Spatial Distributions of Reference Sample

To assess the accuracy of Annual NLCD, Collection 1.0, a team of image interpreters review Landsat imagery, high resolution aerial photography, and other supporting data to make annual land cover classifications for thousands of 30-meter by 30-meter plot locations for the full 1985-2023 time series. This process is broken down into two phases. Phase one includes 5,000 randomly distributed plots, which are interpreted before the Annual NLCD map products are even finished. In the second phase, the final map products are used to create a stratified sample of an additional 5000 plots that capture expanded dynamics of the complex dataset. The two sets of interpreted plots make up the Collection 1.0 reference data set and are used to assess the accuracy of the map products.