Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Turquie → SPI and artificial intelligence-based analysis of drought conditions in the Black Sea region

Gümüşhane Üniversitesi (2022)

SPI and artificial intelligence-based analysis of drought conditions in the Black Sea region

DENİZ ÖZTÜRK Yasemin

Titre : SPI and artificial intelligence-based analysis of drought conditions in the Black Sea region

Karadeniz Bölgesi’ndeki kuraklık koşullarının SPI ve yapay zeka tabanlı incelenmesi

Auteur : DENİZ ÖZTÜRK Yasemin

Université de soutenance : Gümüşhane Üniversitesi

Grade : Master Thesis 2022

Résumé
In Turkey, many scientific drought analysis studies have been carried out since the 1940s. In these studies, one of the least studied regions in Turkey is the Black Sea Region. Although the Black Sea Region is the rainiest region in Turkey, there are oscillations in precipitation amounts over the years and significant precipitation differences between the coastal and the interior areas. In this study, the drought periods between 1960 and 2020 in the Black Sea Region were examined using SPI and artificial intelligence methods using the data of 26 meteorological stations. According to the SPI values, there is a transition towards more humid conditions throughout the region. This situation is seen both in the monthly SPI heat maps and in the average SPI values. Although there is a trend towards more humidity throughout the region, the period of 1972-1987 was the driest period in the last 60 years, and 1996, 1969, 1974, 1984, 1986, 1993, and 2020 years were moderately and severely drought. In 2020, which was quite arid in Turkey, drought was experienced more prominently, especially in the inner parts of the Black Sea. According to 14 different artificial intelligence prediction methods made using 6 different climate variables, Random Forest, IBK, and Kstar models gave the best drought predictions in the Black Sea region. The lowest accuracy rate belongs to the Naive Bayes method.

Présentation

Version intégrale

Page publiée le 3 janvier 2023