2000 Canopy Area Data

Land Use/Land Cover and Impervious Surface

Canopy cover (percent) for Baldwin County, Alabama.

Process Documentation

The National Land Cover Database 2001 tree canopy layer for mapping zone 46 was produced through a cooperative project conducted by the Multi-Resolution Land Characteristics (MRLC) Consortium.

The canopy land cover data sets are single band raster images. All data are projected to Albers Conical Equal Area using the NAD83 Datum, GRS 1980 Spheroid, with a spatial resolution of 30 meters.

The method employed to map tree canopy density for NLCD 2001 consists of three key steps: deriving reference data of tree canopy density from the high spatial resolution images, calibrating density prediction models using reference data and Landsat spectral bands, and extrapolating the developed models spatially to map per-pixel tree canopy density. This method is described in detail in Huang et al. (2001). This approach was applied to mapping zone 46 by NARSAL at the Institute of Ecology, University of Georgia.

To develop adequate training data for tree canopy density estimation, 38 1-m resolution DOQQs across mapping zone 46 were classified into either tree canopy or non-tree canopy areas, and then values summed within each 30 meter Landsat pixel cell to obtain the percentage of tree canopy.

The training data were selected based on the degree of variance each training data set possesses with regards to Landsat ETM+ imagery used for mapping. Each added training data set decreased the amount of variance present within the Landsat data as a whole. The process was repeated until the required number of samples had been allocated. By combining the training data with Landsat spectral data and other ancillary data, tree canopy prediction models were developed using "Cubist". After the best combination of possible input data layers had been determined and the final percent canopy density layer produced, a mask was made from the final land cover classification of the mapping zone and was used to eliminate those pixels from the canopy density file that fall into non-forested areas. The mask includes 5 land cover classes (deciduous forest, coniferous forest, mixed forest, woody wetlands, and developed open space).