Simulating spectral images of natural waters and submerged objects from a bird’s-eye view

Abstract

We describe a new approach to modelling two-dimensional ‘bird's-eye’ images of natural waters, sensitive to the spectral absorption and scattering over the visual range. Constructed using the mathematical framework of invariant imbedding to efficiently and economically resolve the associated radiative transfer equations, this method provides a fast routine to build physically accurate simulations. We characterise the quality of the water through surface chlorophyll content, instead of the traditional Jerlov water types. Simulations of ‘Macbeth’ colour identification chart at different depths are compared with real, hyperspectral images recorded in a tank.