Aplicabilidade de curvas-chave para modelagem de produção de sedimentos a partir da Modified Universal Soil Loss Equation (MUSLE)

Abstract

The intense use of soil and water resources in watersheds can generate considerable impacts on the environment, being closely associated with erosive processes. In this way, consequently, knowledge and quantification of sediment transport becomes important. The main objectives of this research were to: i) evaluate the use of sediment curves in sedimentometric stations in Rio Grande do Sul; and ii) evaluate the ability of the Modified Universal Soil Loss Equation (MUSLE) to quantify sediment yield in the Cadeia river watershed (CRW), by using observed data (water level and streamflow) combined with sediment curves. The adjustment of the sediment curves to a power function were tested three analytical scenarios of data manipulation: complete data, subdivided into 10-year periods and subdivided into streamflow intervals, and two proposals of input data: data of suspended sediment concentration and suspended sediment load. In CRW, sediment yield for each rainfall event was estimated and compared with the respective observed value according to two proposed sediment curves, derived from the set of monitoring information (rainfall, water level and streamflow). This information also served to obtain the volume of direct surface runoff and the peak discharge. Erodibility, topographic and facing factors were obtained and coefficients a and b were analyzed for three different scenarios, i) Williams (1975); ii) Avanzi et al. (2008); and iii) calibration for the watershed. The coefficient of determination, Average Percent Relative Error and Nash Sutcliffe coefficient demonstrated that for the suspended sediment concentration data greater uncertainties were observed.The best results of the estimate were observed when the model of the tested sediment curve was applied to the complete suspended sediment load data. The results showed that the estimate of sediment yield through MUSLE considering scenario i) is overestimated compared with the values observed for all analyzed events. On the other hand, the scenario ii) seems to generate values closer to the observed values, however, the characteristics of the watershed where these coefficients were calibrated must be considered. MUSLE was able to predict sediment yield when coefficients a and b were calibrated for CRW, presenting acceptable estimates according to the Nash Sutcliffe coefficient, showing the importance of calibrating the coefficients for CRW.