A Multi-scale Segmentation Approach to Mapping Seagrass Habitats Using Airborne Digital Camera Imagery

• DOI: 10.14358/pers.72.6.665
• Publication Date: 2013-11-27T02:37:44Z
• Authors: Richard G. Lathrop; Paul Montesano; Scott Haag
• Keywords: Digital mapping; Digital camera

The purpose of this study was to map the areal extent and density submerged aquatic vegetation, principally seagrasses, Zostera marina Ruppia maritima, as part ongoing monitoring for Barnegat Bay, New Jersey National Estuary Program. We examine utility multiscale image segmentation/object-oriented classification using eCognition software seagrass across our 36,000 ha area. multi-scale segmentation/ object oriented approach closely mirrored conceptual model spatial structure habitats successfully extracted features ecological interest. agreement between mapped results original field reference 68 percent (Kappa � 56.5 percent) four category 83 63.1 presence/absence map; independent data 71 43.0 a simple map. While aerial digital camera imagery employed in had advantage flexible acquisition, suitable scale, fast processing return time, comparatively low cost, it inconsistent radiometric response from image. This inconsistency made difficult develop rule-based that universally applicable 14 individual mosaics. However, within scene mosaics, “manual classification” mode provided time effective mapping habitats.