Desempenho do modelo SWAT para diferentes critérios de geração de unidades de resposta hidrológica

Abstract

The subjectivity in the use of the SWAT model, regarding the spatial discretization of the basin in subbasins along with the hydrologic response units (HRUs) generation, was the issue that motivated this study. It was compared monthly streamflow obtained by SWAT model simulated to a watershed with 2030 km2 of drainage area (Grande River, MG State), taking into account different criteria for HRUs generation as well as spatial discretization in sub-basins. Based on the results, it was conclude that the configuration of the SWAT model with a higher number of sub-basins and multiple HRUs have produced better results for both calibration and validation procedures, in monthly time step, than the configuration based on both less spatial detailing of the drainage network and only a HRU by sub-basin. It was also observed that a fine discretization of the watershed into sub-basins and HRUs has provided better statistical of precision. Among the three criteria for HRU generation tested in this study, there were observed slight differences between the results of the methods based on a single HRU by sub-basin (Predominant HRU and Predominant Classes). However, it was observed that there was a significant improvement on the model’s performance given by the statistics of precision NS (Nash-Sutcliffe coefficient) and PBIAS (bias), when considered the method that generates a greater number of HRU by sub-basin (Multiple HRUs). Thus, it was possible to conclude that the finest spatial discretization has provided better results in monthly hydrologic simulation.