Multi-Output Neural Networks for Estimation of Synthetic Unit Hydrograph Parameters: A Case Study of a Catchment in Turkey
A. Güven, A.Y. Günal and M. Günal
Uviversity of Gaziantep, Department of Civil Engineering, Gaziantep, Turkey
Full Text PDF
For developing unit hydrographs of catchments, the detailed information about the rainfall and the resulting flood hydrographs are needed. Such information, however, is available only for a few locations and for the remote locations such information is normally very scanty. In this study, Snyder based synthetic unit hydrographs were developed by using both, the digitized map and the digital elevation model of a case study of a small catchment in Turkey. Multi-output neural network technique was applied to predict three unit hydrograph parameters: peak discharge qp, time to peak tp and time base tb of a number of unit hydrographs observed in the catchment, based on most relevant geomorphological and meteorological parameters. Multi-output neural network was observed to outperform: the conventional synthetic unit hydrograph methods. The advantage of the proposed multi-output neural network is based on the fact that it predicts the three parameters of the unit hydrograph, based on a single model, compared to the conventional neural network technique, which utilizes a model for each parameter.

DOI: 10.12693/APhysPolA.132.591
PACS numbers: 92.40.FB, 92.40.QP, 07.05.MH