Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi

Yazarlar: Murat ÇOBANER, Emin BABAYİĞİT, Bilal BABAYİĞİT

Konular:İnşaat Mühendisliği

DOI:10.28948/ngumuh.295566

Anahtar Kelimeler:Meteorological data,Groundwater level estimation,Multiple linear regression,Genetic programming

Özet:    The significant portion of freshwater needs is met by ground water. Therefore, paying attention to the monitoring of groundwater quality and quantity data and to reveal the current situation is necessary for the calculation of the potential for water development and water management policies. Genetic programming (GP) is an evolutionary-based approach and is used as an alternative to artificial intelligence techniques in recent years. GP has the advantage of revealing a mathematical model by using the relationships between variables. In this study, various mathematical models have been developed to predict the groundwater levels by using meteorological data and previous days of groundwater levels. To the development of mathematical models, a new GP approach, multi-gene genetic programming (MGGP), was implemented. The daily data obtained from Karacaviran observation wells and Develi meteorological station covering the years of 2007-2009 were used for the creation of the models. The accuracy of the generated models to predict the ensuing a month (30 days) groundwater levels were evaluated and compared with the multiple linear regression models. The results obtained with MGGP models were found to be better than multiple linear regression models based on four different criteria. Quite simple and useful models using MGGP have been revealed.

---