High resolution landcover for the Western Gulf Coastal Plain of Louisiana

The dataset includes Land Use/Land Cover types throughout the Chenier Eco-Region in Southwest Louisiana. Using the 2015 NAIP dataset (1m) as the basemap, E-Cognition image objects were derived from the multiresolution segmentation algorithm at 75 and 250 segments. Attempts to refine the data training methods using E-cognition, to extrapolate automating categories of this information to the entire map resulted with exceedingly low accuracy. Therefore, a raster was produced by piecing together several data resources, which provide reliable data for specific LULC categories. This process involved stitching together more reliable sources for specific categories to apply to higher resolution (75) segmentation product. Reference datasets include; 12,000 aerial points assigned to image objects derived from 75 segmentation settings (previously used to develop scripts for data training), mask created from NWI 2008 including water, wetland forested, upland forested and scrub/shrub categories, BOEM marsh classes, NLCD urban areas, and CDL data. The raster produced from this process was applied to the vector image objects derived from the 250 segmentation settings, using a majority filter (greater than 50 percent). The series of draft shapefiles were manually edited and merged, resulting in the final dataset.