Application of Empirical Mode Decomposition, Feedforward Backpropagation Neural Network, and Cascade Forward Backpropagation Neural Network For Flood Routing: A Case Study of Ankara, Mera River

Katipoğlu O. M., Sarıgöl M.

INSAC International Researches Congress on Natural and Engineering Sciences (INSAC-IRNES'23), Konya, Turkey, 18 - 19 March 2023, pp.225-226

  • Publication Type: Conference Paper / Summary Text
  • City: Konya
  • Country: Turkey
  • Page Numbers: pp.225-226
  • Erzincan Binali Yildirim University Affiliated: Yes


Floods are among the most costly natural disasters worldwide. Flood control, one of the important engineering problems, can be solved by modeling floods correctly. Furthermore, flood routing is vital in helping reduce floods'

impact on people and communities by allowing timely and appropriate responses. In this study, the empirical mode decomposition signal decomposition technique is combined with feedforward backpropagation neural network and

cascade forward backpropagation neural network machine learning techniques to model the 2014 floods in Ankara, Mera River. The flood data is split to avoid the model's underfitting and overfitting problems. While establishing the model,

70% of the data was divided into training, 15% testing, and 15% validation. In the design of the artificial intelligence model, the streamflow observation station D12A102, located in the upstream part of the Mera River, was used as input,

while the streamflow observation station D12A242 downstream was selected as the target. Graphical indicators and statistical parameters were used for the analysis of model performance. As a result of the study, it has been determined

that the EMD signal decomposition technique improves the performance of machine learning models. The study's outputs can assist in designing flood control structures, such as levees and dams, that can help reduce flood risk.