Avaliação comparativa de modelos hidrodinâmicos para previsão de inundações: um estudo de caso do município de Getúlio Vargas-RS

Abstract

Flood events effects upon areas range from lives lost to property damaged. Thus, flood area mapping is an essential activity for supporting decision makers on land use planning and management. This activity may be carried out through the application of a hydrodynamic mathematical model. As results, the model can deliver not only spatially distributed water depths, but also flow velocities and directions. In regards for the fast development and easier access to the hydrodynamic models’ packages, observed in the past decade, their application has become more often in engineering studies and projects. Thereby, due the expressive diversity of models with different methods and discretizations, it is necessary to compare and evaluate the models’ capabilities and results. The objective of this study is to analyze the performance and uncertainties of three hydrodynamic models currently free available: HEC-RAS, FLO2D and Iber. These softwares are used to map flooded areas. The study area comprises the municipality of Getúlio Vargas-RS, which is prone to floods, with recent occurrences in May 2017, October 2016, July 2015, June 2014, November 2013 and a historical flood in May 1992 that caused severe damages to the riverside population. To fulfil the study main objective, the methodology was composed of: 1) study area characterization by means of field surveys, 2) deployment of an Unmanned Aerial Vehicle (UAV) to compose a Digital Elevation Model (DEM), 3) discharge regionalization to the study area, 4) sensitivity model analysis to input parameters and numerical stability evaluation due to initial conditions and parameters, 5) flood simulations for several discharge values scenarios, and 6) mapping of hazard areas. It was possible to obtain a 0,34 m resolution DEM and discharges to the study area for specific return periods. According to the evaluation HEC-RAS has an application operationality better, and a better performance, as well. It was found that time steps longer than 10 seconds caused numerical instability as well as grids larger than 25 meters. Maps were generated for the different flood return periods for every hydrodynamic model. Flood maps were consistent with observed water level data during flood events. The mapping of the hazard areas was also obtained, and it shows that the areas near the channel were classified as high level, whereas vicinal regions were classified as medium and low levels.