Combining Ikonos and Bathymetric LiDAR Data to Improve Reef Habitat Mapping in the Florida Keys

Abstract

Mapping reef habitat at a detailed level is a challenge when using single-source remote sensing data such as Ikonos imagery. In this study, Ikonos and bathymetric LiDAR data were combined to improve reef habitat mapping in the Florida Keys, USA. An object-based ensemble approach was applied to integrate two remote sensing data sources for generating accurate and informative reef maps of geomorphological structure types and detailed habitats. LiDAR can significantly increase reef classification by providing important bathymetry, reef habitat elevation, and intensity information. A synergy of Ikonos imagery and all LiDAR-derived features could effectively classify reef structure (3-class) with an overall accuracy (OA) of 93% and Kappa value of 0.8, and map detailed reef habitats (10-class) with an OA of 79% and Kappa value of 0.7. Ensemble analysis of three machine learning outputs did not significantly improve the classification but generated a complementary uncertainty map to identify regions with a robust classification and areas difficult to map. Fusion of Ikonos and bathymetry LiDAR is a promising alternative to traditional in-situ and manual interpretation methods for detailed reef habitat mapping.