Abstract: Remotely sensed bathymetric technology has awide application for the water-depth measurement in a coastal zone, an estuary and other water bodies of different types.Owing to the turbid characteristic of water, the bathymetry in nearshore and estuary areas by means of visible-light remote sensing technology is comparatively difficult.The improvement for the inversion precision of water depth in these areas mainly relies on two aspects:(1)the reliability of bathymetric inversion model; and (2) considering that how sunpended matter in a water body influences the reflection spectrum of water.Taking the Changjiang Estuary as a study area, firstly four appropriate water depth inversion factors were chosen by the correlation analysis between real water depth data and corresponding pixel reflectance value.And then, linear, nonlinear and BP artificial neural network models for the bathymetric inversion were established and compared in detail.On the basis of these models and the study results about the spectral characteristics of suspended sediments, a remote sensing-based bathymetric inversion model was established which was suitable for water depth extraction in shallow, turbid estuary and offshore areas.Inversion results showed that the average relative error is less than 15%.Between the water depths of 7~14 m, the precision is much better and the average relative error only reaches 8.5%.