Estudo da influência da distribuição espacial e temporal da precipitação na resposta hidrológica de bacias urbanas de pequeno porte

Abstract

With the increasing urban occupation in flood susceptible areas, storm water management is becoming more and more complex. Therefore, information about storm characteristics, in high spatial and temporal resolution, is important for an adequate rainfallrunoff modeling that can be used to improve the conditions of flood forecast in cities. In the light of this problem, the present study aimed at assessing whether the spatio-temporal variability of precipitation influences the hydrological response of a small urban watershed. Two approaches were used: the first was to verify the spatio-temporal distribution of the observed extreme events in the catchment and the second was to model events with different precipitation time and space distributions. For this research, I chose the Cachoeirinha catchment, located in Belo Horizonte, MG, due to its recurrent flood records. In order to undertake the methodology, I firstly selected the events that generateda flood alert in the catchment. To do so, precipitation data with 10-minutes temporal resolution from 42 monitoring stations in Belo Horizonte were used. Secondly, rainfall fields of the events were created for each 10-minutes interval, by applying the geostatistical technique of Inverse Distance Weighting (IDW) using the Zonal Statistics as Table from ArcMap (version 10.5). By doing that, it was possible to obtain the average rainfall in each sub-catchment for each rain event and time interval, which was then used as input in the model. Following, I adapted and used the rainfall-runoff model Storm Water Management Model (SWMM) for subsequent simulations as in Siqueira, Moura and Silva (2019). The model composed of 35 sub-catchments was manually calibrated with water level data from some observed events. After calibration, a new group of observed events was selected, which were simulated in the model and generated synthetic events of the same duration and total precipitate as the chosen observed events, rotating the rainfall fields of the observed events. In addition, a third objective was also explored, by analysing whether there was a gain in using spatially distributed rainfall in comparison to the lumped modeling approach. From the analysis outputs, it was possible to conclude that the hydrological response of the Cachoeirinha catchment is influenced by the spatial and temporal distribution of the precipitation and that the model correctly represented these influences. The changes in the spatio-temporal distribution of the precipitation mainly affect the time to the peak of the hydrograph. It was also concluded that most of the observed events with flood alert have the same direction of the main watercourse, coming from upstream to downstream in the catchment. Additionally, it was possible to conclude that the simulations performed better with the use of the distributed approach than the lumped one. However the improvement, assessed using the Nash coefficient, was not very significant.