Diploma Thesis Defense by Ms Olianna Akoumianaki

«Space-time analysis of rain data of Crete using statistical methods»

Thesis Title: «Space-time analysis of rain data of Crete using statistical methods»

Monday 7 October 2019, at: 11:30, Venue: Hall Κ2.A3

Examination Committee

  • Professor George Karatzas (advisor)
  • Associate Professor Nikolaos Paranychianakis
  • Dr. Emmanouel Varouhakis

Abstract

This thesis performs a space and time statistical analysis and assessment of the rainfall data on the island of Crete, Grece. Initially, an extensive description of the study area is carried out, which analyzes the geographical, morphological and hydrological characteristics of Crete island. The main features of the climate, its geological background is also described and reference is made to the main uses of land and water, as well as to the surface waters and hydrological basins of the island.

In the following chapter, the database of the rainfall data is presented, listing fifty-eight rainfall stations scattered in the four prefectures of Crete, for the years 1974-2018. The rainfall data were collected from each station separately and analyzed through Excel spreadsheets, as well as through the Hydrognomon software to provide a complete database for direct data analysis at each station. The descriptive statistical analysis of precipitation is following, using mainly plots, to examine the temporal distribution of rainfall. More specifically, charts are presented with the average annual total rainfall for the four prefectures of Crete, for Εastern and Western Crete. Correlation analysis was applied between the four prefectures to examine their relationship on rainfall variability. Also, the rainfall of Crete stations was compared with the Mediterranean Oscillation Index (MOI), which is defined as the normalized pressure difference, between Algiers and Cairo for the period of available data (1973-2017), in order to examine any correlations.

Gumbel allocation charts identify the return periods of maximum rainfall and the expected amount of rainfall and then with the use of Matlab software, significant years are identified, as well as the years with the highest fluctuation for each case studied. Statistical hypothesis for rainfall data is also performed, where t-test and non-parametric Mann-Kendall test are applied. Finally, through the data clustering process, a new method based on neural network technology, the Self-Organizing Maps (SOM) method, is applied, where results of spatial maps are obtained, as the data are sorted and mapped based on their homogeneity. Within the data clustering, the spatial variation of rainfall over the years 2010-2018 is also mapped using the Kriging method and an exponential variogram, which was coded in R environment.

Finally, after presenting the results of the methods used, the main conclusions regarding the spatial distribution of rainfall over the whole island are mentioned. More specifically, the following important conclusions are drawn:

In Western Crete, there is more rainfall than the East and the rainfall increases with the altitude. Also, the highest correlation in rainfall variations identified between Rethymnon and Heraklion, while the lowest between Chania and Lasithi. For all the stations on the island of Crete, it is estimated that during the hydrological years 1974-2018, the amount of precipitation followed a downward trend of 2.15 mm/yr, while the average of the total annual precipitation on the island of Crete, is up to 732.46 mm. The highest rainfall is observed at Askifou station in Chania, while the lowest is observed at Heraklion port station. Finally, it was found, that for most of the prefectures, 2015 is the year with the highest fluctuations and extreme values in rainfall.