Select a visible 30 × 30 m cell to retrieve a training scene matched by predictive KNN inputs, H/CR/N and Lovell.