Análise das inundações no Vale do Taquari/RS como subsídio à elaboração de um modelo de previsão

Abstract

This study aimed identify the fundamental hydrological variables in the formation of floods and develop a forecast model of floods in real time and with simple operation, for the region of Vale do Taquari, located in Rio Grande do Sul. Thus, at first the study area focused on watershed of the Taquari-Antas river and, subsequently, in the stretch comprising the region of Vale do Taquari, especially among the municipalities of Muçum and Estrela, the most affected by the floods. To achieve the aim was evaluated the existing hydrological monitoring network in the watershed of the Taquari-Antas river; the hydrological behavior of this and finally elaborated a model of simple operation for forecast maximum level of flood in the municipality of Estrela. As hydrological data source used the Hydrological Information System the National Water Agency (HIDROWEB) and records made by other local institutions. The evaluation of the hydrological monitoring network was based on the recommendations of the WMO. The analysis of the hydrological behavior stems from evaluation of morphology, time of concentration of the main sub-basins and points along the main river and recurrence of the highest elevations in the most critical stretch to floods. Utilized the following software throughout the study: ArcGis version 10.1; EasyFit version 5.6; Action; and all calculations procedures were conducting in Excel. As for the model, it has been produced by the multiple linear regression method. 59 tuning options were tested through combinations of the series of maximum level star with the explanatory variables: river levels series in sections upstream Estrela (Encantado and Muçum), which occurred on the day of the Estrela event, on earlier and concerning the maximum amount recorded during the event; as well as data series of cumulative average rainfall in the sub-basin in "x" days prior to the event. There was the best fit through quality measures: mean square error, standard error, Nash-Sutcliffe coefficient; determination and Pearson correlation. The best model found was submitted to verify the compliance of the hypothesis of the method. Based on the results it is concluded that the linear regression model composed by the equation identified as function nº. 54 modified provided the best results for forecast floods in Estrela. This model is based on the maximum levels upstream in Muçum and Encantado municipality and, due to the peak time between the cities be less than one day, it is recommended to apply hourly basis. It was found that the average rainfall no statistical significance in the composition of the regression model. Still, it was found that: (i) the watershed of the Taquari- Antas river presents rapid response to most of the rainfall events with high volumes; and (ii) there is need of network with better quality, both spatially and temporally. In summary, this study gives satisfactory results to the academic area and in the future could provide a significant social contribution, given its continuity and applicability.