Combining sampling campaigns with satellite spectral data a single band algorithm was developed to retrieve water turbidity, in Paraná River middle basin, North-East Argentina. Water samples were collected and characterized by in situ and laboratory physicochemical measurements. ACOLITE software was used to atmospherically correct Sentinel-2 products, obtaining spectral water surface reflectance. Several models were trained and tested, calculating performance metrics as the selection criteria. The best model was a linear expression between turbidity and spectral band at 704 nm, both transformed by natural logarithm. Turbidity maps followed the same sediment gradient across Paraná River, with high values in the West shore and low presence in the East side. A turbidity regression model can be used to describe water quality, a relevant parameter for water potabilization.