An artificial neural network model for flood forecasting, case study in Jordan

  • Alaa Hawamdeh, Mustafa Al Kuisi


This work presents the flood forecasting over wadi al wala watershed using Artificial Neural Network (ANN). This approach of machine learning applied to predict flood flow based on a long historical record from 1980 to 2018  of  meteorological data (daily rainfall, relative humidity, wet bulb temperature and wind speed) using python programming language. Flood prediction is a non-linear problem and Multi Linear Perceptrons (MLP) neural network type is suitable to be used for this purpose. Various models of different selected numbers of nodes and layers were determined in this study since there is no specific rule to define the architecture of the ANN. The statistical analysis (mean squared error(MSE) and R-squared(R2))  were used for testing the validity of the model by comparing the actual flood values to the predicted values. Results show that the models successfully predicted the flood flow over the watershed.