Author: Alpaslan YARAR, Mustafa ONÜÇYILDIZ

Publishing Date: 2009

E-ISSN: 2147-9364

Volume: 24 Issue: 2

ABSTRACT:

There are some data, which are known but ignored or cannot be measured, in hydro-logical modeling studies. Modeling instruments like Artificial Neural Networks, give efficient results in absence these data. In this study, water level changes of Beyşehir Lake that is the main water resource of Konya Plain Project, was studied with Artificial Neural Networks method. DSİ carried out the determination of level values with Artificial Neural Networks using Inflow – Loss flow, Rainfall, Evaporation, Drawn flow and Level measurements between years 1962 – 1990,  and the obtained values were compared with the results of the traditional methods. 

The best result was obtained by Scaled Conjugate Gradient model with 0.056285 as lower error, under 1 hidden layer, 7 hidden nodes and 500 epochs, with made application. With this study, performed for Beyşehir Lake, it was aimed to obtain the results in a very short time by  eliminating  the  difficulties  and  problems  faced  during  the  traditional  evaluation  of  level measurements.

Key Words: Artificial Neural Networks; Beyşehir Lake; Level Changes; Water Equilibrium; Water Bugget.

FullText