We mosaicked twelve LandSat-8 OLI satellite images taken during the summer of 2014, which were used in an object based image analysis (OBIA) to classify the landscape. We mapped seventeen of the most dominant geomorphic land cover classes on the ACP: (1) Coastal saline waters, (2) Large lakes, (3) Medium lakes, (4) Small lakes, (5) Ponds, (6) Rivers, (7) Meadows, (8) Coalescent low-center polygons, (9) Low-center polygons, (10) Flat-center polygons, (11) High-center polygons, (12) Drained slope, (13) Sandy barrens, (14) Sand dunes, (15) Riparian shrub, (16) Ice, and (17) Urban (i.e. towns and roads). Mapped products were validated with an array of oblique aerial/ground based photography (Jorgenson et al., 2011) and 249 high resolution SPOT-5 images covering >80% of the ACP. We used a stratified random sampling accuracy assessment design, where peaty and sandy lowlands contained 700 and 300 reference sites, respectively. Overall map accuracy was 76% and Cohen’s Kappa coefficient was 0.73
Click on title to download individual files attached to this item.
Potential Metadata Source