Seabed prediction from airborne topo-bathymetric lidar point cloud using machine learning approaches

Predicting the seabed from unfiltered bathymetric lidar data is a very complex task and a critical issue in bathymetric data processing especially with the objective of nautical charting. This is challenging to ensure a high level of quality and security for the needs of a national hydrographic office. This paper proposes a methodology to predict the seabed based on machine learning, which could be useful to automate outlier detection and control the topo-bathymetric lidar point cloud datasets. Several predictive methods have been investigated to predict the seabed from our 2D + 1D data structure. A characteristic dataset of Corsica region was used as a case study for this predictive workflow.