Assessment and prediction of water quality parameters in Lake Köyceğiz using artificial neural network approach

---

Tezin Türü: Yüksek Lisans

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Çevre Mühendisliği Bölümü, Türkiye

Tezin Onay Tarihi: 2015

Öğrenci: OYA ÖZÇELİK GEDİK

Danışman: EMRE ALP

Özet:

Water quality monitoring plays a significant role on water resources management and planning. European Union (EU) Water Framework Directive aims to achieve “good status” for all waters.. Within the adaptation period to EU, Turkey aims to improve its water quality monitoring network; however, this will lead to time, budget and human resources problems. Purpose of this thesis is the application of a method that will provide water quality assessment (WQA) under limited budget and data conditions. The method was applied to Lake Köycegiz that is located in Muğla at junction point of Mediterranean and Aegean regions and has nearly 5500 ha surface area. WQA based on multivariate statistical analysis (hypothesis testing and principal component analysis (PCA)) was conducted and water quality status of Lake Köyceğiz and its tributaries were determined based on Surface Water Quality Management Regulation. The results showed that the lake is eutrophic and although there are seasonal differences for water quality parameters, there is no spatial difference between different locations of Lake Köycegiz. In addition, PCA explains the main pollution causes to the lake as fertilizer use in the area or wastewater discharge. Artificial neural network (ANN) approach was performed to predict water quality parameters in the lake using monthly measured water quality parameters of tributaries as input. Different input combinations and performance criteria were tried to find the best predictions. Results revealed low error and high correlation values between measured and estimated parameters. These results indicate great potential of ANNs to predict water quality parameters.