Clustering of historical floods observed on Iguaçu River, in União da Vitória, Paraná

ABSTRACT The occurrence of flood events has become more frequent, and, in Brazil, there are regions that suffer with the repetition of those events. In União da Vitória, located in Paraná State, Brazil, those phenomena are commonly recorded, generating series of consequences for each flood event, such as financial losses, environmental damages, home losses and deaths. So, since it is not possible to avoid the occurrence of floods, it is necessary to reduce its impacts, and in a scenario of repeated flood events, as observed in União da Vitória, it is justified the clustering of historical floods, aiming to improve the knowledge about the river basin hydrological behavior and to assist in the determination of hydrological models parameters. Clustering analysis aims to establish sets of events with similar characteristics, and, for this, based on fuzzy logic, the present study uses the fuzzy c-means method to cluster Iguaçu river floods, observed in União da Vitória, using a set of different flood severity indicators. The classification defined four clusters, according to different flood severity levels, so-called: low; medium; high; and, disaster or catastrophe. Therefore, by the analysis of similar features among different clusters of events, it is further possible to study the flood formation mechanisms, contributing to the reduction of its impacts, through real-time flood alert and forecasting systems, for instance.