Water-depth-zoning inversion based on the relationship between two-band radiance data and the depth-invariant index

Abstract

Water depth is important in hydrology and is often estimated by remote sensing. Studies have shown that different substrates have a certain influence on satellite derived bathymetry (SDB). In this work, the depth is estimated after the depth-invariant index used to perform substrate classification. This paper proposes a new depth-invariant index based on the relationship of dual-band radiance, and it is compared with the depth-invariance index proposed by Lyzenga (1978) (note that: the index is not used for SDB). Experiments is carried out in shallow water around North Island and East Island of Xisha Islands. Results show that for the two study sites, application of the indexes provide good classification results and reduce the estimation error. The classification accuracy of the proposed index is 2.90% and 0.38% higher than that of the Lyzenga’s model for the two sites, respectively. Compared with the mean absolute error of depth estimation without classification, that of SDB after classification with the proposed index is reduced by 0.10 m and 0.07 m for the two substrate types, while the estimation error of SDB after classification with Lyzenga’s model is 0.11 m and 0.10 m lower than that of the unclassified. Estimation accuracy of water depth over coral reefs showed great improvement, but over coral sands, it showed only slight improvement when compared with SDB based on the unclassified.