Offer TitlePrediction of Instantaneous Peak Streamflow using IF-DNN, Performing Connectivity Analysis with PTE
Organisation NameErciyes University
Family Namelatifoğlu
First Namelevent
Accademic TitleAssistant Professor
Department / SectionCivil Engineering/ Hydrology
Link to 'research gate' or equivalent0000-0002-2837-3306
Keywordswater, drought, rainfall, runoff, sediment
Descriptiondata supplier about hydrology and meteorology,hidrologists
Project DescriptionIn the proposed project, it is aimed to perform a high-performance estimation of the river flow general data and the peak flows that occur in the river flow and are very difficult to predict. For this purpose, it is aimed to develop a method using deep neural networks based on multiple data, including parameters that have not been used before in the literature. In addition, it is aimed to increase the estimation performance by analyzing the sudden peaks in the data with the sudden frequency change by using the Instantaneous Frequency (IF) features, VMD, LMD and Hilbert-Huang methods, which use the decomposition method according to the instantaneous frequency characteristics of the data, which has not been used in the literature before, in the estimation of daily river flow data. In the estimation of the peak flows, not only the past flow data of the predicted river is used, but also the flow data of the side rivers feeding this river, the lake level data where the river flows, and the meteorological precipitation, temperature, humidity, evaporation and wind speed data are also aimed to be used in the estimation study. In the project study, which is planned to be realized, studies will be carried out to define, characterize and determine the amount of inter-regional connectivity of the river flow in different regions depending on multiple parameters. The Phase Transfer Entropy (PTE) approach will be used to show the connectivity of the interregional river flow relationship.
Opportunities for NetworkingIn this project, a powerful prediction method will be created using advanced signal processing methods and Deep Neural Networks (DNN). Thanks to the model to be developed, a preliminary study will be put forward for the prediction of hydro-meteorological data together with the precise estimation of extreme situations.