İklim ve hidrolojik modeller ile Rize'de bulunan su havzalarının taşkın risk tayini

Abstract

Bu çalışmada, Türkiye'nin en çok yağış alan bölgesi olan Rize ili sınırları içerisinde yer alan havzalarda, mevcut durum ve çeşitli iklim değişikliği senaryoları için taşkın riskinin belirlenmesi amaçlanmıştır. Bu amaç doğrultusunda, ECHAM5 küresel sirkülasyon modelinin, RegCM3 bölgesel modeli ile dinamik olarak küçültülen, A2 emisyon senaryosuna ait günlük yağış çıktıları kullanılarak, havza ve taşkın modelleme çalışmaları gerçekleştirilmiştir. Modelleme çalışmaları, iklim modelinin referans dönemi (1961-1990) ve 3 farklı iklim değişikliği projeksiyon zaman aralığı (2013-2039, 2040-2069, 2070-2099) için ayrı ayrı gerçekleştirilmiştir. Çalışmanın ilk aşamasında, 10 farklı havzada yağış-akış modelleri oluşturulmuştur. Yapılandırılan modellerin altyapıları Coğrafi Bilgi Sistemi (CBS) ortamında hazırlanmıştır. Havza modelleri kalibrasyon çalışmalarında, ölçülmüş yağış ve akış verileri kullanılarak, farklı modelleme yaklaşımları ve model parametreleri test edilmiştir. Oluşturulan nihai modellerin sonucunda, her bir havzada, 4 farklı zaman periyodu için 100 ve 500 yıl dönüş aralıklı taşkın hidrografları elde edilmiştir. Çalışmanın ikinci aşamasında, havzaların çıkış noktalarında taşkın riskinin belirlenmesi amacıyla taşkın modelleme çalışmaları gerçekleştirilmiştir. Modelleme çalışmalarında 2 boyutlu yaklaşım kullanılmış olup, 7 farklı dere güzergahı üzerinde topografya ile uyumlu hesap ağları (grid) oluşturulmuştur. Havza modelleme çalışmaları sonucunda elde edilen hidrograflar, taşkın modellerinde girdi olarak kullanılarak, zamanla değişken hidrolik analizler gerçekleştirilmiştir. Çalışmanın sonucunda, modellenen dere güzergahları boyunca, mevcut durum ve iklim değişikliği senaryoları için, 100 ve 500 yıl dönüş aralıklı maksimum taşkın sınırları, su derinlikleri, akıntı yönü ve hızları elde edilmiştir. Ayrıca, iki farklı ağ çözünürlüğü kullanılarak gerçekleştirilen modelleme çalışmaları sonucunda, ağ çözünürlüğünün model sonuçlarına etkisi de belirlenmiştir. Bu çalışmada, iklim modeli günlük yağış verilerine dayalı havza ve taşkın modellerine ek olarak, zamansal ölçek küçültme yöntemi kullanılarak, Rize'deki şiddet-süre-tekerrür (ŞST) eğrilerinin iklim değişikliği etkisi altında değişimi belirlenmiştir. ŞST eğrilerinin değişimi iki farklı kötümser sera gazı emisyon senaryosu (SRES A2 ve RCP 8.5) altında incelenerek, ŞST eğrilerinin gelecekteki değişimiyle birlikte, farklı senaryolara göre nasıl değişkenlik gösterdiği analiz edilmiştir. Sonuç olarak, iklim modeli günlük toplam yağış verilerine dayanarak yapılan çalışmalarda, tüm havzalarda iklim değişikliğinin etkisiyle yağış değerlerindeki düşüşe bağlı olarak, 2069 yılına kadar pik debi değerlerinde azalma meydana gelmiştir. 2069-2099 yılı arası projeksiyon verileri ile birlikte yüzyılın sonunda debi değerlerinde artış olsa da mevcut durumun altında kalmıştır. İklim değişikliğinden en az etkilenen havzalar İyidere ve Taşlı Dere havzaları olarak elde edilmiştir. Taşkın modelleme çalışmaları sonucunda, İyidere için gerçekleştirilen tüm senaryolarda taşma meydana gelmiştir. 500 yıllık taşkın modeli sonuçlarına göre Taşlı Dere haricinde tüm derelerde mevcut durum için taşkın riski bulunmaktadır. Modelleme çalışmalarında ağ çözünürlüğünün artması, taşkın sınırlarında ve su derinliklerinde azalmalara neden olmuştur. Zamansal ölçek küçültme yöntemi ile elde edilen 24 saatten daha düşük süreli yağış yükseklikleri incelendiğinde, SRES A2 senaryosuna göre, 50 yıldan daha düşük tekerrür aralıkları için tüm gelecek zaman periyotlarında yağış şiddetlerinde artış elde edilmiştir. 50 yıldan daha uzun süreli tekerrür aralıklarına ait yağış şiddetlerinde ise yüzyılın başında artış görülürken, yüzyılın sonuna doğru azalmalar elde edilmiştir. RCP 8.5 senaryosu altında elde edilen sonuçlara göre, tüm tekerrür aralıklarındaki yağış şiddetlerinde yüzyılın başında düşüş gözlemlenirken, yüzyılın orta ve son bölümünde ise yağış şiddetlerinde artış görülmüştür.

