A model to predict spatial, spectral and vertical changes in the average cosine of the underwater light fields: Implications for remote sensing of shelf-sea waters

Abstract

An optical model is developed using experimental data of Inherent Optical Properties (IOP) from oceanic, coastal and productive lagoon waters in order to calculate vertical and spectral profiles of the average cosine in a wide variety of waters within coastal and shelf-sea environments. The results are compared with those generated using a radiative transfer numerical model based on the invariant imbedding technique (HydroLight) with realistic depth-dependent IOPs and appropriate surface and bottom boundary conditions and the results from three existing models (Haltrin, 1998; Timofeyeva, 1971; Talaulikar et al., 2014). The average cosine predicted by the new model shows good agreement with the values obtained directly from radiative transfer calculations for a broad range of the IOPs and solar zenith conditions. Good correlations with excellent linearity with significantly low errors demonstrate a good deal of confidence of the model for accomplishing further applications. Since knowledge of spatial and temporal structures of the average cosine is of great importance to our understanding of the particle dynamics of pelagic ecosystems and coastal processes, efforts were made to apply the present model to both multispectral MODIS-Aqua imagery and hyperspectral (HICO) images acquired over the Arabian Sea and coastal lagoons of the Bay of Bengal dominated by river plumes and phytoplankton blooms. Maps of the average cosine derived from these data demonstrated significant changes in the magnitude and spectral behavior of the average cosine (from nearly featureless to strong spectral features and inflections) from different water types. Substantial changes in its spatial and spectral structures associated with highly productive waters, phytoplankton blooms and sediment plumes, as compared with open ocean areas, are well supported by the theoretical and experimental studies. The advantages of the new model in comparison with existing models are its capability to predict the vertical, spatial, spectral and temporal structures of average cosine with greater accuracy.