We present a novel combination of full reference (FR) and no reference (NR) quality models to estimate quality of experience of TV viewers in different locations of a digital TV station coverage area. An NR method is employed at a reference node with optimal reception. In this node the video impairments are only due to the acquisition and the TV station coding process, since the received digital TV signal is the same as the broadcasted. An FR method runs at the other receiving nodes, located at different geographic spots, where signal can be hit by different patterns of packet losses due to the transmission process. The FR method compares the signal received at these nodes with the one received at the reference node. The overall video quality in each geographic spot is calculated by combining the original quality estimation obtained with the NR method and the effects of transmission impairments obtained with the FR method. We show how this design overpasses the performance of previous NR developed methods.