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

Accueil du site → Master → Turquie → Defining Extreme Indices By Using High Resolution Regional Climate Model Simulation

İstanbul Teknik Üniversitesi (2014)

Defining Extreme Indices By Using High Resolution Regional Climate Model Simulation


Titre : Defining Extreme Indices By Using High Resolution Regional Climate Model Simulation

Yüksek Çözünürlükteki Bölgesel İklim Modeli Simülasyonu İle Ekstrem İklim İndekslerinin Belirlenmesi

Auteur : BATIBENİZ Fulden

Université de soutenance : İstanbul Teknik Üniversitesi

Grade : Master of Science (MS) 2014

Résumé partiel
Extreme events connected with changing climate is vitally important for human and natural systems. Therefore, the main aim of this study is to analyse extrem events, calculating indexes and possible trends of extremes, decreasing risk factor to reduce damage and testing model skills in respect of capturing the extreme events. In this study, daily time-scale analyses have been performed to investigate model skills. Domain of experiment includes Turkey and surrounding area. Önol’s previous Regional Climate Model (ICTP-RegCM3) simulation results (2010) ; daily maximum and minimum temperatures and daily precipitations have been used. Model forced with NCEP/NCAR re-analysis data for period including 1961-2008 and horizontal resolution is 10 km (Önol, 2010). Furthermore, extremes such as heavy precipitation, heat wave etc. have been calculated and analysed. In this study, daily maximum and minimum temperatures and daily precipitation data of Turkey’s 187 gages, other than simulations, have been used to calculate extreme indices and possible trends. Through this way, the evaluations of model skills about capturing ability of extreme events will be tested. The indices selected from among World Meteorological Organization’s (WMO) and World Climate Research Program’s (WCRP) extreme indices with regard to impact on human and natural resources. Accordingly, Extreme hot days (TX35), summer days (SU), warm nights (TN90p), warm days (TX90p), Extreme warm days (TX99p) are picked for temperature indices and very wet days (R95p), wet days (RR1), heavy rainy days (R10mm), excessive heavy rainy days (R20mm), consecutive dry days (CDD) are calculated for precipitation indices. Model results and 187 gages of Turkey observed inter-annually to investigate frequency and intensity of extremes. Additionally, extremes analysed by Q-Q plots of station points and corresponding points of model results. The analysis have been performed for upper part of 95. Percentile in precipitation and for upper part of 90. Percentile in maximum and minimum temperatures. Results demonstrate that the simulation and observations are highly consistent for both extreme daily precipitation and extreme daily temperature. Furthermore, time series of all indices of temperature and all indices of precipitation have been observed and correlation coefficients have been calculated. Recent increment at temperature trends have been captured by model results quite well. Additionally, model and observation results are consistent. Similarly, time series of precipitation shows decreasing trend both in model and in observation. Long-term averages calculated from selected indices demonstrates that the model is consistent with observations both in precipitation and temperature variables. Additionally, the biases show that general patern of precipitation and CDD are captured by model. Hence, model overestimates all precipitation indices. Especially on wet days (RR1) model has positive biases ranges in 50-150 days. Inter-annually analysed temperature results also consistent with observations except Extremely Hot Days Indice (TX99p). All indices results generated by model vary as an acceptable range. Therefore, spatial consistency of model and observation has been proved. In order to reach information about places where there is no station, extreme indices maps have been generated. These maps capture general pattern of precipitation quite well, despite model has overestimations on RR1.


Version intégrale (5,2 Mb)

Page publiée le 14 décembre 2020