This work presents and validates an intra-day operational solar forecast for the southern cone of South America based on GOES16 satellite images. The forecast is evaluated over 12 photovoltaic (PV) generation sites in the northwest of Uruguay. The irradiance predictions are obtained from the combination of different satellite cloudiness forecasts and a standard satellite-to-irradiance conversion model, both specifically adapted to the region. The core of the system is a Cloud Motion Vector (CMV) satellite technique, considered in different spatial resolutions. The optimal linear combination of a single-pixel and space-averaged CMV forecast with the space-averaged persistence is proposed and analyzed. The performance assessment of the individual sub-forecasts and their combination is presented for each time horizon between 10 minutes and 4 hours, with a 10 minutes time step.