In general, hydrological extremes mostly induced by climate change has direct economic and social impacts on society and, thus has long been continuing to be the topic of investigations conducted in a wide spectrum of scientific communities ranging from climatology to public health. Of course, many of the most direct impacts occur by means of the hydrologic cycle for which climate is a driving force. It is becoming increasingly important to be able to accurately predict flooding because it accounts for the greatest losses attributable to natural disasters in the world. Intensive precipitation events in Rize province usually produce damaging floods. Hence, there is a need for estimating peak flow rates from storm events to develop a detailed floodplain management plan and reliable information to promote proper use and management of flood plains in order to protect property and lives and to work on other development plans. The purpose of this study is to produce a reliable watershed and floodplain model for the determination of risk for flood prone areas in Rize province using historical data and produced future data based on climate change scenarios. A model consisting of the combination of Watershed Modeling System (WMS) for peak discharge determination and TuFlow for floodplain analysis were employed. TuFlow which is 1D/2D hydraulic model based on finite difference approach, is known as one of the most popular hydrodynamic software employing shallow water equations. It is suitable for flood studies due to its stability and flexibility. Two-dimensional computations are carried out on Cartesian grids. The model includes the capability to incorporate one-dimensional sections to represent open-channels, rectangular or circular pipes, weirs, or bridges while two-dimensional part is for floodplain modeling. Boundary conditions can be applied to either the 1D and/or 2D portions of the model. Rize province is located in the northeast part of Turkey between Black Sea and North Anatolian Mountains. It has a mild and humid climate with warm summers and cool winters by the influence of Black Sea. The yearly average temperature is about 14oC over the coastline region of Rize. Climate gets colder towards to mountains. Rize is the rainiest city in Turkey, having a total annual precipitation over 2300 mm. Average of rainy days in Rize is about 170 days in a year with highest rainfall rates in autumn and lower rainfall rates in spring. Rize has a drainage area of 3920 km2. The entire study domain divided into 10 small to mid size basins across the province, since major flood damages historically occurred causing serious casualties and physical damages. Streamflow and precipitation gaging stations throughout Rize province are operated by State Water Works (DSI) and State Meteorological Service (MGM), respectively. DSI collects daily streamflow and monthly precipitation data whereas MGM operates precipitation station on daily base. This study comprises two parts: hydrological and hydraulics or floodplain modeling. The former is set using WMS in order to calculate peak flood discharges in present conditions (observed period), covering the period 1961-1990. To understand climate change impacts on flood risk analysis in the study domain, peak discharges were predicted using data for the periods 1961-2039, 1961-2069, and 1961-2099 from climate scenarios. Hydrological model study were carried out using Watershed Modeling System (WMS) software developed by the Environmental Modeling Research Laboratory of Brigham Young University in cooperation with the U.S. Army Corps of Engineers Waterways Experiment Station. WMS is a computer software application in computing physical parameters of watersheds and rainfall-runoff modeling. WMS requires data concerning weather, topography, land use and soil types. The digital elevation data of the basins in Rize province is obtained from Geomatics Division of Istanbul Technical University with a resolution of 30mx30m. Basins located in study area are in mountainous regions with high steeps. The average slope of the basin was calculated as 0.409. Within the basins the hydrologic response units were also calculated based on the combinations of land use and soil type whose maps were generated and overlaid on each other to calculate the average CN values of the basins. Land use map provided by Geomatics Division at Istanbul Technical University. Soil type maps for the study area was not available during research time. Therefore, infiltration rates