Application of Artificial Neural Networks and Regression Analysis to L-Moments Based Regional Frequency Analysis in the Eastern Black Sea Basin, Turkey

ANILAN T., Satilmis U., Kankal M., YÜKSEK Ö.

KSCE JOURNAL OF CIVIL ENGINEERING, vol.20, no.5, pp.2082-2092, 2016 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 20 Issue: 5
  • Publication Date: 2016
  • Doi Number: 10.1007/s12205-015-0143-4
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.2082-2092
  • Keywords: regional flood frequency analysis, L-moments, artificial neural networks, eastern black sea basin, UNGAUGED SITES, FLOOD, DISTRIBUTIONS, STATISTICS, SYSTEM
  • Karadeniz Technical University Affiliated: Yes


This study presents the use of L-moments based regression analysis and Artificial Neural Networks (ANN) for forecasting maximum annual flows of Eastern Black Sea Basin, Turkey. Homogeneity of the region is determined by discordancy (D-i) and heterogeneity (Hi) measures based on L-moments. Several distributions are fitted to the data of the 33 stream gauging stations. Return periods (T) corresponding to each flow rates are calculated using the probability density functions of best fit distribution of the region. Using these T values and also drainage area, main stream slope, elevation, stream density, and mean annual rainfall values as independent variables, regression and ANN models are adopted to the data. Mean relative error, mean absolute error and root mean square error are applied for evaluating the performance of the models. Error values indicate that ANN method yields better results for estimation of maximum flows.