Overview / Usage

Hydrological models are mathematical models that aims to reproduce the hydrological behavior of a basin or sub-basin that have several applications in urban and hydrological planning. However, these models require the use of very specific data that are often not available, in addition, they are very sensitive to data and difficult to calibrate. In this context, the objective of this research is to estimate whether machine learning algorithms are efficient to replace hydrological models. Thus, in this work, deep learning hybrid models with networks such as Gated Recurrent Unit (GRU) and Long Short Term Memory (LSTM) will be used to perform the flow prediction with the use of precipitation data. Finally, an estimate of the performance of each model will be made by comparing it with real flow data and calculating the root mean square error (RMSE). In order to make the comparison between the models, the best model will be the one with the lowest RMSE.