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

Accueil du site → Master → Turquie → Estimating Climate Extremes For Turkey And Its Region

İstanbul Teknik Üniversitesi (2012)

Estimating Climate Extremes For Turkey And Its Region


Titre : Estimating Climate Extremes For Turkey And Its Region

Türkiye Ve Bölgesi İçin İklim Uç Değer İstatistiklerinin Kestirimi

Auteur : YILMAZ Yeliz

Université de soutenance : İstanbul Teknik Üniversitesi

Grade : Master of Science (MS) 2012

Extreme climate events have high socio-economic impacts all around the world, in recent years. Especially in last decade (2001-2010), studies on extreme climate events have been increasing. According to report of Turkish State Meteorological Service (TSMS), 555 extreme climate events had recorded since 1940 in 2010 for Turkey. In this study is aimed to extract such information about estimating the distribution of extreme events by using station data and dynamically downscaled climate projections for Turkey and its region. Another point is to find answers for questions such as how important these extreme climate events for Turkey. Analyses are mainly focused on extremes in temperature and precipitation. For this purpose, Extreme Value Analysis (EVA) was used to estimate extreme value statistics. EVA has been used in many disciplines such as hydrology, earth sciences, finance,insurance, metallurgy, environmental research and meteorology etc. In this thesis, Generalized Extreme Value (GEV) distribution models was used for analyses. GEV model was fitted to daily maximum temperature, daily minimum temperature and daily total precipitation for Turkey and its region. Moreover, GEV method allows to analyzing return values, return level, at different time scales such as monthly, seasonal, annual, etc. Return level means that it is exceeded by the maximum value in any particular time scale with probability. In the study of Bozkurt et al., results of global climate models (GCMs) such as ECHAM5, CCSM and HadCM3 are downscaled to force at the boundaries a regional climate model (RCM), RegCM3, to obtain dynamically downscaled climate fields at a resolution of 27 km for the historical (1961-1990) reference period and the 21st Century (2000-2099). EVA is applied to these model outputs and compared with results of NCEP/NCAR Reanalyses data for reference period. All of these analyses were done under the stationary assumption. But it is known that climate data are nonstationary. In extreme value analysis, assumption of time-dependent models is more realistic. The nonstationary extreme value analysis is a developing research area. In this study, probability weighted moments method was used to estimate the parameters (location, shape and scale) of GEV distributions under the assumption of stationary. Uncertainties for GEV parameters were estimated through resampling methods to measure the accuracy of parameters. Resampling methods such as jackknife was applied to reference and projected climate data.


Version intégrale (4,3 Mb)

Page publiée le 20 décembre 2020