Please use this identifier to cite or link to this item: http://hdl.handle.net/11527/17969
Title: Fırat-dicle, Çoruh Ve Doğu Karadeniz Havzalarında Ekstrem Değer Dağılımı İle Taşkın Frekans Analizinin Yapılması
Other Titles: Flood Frequency Analysis With Generalized Extreme Value Method For Firat-di̇cle, Coruh And Eastern Black Sea River Basins
Authors: Bihrat, Önöz
Cem, Alpan
301131037
Enerji Bilim ve Teknoloji
Energy Sciences and Technologies
Keywords: Sel debisinin hesabı
Enerji
Flood routing
Energy
Issue Date: 1-Jun-2018
Publisher: Enerji Enstitüsü
Energy Institute
Abstract: Akımların incelenmesi akarsu havzalarında bulunan enerji yapıları için büyük önem taşımaktadır. Taşkın değerleri baraj rezervuarlarının biriktirme kapasitesinin hesabı ve taşkın yapılarının projelendirilmesinde kullanılmaktadır. Bu nedenle geçmiş akım kayıtları incelenerek taşkın akımları değerlendirilmesi gelecekteki su kaynakları projelerinin planlanması açısından büyük önem arz etmektedir. Ayrıca gelecekte oluşabilecek taşkınların bilinmesi su yapılarının optimum bir şekilde planlanması ve işletilmesinin yanı sıra özellikle hidroelektrik santrallerin planlanması ve işletilmesinde büyük önem kazanmaktadır. Bu çalışmada, güncel verileri Devlet Su İşleri (DSİ)'den alınan Türkiye'nin önemli su kaynaklarının bulunduğu ve gelecekte yeni enerji projelerinde yoğun olarak kullanılması planlanan Çoruh, Doğu Karadeniz ve Fırat-Dicle havzalarının farklı yerlerinde bulunan toplam 38 adet akım gözlem istasyonunun yıllık anlık maksimum akış verileri değerlendirilmiştir. Taşkın debileri kullanılarak Türkiye'de diğer dağılımlara göre daha güvenilir sonuç verdiği kanıtlanmış olan ekstrem değer dağılımı kullanılarak taşkın frekans analizi yapılmıştır. Taşkın frekans analizi ile elde edilen tahmin değerleri ile ölçüm değerleri yıllık anlık maksimum akış-dönüş aralığı, kuantil-kuantil grafiği, eklenik dağılım grafiği ve olasılık dağılım grafikleriyle karşılaştırılmış ve modelin tutarlılığı Nash-Sutcliffe verimlilik, ortalama hata kareleri kökü ve R2 deterministik katsayı hesaplanarak değerlendirilmiştir. Dağılımın taşkın değerini ölçüm değerlerinden yüksek ya da düşük olma eğilimini bulabilmek için bias oranı hesabı yapılmıştır. Tüm bölgelere ait taşkın zarf grafikleri maksimum debi ve yağış alanları kullanılarak çizilmiş ve istasyon verileri karşılaştırılmıştır. Çizilen grafikler yorumlandığında ve GEV modelinin üç havza için de güvenilir sonuçlar verdiği sonucuna varılmıştır. İstasyonların büyük bir kısmında 0,95'in üzerinde NSE ve R2 değerleri hesaplanmıştır. Eklenik dağılım ve olasılık dağılım grafiklerinde GEV dağılımının ölçüm sonuçlarına yakın eğriler oluşturduğu gözlemlenmiştir. Tüm istasyonlarda taşkın frekans analizi ile elde edilen olasılık denklemleri kullanılarak 100, 200 ve 500 yıllık dönüş aralıklarına göre taşkın tahminleri yapılmıştır. Çalışmanın daha sonraki aşamasında taşkın debisi kayıtları uzun süreli olmayan veya akım gözlem istasyonlarının bulunmadığı yerlerde taşkın tahminlerinin yapılabilmesi için bölgesel analiz yapılmış ve her havza için ayrı bölgesel GEV eğrileri çizilmiştir. Yapılan tüm bu analizler sonucunda havzalarda gerçekleştirilecek enerji projelerinin planlama ve işletmesinde kullanılabilir önemli veriler elde edilmiştir.
