Classification of SHOALS 3000 bathymetric LiDAR signals using decision tree and ensemble techniques

Coastal Management is a complex issue that is facing policy makers and scientists around the world. Monitoring the coast can be difficult given the vast stretch of water which needs to be covered. Ship-based surveys can take weeks to perform mapping of the coastal zone. LiDAR bathymetry is a tool which shortens the survey time at reduced survey cost. The objective of this paper is to describe the parameterization of the LiDAR waveform from the SHOALS 3000 LiDAR system and show how it can be successfully used for classification of bottom habitat. Decision tree techniques and ensemble methods are used for classification purposes. The Rotation Forest ensemble method provided the greatest overall classification accuracy of 91% averaged over all fractions of the training data.