainty the existence or absence of the effect of NAO index on rainfall in Greece.

Available Files