Ponentes
Descripción
Faculty of Engineering, University of Asuncion, Paraguay. Email: maguilar@fiuna.edu.py, hvelazquez@fiuna.edu.py
This study presents a methodology for predicting maximum flood levels in Mburicao Stream, Asunción, Paraguay, with a one-hour lead time. This information is crucial for issuing timely citizen alerts and reducing flood risks. We leverage high-resolution water level data, collected every 10 minutes from near San Ignacio de Loyola School, combined with rainfall data from two stations: one near the SND Arena and another at the International Airport. The analysis focuses on data from 2021 to identify historical flood events. A linear regression model is then developed to predict future maximum water levels based on this dataset. The methodology incorporates data preprocessing, flood event identification, and model development with performance evaluation using Root Mean Squared Error (RMSE). Our preliminary results demonstrate the model's ability to provide reliable predictions, which is a critical component for effective citizen alert systems. By enabling timely warnings, this approach can significantly reduce the risk of flood-related accidents and fatalities.
Keywords: Flood prediction, Machine learning for flood forecasting, Linear regression for water level prediction, Citizen alert systems for flood warning
