NLCD 2011 USFS Tree Canopy analytical (PUERTO RICO)
The U.S. Forest Service analytical canopy product is designed for users that require more analytical vigor in their applications. This product has two layers: percent tree canopy cover (layer one) and standard error (layer two). For layer one, the percent tree canopy cover layer, the file pixel values range from 0 to 100 percent, with each individual value representing the area or proportion of that 30m cell covered by tree canopy. No masking of obvious non-tree areas is performed for this product. For layer two, the standard error layer, the file pixel values range from 0 to 45 percent, with each individual value representing the standard error for the model in that 30m cell. The higher the value, the greater the model uncertainty in estimating the canopy value for that cell. The percent tree canopy cover layer was produced using a Random Forests (trademarked by Leo Breiman and Adele Cutler) regression algorithm and the standard error layer was calculated from the variance of the canopy cover estimates from the random forest regression trees. This mapping methodology is different from the previous NLCD 2001 canopy product, which results in a current product that is not designed to be comparable to NLCD 2001 canopy for change detection.