Studies on water flows in rivers have high importance for energy buildings in the basin. Flood expectations are used in dam reservoir calculations and flood protection projects. Using historic flood data at a basin is extremely important for planning water projects in future. Flood estimations are very critical for planning and managing energy buildings, especially power plants in the basin. Floods occur as rivers flow rate exceeds the capacity of the river channel. Then water overtops, resulting in some of the water escaping its boundaries. Climate, geological properties, vegetation cover and human effect are main factors that effect floods. Excessive rain in small period of time can cause water level to rise and result in a flood. Sudden change in temperature can cause ice and snow on basin to melt suddenly and raise water level which can cause a flood. Slope of the area and soil structure are two main geological factors of flood. Porous soil can absorb higher amount of water which can prevent water level rising. In river basins with high slope water level will rise more quickly then areas with lower slope. Vegetation cover can also prevent floods. Some of the water after rain is absorbed by plants and some of the water gets stuck on plant surfaces which can get vaporized. Soil is also more likely to be porous at heavy vegetation areas. With development of people, cities get bigger and people need more industrial areas. To build cities and industrial areas people destruct nature by covering soil with concrete which increases flood probability. Flood frequency analysis is a method used by hydrologists to predict flow values according to specific return periods or probabilities for a river. This method was first introduced by Gumbel. Flood frequency analysis is used to calculate statistical information as mean, skewness and standard deviation by using annual maximum water flow data that is available at gauging stations. These calculated values are then used for drawing frequency distribution graphs. These graphs are used to estimate the design flow values according to return periods which can be used for energy building projects design and management. Flood frequency analysis is important for evaluating optimum design parameters for hydroelectric power plants, and preventing underdesigning or overdesigning a project. Flood frequency is also used for estimating recurrence of floods. Flood values calculated with analysis are used in designing structures as dams, bridges, highways, levees and industrial buildings. Estimated values calculated with frequency analysis are useful in providing a measurement parameter to analyze the damage of the floods. Flood frequency estimates are also used for flood insurance and flood zoning. Accurate flood estimations calculated with flood frequency analysis helps engineers in designing safe structures and reducing economic losses caused by floods. In this study, all available annual maximum water flow data of 38 gauging stations are used. Data is taken from DSİ. 20 of these gauging stations are located in varying locations of Fırat-Dicle River Basin. 9 of them are in various rivers of Eastern Black Sea River Basin. 9 of them are on Çoruh River Basin. These three river basins are important water resources of Turkey and there are many on going and planned energy projects for future on them. Annual maximum water level data is used to conduct flood frequency analysis and generalized extreme value method is chosen because according to the previous studies generalized extreme value(GEV) method is proven to be the best fitting method for rivers of Turkey. Three river basins used in this study are important water sources of Turkey. Three of the dams with highest installed capacity are located on Fırat River Basin which are Atatürk Dam with 2400 MW, Karakaya Dam with 1800 MW and Keban Dam with 1330 MW installed capacity. Fırat has the %30 of annual flow of Turkey. Three river basins used in this study has total of %43 of Turkey's annual flow. There are many statistical distributions used for random natural occurances. In this study generalized extreme value distribution is used. According the previous studies GEV distribution is proven to be a better fit with measured flood values in various river basins around the word and previous studies also show that GEV is the best fitting distributions in Turkeys flow values. Values calculated with GEV method are compared with measured annual maximum water flow values measured for each station by using annual maximum water flow-return period, quantile-quantile graph. Quantile-quantile graphs are drawn by calculating GEV flood prediction for the same probability of exceedance with measured values of gauging stations. This graphics can also show deviation of flood prediction calculated with GEV distribution from measured value. Both annual maximum water flow-return period and quantile-quantile graphs showed fitting results for all stations. Cumulative density function and probability density function graphics are also drawn for each gauging station. To draw these graphs frequency of measured values are calculated for class intervals and compared with curves drawn by using probability density function and cumulative density function. Annual maximum values are used for cumulative distribution graphs and fitting graphics are seen. To draw probability distribution graphics annual maximum water flow values are converted to water level. Peak points of the GEV curve and measured data probability frequencies are shown to be fitting. Fit of GEV predictions are calculated with R2, Nash-Sutcliffe efficiency method, root mean squared error. Tendency of GEV model for predicting lower or higher floods then measured values is calculated with percentage bias. By using maximum measured flood value and gauging stations source area, zarf graphs are drawn and all gauging station datas are compared. Graphics are reviewed and GEV method is proven to be fitting for all three basins. For most of the gauging stations Nash-Sutcliffe efficiency and R2 are calculated higher than 0,95. Percentage bias values are mostly positive values which indicates that GEV distribution is more likely to calculate lower values then measured flows. Most of the Pbias values are below 1 which shows there is not a considerable deviation between GEV estimations and measured values. Cumulative density function and probability density function graphics shows fitting results for GEV predictions and measured values. By using cumulative density function formulas found for each station 100, 200 and 500 year return period flood values are calculated. Calculated flood values have critical important value for any project planning and managing on the basins. Amount of available data is very important to have reliable flood calculations. Unfortunately every river do not have gauging stations on them and even some rivers have gauging station, there is not enough amount of data to conduct successful flood frequency analysis. Regional flood frequency analysis is used for the estimation of floods at sites where little or no data are available. It involves the identification of regions of hydrologically homogeneous catchments and the application of a regional estimation method in the identified homogeneous region. For areas with little amount of recorded annual flood value and areas with no gauging station, regional GEV graphs are drawn. With these graphs flood values can be calculated for any return period at areas which average flood value can be calculated with different regression methods. In the result of this flood frequency analysis important values are calculated for planning and managing energy projects for Fırat-Dicle, Eastern Black Sea and Çoruh River Basins.
Description: Tez (Yüksek Lisans) -- İstanbul Teknik Üniversitesi, Enerji Enstitüsü, 2018
Thesis (S.M.)--Massachusetts Institute of Technology, Dept. of Physics, [DATE]
URI: http://hdl.handle.net/11527/17969
Appears in Collections:Enerji Bilim ve Teknoloji Lisansüstü Programı - Yüksek Lisans

Files in This Item:
File Description SizeFormat 
301131037.pdf5.43 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.