Water depth estimation from Worldview-2 image with back propagation neural network in coastal area

Bathymetry provides important information for marine environment with various hydrological applications. Traditional acoustic echo-sounding techniques have been developed for several decades which can measure water depth more than hundreds of meters. However, they have their limitation in the shallow area when it is difficult to access by ships. Recent airborne bathymetry LIDAR systems can access coastal area without restriction and can measure water depth up to tens of meters in clear water, but the operational expanse is high. The economic approach for bathymetry estimation in coastal area is optical satellite images. They can survey a large area with single or multiple satellite images and the penetration of visible light in water merely reaches 30 meters. In this study, a three-layer back propagation neural network is proposed to estimate bathymetry with limited number of training samples. The experiments show the mean square errors are less than 5 meters.