Aplicação de índices espectrais na avaliação do aporte de sedimentos ao reservatório da Usina Hidrelétrica Itumbiara (Brasil)

Abstract

This study investigated the sediment input to the reservoir of the Itumbiara Hydroelectric Power Plant (GO/MG) through the application of the Normalized Difference Turbidity Index (NDTI), using Landsat 8 satellite imagery and field data such as precipitation and suspended sediments. The analysis revealed that the methodology was effective in identifying seasonal patterns in water turbidity, with higher turbidity observed during the rainy months. The correlation between NDTI data and suspended sediments demonstrated the accuracy of the index in monitoring turbidity. The application of NDWI to isolate the drainage area was crucial for ensuring a more precise analysis, excluding areas outside the water body. The classification of turbidity levels based on statistical criteria proved adequate, but the research suggests the need for refinements to better capture spatial and temporal variability. The Random Forest model, used to validate the NDTI, highlighted the most relevant variables influencing turbidity, such as reflectance bands and derived indices. Despite the positive results, it became evident that the current model is not sufficient to fully explain sediment dynamics, requiring the inclusion of factors such as land use, topography, and rock types, as well as a better understanding of fluvial processes. The proposed methodology, combining spectral indices and sediment data, proved promising and could be applied to other hydroelectric reservoirs, contributing to the advancement of water quality monitoring and resource management.