Modelling spatially distributed soil losses and sediment yield in the upper grande river basin – Brazil

Abstract

Water erosion negatively affects soil fertility, soil structure, and water availability to plants. Moreover, off-site erosion effects contribute to the sedimentation and eutrophication of water courses. The Grande River is one of the main tributaries of the Paraná River, and an important source of hydroelectric power in Brazil. The Upper Grande River Basin covers an area of 15,705 km2, mostly occupied by rangelands. Shallow and little permeable Cambisols are the predominant soil class in the basin, which, combined with the intensive and highly concentrated summer rainfall, characterize an erosion-prone scenario. The aim of this study was to model the soil losses and the sediment yield in the Upper Grande River Basin. It also sought to quantify the sediment delivery to the two main hydroelectric power plant reservoirs in the basin: Camargos/Itutinga and Funil. Geographical Information Systems (GIS) were used to apply the Revised Universal Soil Loss Equation (RUSLE) and the Sediment Delivery Distributed model (SEDD) in the study area. The models were calibrated using sediment transport data obtained from a river gauging station located in a subwatershed. RUSLE predictions estimated that the average soil losses in the Upper Grande River Basin were of 22.35 t ha− 1 yr− 1, and that bare soils, eucalypt and agriculture suffered the highest erosion rates among the identified land use classes. The average specific sediment yield in the basin was of 1.93 t ha− 1 yr− 1. According to the model calibration, the specific sediment yield predictions showed an error of 0.01 t ha− 1 yr− 1, or 0.6%. Agriculture and eucalypt forests, which compose approximately 10% of the study area, contribute to more than 40% of the sediment yield in the basin. The model predictions estimated that 1.45 million t yr− 1 of sediments are delivered to the Camargos/Itutinga power plant reservoir, whereas the Funil power plant reservoir receives a sediment input of 1.68 million t yr− 1. Although model calibration yielded small errors in relation to the observed sediment measurements, the relative lack of available data has impaired a more thorough validation of the employed models. Nevertheless, the results indicate that the RUSLE/SEDD approach may be useful for analyzing sediment transport in Brazilian watersheds, where limited input data is available.