The Extended Stumpf Model for Water Depth Retrieval From Satellite Multispectral Images

Abstract

The commonly used water depth retrieval models do not consider the influence from instantaneous tidal fluctuations during satellite flying, which causes bathymetric errors. For this reason, the wavelet spline instantaneous tidal correction model, which is applied to calculate accurate tidal heights for any moment of satellite flight, is established for correction of water depth retrieval from satellite multispectral images. Two study areas, Saipan Island and Molokai Island in the USA, are selected, and Landsat 8 images are used to verify the accuracy achieved by the proposed model. Fifteen bathymetric check points and three check lines in each of study area are used to further verify the absolute and relative accuracy achievable for the proposed model. The experimental results from our model demonstrated that the mean absolute errors (relative to reference data) in study area, Saipan, and Molokai are 1.42 and 1.5 m, both of which decrease by 33.33% and 27.54%, respectively. The root mean square errors (RMSEs) are 1.67 and 1.83 m, both of which decrease by 31.84% and 27.09%, respectively. Therefore, it can be concluded that the proposed model can largely increase the accuracy of water depth retrieval by satellite multispectral images.

Keywords

• Bathymetry
• instantaneous model
• light detection and ranging (LiDAR)
• multispectral image
• retrieval
• tidal fluctuation