In this study, we explore repeatability in photo-interpreted imagery from the National Agriculture Imagery Program that was sampled as part of the National Land Cover Database 2011 Tree Canopy Cover pilot project. Data were collected in 5 diverse pilot areas in the US, including one each in Oregon, Utah, Kansas, Michigan and Georgia. Repeatability metrics. The intra-class correlation coefficient as well as repeatability graphics are explored for each pilot area. In addition, we conduct a simulation to illustrate the effect of varying tolerance and compliance rates on the predictive maps of tree canopy cover.