Fusion of bathymetric LiDAR and hyperspectral imagery for shallow water bathymetry

We propose combining a forward model based support vector regression and the semianalytical radiative transfer model to determine shallow water characteristics. The derived water depths were compared to both LiDAR derived water depths and field measured water depths. The bathymetry results show that both LiDAR and hyperspectral imagery are unable to retrieve water depth for deeper water (>7 m) due to the water attenuation. Fusion was also performed with the LiDAR bathymetry as a constraint on the hyperspectral imagery; the constraint varies the estimated water characteristics but we were not able to independently assess the performance because no measurements of water column characteristics were available. The retrieved hyperspectral bathymetry yielded a standard deviation of 20 cm when compared to LiDAR bathymetry.