were determined based on the slope maps. Those areas, which have high slope rates, were assigned low infiltration rate or vice versa. SCS unit hydrograph method were applied to develop a hydrograph. The basin average precipitation depth was used for each basin. Grid-based procedure was adopted to determine basin average precipitation depths for the following four different time spans: 1961-1990, 1961-2039, 1961-2069, 1961-2099. The study area is well known for receiving heavy storms. Critical rainfall duration maps have been available for Turkey in consideration of observed rainfall durations, rainfall amounts and topography. According to rainfall duration maps, Rize province receives critical rainfall with duration up to 24-hour during severe storms. Therefore temporal distribution of 24-hour precipitation was considered in this study. Daily precipitation total values converted to instant values using time distribution curves available for Turkey. Finally, the flood hydrographs were obtained for four different scenarios from the model. In this study, watershed and flood modeling studies were carried out by using grid based precipitation outputs of the ECHAM5 /A2 global circulation model, dynamically downscaled by the RegCM3 regional model. Grid-based rainfall values were converted to basin average rainfall. The basin average rainfall values were obtained using basin area ratios like Thiessen method. The approach of this study is different from classical Thiessen approach as using grid polygons instead of Thiessen polygons. According to the climate change scenarios applied in this research, decreasing trends were observed in the basins in Rize until 2069. Afterward increasing trends were detected in all the basins. Two-dimensional flow modeling was constructed for floodplain mapping with the help of TuFlow, which is an integrated software system designed for interactive use in a multitasking environment. The system is comprised of a graphical user interface, separate hydraulic analysis components, data storage and management routines, and graphics and reporting facilities. The floodplain mapping is split into three primary branches: processing of cross sections and watershed elevations, constructing hydraulic structures, defining flood hydrographs and post processing of hydraulic results and floodplain mapping. A river system is defined by the square grids for the both channel and floodplain. A grid size of 10m×10m was first constructed for the channel and floodplain system. Later, the resolution increased to 5m×5m in examining the grid size resolution for the floodplain mapping determination. Flows and boundary conditions were obtained from the WMS model using the 100-year and 500-year return period flows. Detailed information of the river constructions were obtained from Special Provincial Administration in Rize. In this study, 7 river systems containing measurements among 10 basins were analyzed for floodplain delineation. This study focused on reach lengths between 1.5 km and 3.5 km long, which historically experienced frequent floods. The rivers starting from the most downstream part, which are vulnerable to extreme flood damage in the basin, were adopted into the model. A 100-year and 500-year flood hydrographs were employed for floodplain delineation. A total of 28 scenarios were determined and applied to 7 river systems. Four flood hydrographs were applied using the present and the future periods 2013-2039, 2040-2069, and 2070-2099. The floodplain analysis resulted in 6 scenarios out of 28 causing flooding. Iyidere River seems to have most dangerous potential amongst others since flooding has occurred in every scenario. As a result, urgent action and rehabilitation is highly recommended for Iyidere river. In addition, the effect of grid resolution on the model results was also determined by using two different grid resolutions. For this purpose, 5m and 10m regular grids were created on the same digital terrain model generated from 1:1000 scaled CAD files of the Special Provincial Administration of Rize. The produced floodplains with respect to both mesh resolutions were examined and compared with each other. Results showed that coarse grid resolutions might cause overestimation in the floodplain and water depth.